diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18678 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 3108,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-07,
+      "loss": 1.0887,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 1.0513,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-07,
+      "loss": 1.077,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 1.0184,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 1.0352,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 0.9403,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4893617021276596e-06,
+      "loss": 0.9157,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 0.7227,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9148936170212767e-06,
+      "loss": 0.7366,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 0.6735,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.340425531914894e-06,
+      "loss": 0.3923,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 0.3451,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.765957446808511e-06,
+      "loss": 0.3722,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.978723404255319e-06,
+      "loss": 0.1911,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.191489361702128e-06,
+      "loss": 0.1638,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 0.1662,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6170212765957453e-06,
+      "loss": 0.1634,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 0.1519,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.042553191489362e-06,
+      "loss": 0.1642,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 0.1886,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.468085106382979e-06,
+      "loss": 0.191,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 0.1547,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.893617021276596e-06,
+      "loss": 0.1415,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 0.1458,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 0.1522,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.531914893617022e-06,
+      "loss": 0.136,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.744680851063831e-06,
+      "loss": 0.1357,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.957446808510638e-06,
+      "loss": 0.1367,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.170212765957447e-06,
+      "loss": 0.13,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.382978723404256e-06,
+      "loss": 0.1288,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.595744680851064e-06,
+      "loss": 0.1331,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 0.1328,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.021276595744682e-06,
+      "loss": 0.1338,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 0.129,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 0.126,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 0.1246,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 0.1248,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.085106382978723e-06,
+      "loss": 0.1346,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.297872340425532e-06,
+      "loss": 0.1256,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 0.1281,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.72340425531915e-06,
+      "loss": 0.1206,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 0.1256,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.148936170212767e-06,
+      "loss": 0.13,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.361702127659576e-06,
+      "loss": 0.1185,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.574468085106385e-06,
+      "loss": 0.1253,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 0.1341,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1216,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0212765957446808e-05,
+      "loss": 0.1229,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0425531914893619e-05,
+      "loss": 0.1234,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 0.1196,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0851063829787233e-05,
+      "loss": 0.135,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1063829787234044e-05,
+      "loss": 0.1281,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1276595744680851e-05,
+      "loss": 0.1194,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1489361702127662e-05,
+      "loss": 0.1194,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.170212765957447e-05,
+      "loss": 0.1224,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1914893617021277e-05,
+      "loss": 0.1311,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2127659574468087e-05,
+      "loss": 0.1376,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2340425531914895e-05,
+      "loss": 0.1222,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2553191489361702e-05,
+      "loss": 0.1327,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2765957446808513e-05,
+      "loss": 0.13,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.297872340425532e-05,
+      "loss": 0.1147,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3191489361702127e-05,
+      "loss": 0.1224,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3404255319148938e-05,
+      "loss": 0.118,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3617021276595745e-05,
+      "loss": 0.1093,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3829787234042556e-05,
+      "loss": 0.118,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4042553191489363e-05,
+      "loss": 0.114,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.425531914893617e-05,
+      "loss": 0.118,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4468085106382981e-05,
+      "loss": 0.1108,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4680851063829789e-05,
+      "loss": 0.112,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4893617021276596e-05,
+      "loss": 0.108,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5106382978723407e-05,
+      "loss": 0.1182,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5319148936170214e-05,
+      "loss": 0.1068,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5531914893617023e-05,
+      "loss": 0.1072,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5744680851063832e-05,
+      "loss": 0.11,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.595744680851064e-05,
+      "loss": 0.1035,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6170212765957446e-05,
+      "loss": 0.1048,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6382978723404255e-05,
+      "loss": 0.1094,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6595744680851064e-05,
+      "loss": 0.0921,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6808510638297873e-05,
+      "loss": 0.0851,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 0.1043,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.723404255319149e-05,
+      "loss": 0.0928,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.74468085106383e-05,
+      "loss": 0.0919,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.765957446808511e-05,
+      "loss": 0.0811,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7872340425531915e-05,
+      "loss": 0.0788,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8085106382978724e-05,
+      "loss": 0.0776,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8297872340425533e-05,
+      "loss": 0.0912,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8510638297872342e-05,
+      "loss": 0.0791,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.872340425531915e-05,
+      "loss": 0.0847,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.893617021276596e-05,
+      "loss": 0.0787,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.914893617021277e-05,
+      "loss": 0.0779,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9361702127659575e-05,
+      "loss": 0.0785,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9574468085106384e-05,
+      "loss": 0.0757,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9787234042553193e-05,
+      "loss": 0.0757,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999999456770665e-05,
+      "loss": 0.0705,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999978270832498e-05,
+      "loss": 0.0755,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999951109395245e-05,
+      "loss": 0.0656,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999991308342441e-05,
+      "loss": 0.0628,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99998641929613e-05,
+      "loss": 0.0655,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999804438059042e-05,
+      "loss": 0.0669,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999733818782543e-05,
+      "loss": 0.0623,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999965233520854e-05,
+      "loss": 0.0695,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999559987425553e-05,
+      "loss": 0.0637,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999945677553392e-05,
+      "loss": 0.0614,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999342699645774e-05,
+      "loss": 0.0706,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999921775988505e-05,
+      "loss": 0.0724,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99990819563875e-05,
+      "loss": 0.0588,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999893528930066e-05,
+      "loss": 0.0746,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999877775878388e-05,
+      "loss": 0.0612,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998609365008313e-05,
+      "loss": 0.0694,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999843010815691e-05,
+      "loss": 0.0637,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998239988424425e-05,
+      "loss": 0.0783,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998039006017414e-05,
+      "loss": 0.061,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999782716115424e-05,
+      "loss": 0.0546,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997604454065063e-05,
+      "loss": 0.0575,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997370884991842e-05,
+      "loss": 0.0623,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997126454188343e-05,
+      "loss": 0.0682,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999687116192013e-05,
+      "loss": 0.0572,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996605008464564e-05,
+      "loss": 0.0522,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999632799411081e-05,
+      "loss": 0.0699,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996040119159842e-05,
+      "loss": 0.062,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995741383924413e-05,
+      "loss": 0.0617,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999543178872909e-05,
+      "loss": 0.0642,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995111333910238e-05,
+      "loss": 0.065,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999478001981602e-05,
+      "loss": 0.0566,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994437846806384e-05,
+      "loss": 0.0626,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99940848152531e-05,
+      "loss": 0.08,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999372092553971e-05,
+      "loss": 0.0572,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993346178061572e-05,
+      "loss": 0.0705,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992960573225837e-05,
+      "loss": 0.0644,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992564111451444e-05,
+      "loss": 0.0688,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999215679316913e-05,
+      "loss": 0.0611,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991738618821435e-05,
+      "loss": 0.0567,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991309588862683e-05,
+      "loss": 0.0624,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990869703759e-05,
+      "loss": 0.0758,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990418963988306e-05,
+      "loss": 0.0594,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989957370040305e-05,
+      "loss": 0.0661,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989484922416503e-05,
+      "loss": 0.0602,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989001621630196e-05,
+      "loss": 0.0517,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998850746820647e-05,
+      "loss": 0.0667,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988002462682198e-05,
+      "loss": 0.0579,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998748660560605e-05,
+      "loss": 0.0616,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986959897538488e-05,
+      "loss": 0.0702,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986422339051753e-05,
+      "loss": 0.0705,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985873930729883e-05,
+      "loss": 0.0543,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985314673168696e-05,
+      "loss": 0.0513,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998474456697581e-05,
+      "loss": 0.0634,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9984163612770617e-05,
+      "loss": 0.0578,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983571811184297e-05,
+      "loss": 0.0606,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982969162859828e-05,
+      "loss": 0.0572,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982355668451948e-05,
+      "loss": 0.0596,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981731328627208e-05,
+      "loss": 0.0606,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981096144063916e-05,
+      "loss": 0.0568,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980450115452183e-05,
+      "loss": 0.0603,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979793243493887e-05,
+      "loss": 0.0595,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979125528902688e-05,
+      "loss": 0.0689,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978446972404043e-05,
+      "loss": 0.0552,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977757574735164e-05,
+      "loss": 0.057,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977057336645057e-05,
+      "loss": 0.0682,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976346258894502e-05,
+      "loss": 0.0753,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975624342256057e-05,
+      "loss": 0.0627,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997489158751405e-05,
+      "loss": 0.0562,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974147995464593e-05,
+      "loss": 0.0552,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997339356691557e-05,
+      "loss": 0.0574,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997262830268663e-05,
+      "loss": 0.0709,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971852203609204e-05,
+      "loss": 0.0635,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971065270526494e-05,
+      "loss": 0.0638,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970267504293466e-05,
+      "loss": 0.0654,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996945890577686e-05,
+      "loss": 0.0559,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968639475855192e-05,
+      "loss": 0.0632,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996780921541873e-05,
+      "loss": 0.0692,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966968125369522e-05,
+      "loss": 0.0662,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996611620662138e-05,
+      "loss": 0.0566,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965253460099872e-05,
+      "loss": 0.0622,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964379886742344e-05,
+      "loss": 0.0616,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996349548749789e-05,
+      "loss": 0.0581,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962600263327376e-05,
+      "loss": 0.0595,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961694215203424e-05,
+      "loss": 0.0603,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960777344110426e-05,
+      "loss": 0.0599,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959849651044515e-05,
+      "loss": 0.0617,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958911137013596e-05,
+      "loss": 0.0581,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957961803037325e-05,
+      "loss": 0.0652,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957001650147116e-05,
+      "loss": 0.057,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956030679386132e-05,
+      "loss": 0.0715,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955048891809293e-05,
+      "loss": 0.0765,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954056288483273e-05,
+      "loss": 0.0624,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953052870486492e-05,
+      "loss": 0.0741,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995203863890912e-05,
+      "loss": 0.0652,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951013594853087e-05,
+      "loss": 0.065,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949977739432052e-05,
+      "loss": 0.0645,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948931073771433e-05,
+      "loss": 0.0555,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947873599008387e-05,
+      "loss": 0.0607,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946805316291817e-05,
+      "loss": 0.0513,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994572622678237e-05,
+      "loss": 0.0553,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994463633165243e-05,
+      "loss": 0.0504,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943535632086125e-05,
+      "loss": 0.0579,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942424129279318e-05,
+      "loss": 0.0557,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941301824439608e-05,
+      "loss": 0.0506,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994016871878634e-05,
+      "loss": 0.0521,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993902481355058e-05,
+      "loss": 0.0549,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937870109975138e-05,
+      "loss": 0.0507,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936704609314553e-05,
+      "loss": 0.058,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935528312835087e-05,
+      "loss": 0.0548,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993434122181474e-05,
+      "loss": 0.0568,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933143337543238e-05,
+      "loss": 0.0503,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931934661322035e-05,
+      "loss": 0.0512,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930715194464304e-05,
+      "loss": 0.0489,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992948493829495e-05,
+      "loss": 0.0588,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928243894150586e-05,
+      "loss": 0.0532,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9926992063379568e-05,
+      "loss": 0.0436,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992572944734195e-05,
+      "loss": 0.0483,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924456047409517e-05,
+      "loss": 0.0519,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992317186496576e-05,
+      "loss": 0.0552,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921876901405894e-05,
+      "loss": 0.0561,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9920571158136837e-05,
+      "loss": 0.0562,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919254636577235e-05,
+      "loss": 0.0462,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917927338157428e-05,
+      "loss": 0.0464,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916589264319475e-05,
+      "loss": 0.0474,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915240416517134e-05,
+      "loss": 0.0581,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991388079621587e-05,
+      "loss": 0.066,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912510404892865e-05,
+      "loss": 0.0562,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991112924403698e-05,
+      "loss": 0.0606,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909737315148798e-05,
+      "loss": 0.046,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908334619740587e-05,
+      "loss": 0.0581,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906921159336318e-05,
+      "loss": 0.0533,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990549693547166e-05,
+      "loss": 0.0571,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904061949693972e-05,
+      "loss": 0.0468,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902616203562307e-05,
+      "loss": 0.0439,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990115969864741e-05,
+      "loss": 0.0504,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989969243653171e-05,
+      "loss": 0.044,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.0452,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896725647086074e-05,
+      "loss": 0.047,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895226122979426e-05,
+      "loss": 0.0526,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893715848118566e-05,
+      "loss": 0.0527,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892194824144338e-05,
+      "loss": 0.0481,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890663052709276e-05,
+      "loss": 0.0588,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889120535477584e-05,
+      "loss": 0.0627,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9887567274125143e-05,
+      "loss": 0.0521,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988600327033951e-05,
+      "loss": 0.0527,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884428525819906e-05,
+      "loss": 0.0559,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988284304227723e-05,
+      "loss": 0.0481,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881246821434047e-05,
+      "loss": 0.0449,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879639865024573e-05,
+      "loss": 0.0565,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878022174794714e-05,
+      "loss": 0.0516,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876393752502013e-05,
+      "loss": 0.0615,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987475459991569e-05,
+      "loss": 0.0575,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873104718816613e-05,
+      "loss": 0.048,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987144411099731e-05,
+      "loss": 0.0514,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869772778261967e-05,
+      "loss": 0.054,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986809072242641e-05,
+      "loss": 0.05,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866397945318133e-05,
+      "loss": 0.0528,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986469444877626e-05,
+      "loss": 0.0586,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862980234651574e-05,
+      "loss": 0.0488,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9861255304806493e-05,
+      "loss": 0.0429,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9859519661115087e-05,
+      "loss": 0.0389,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985777330546306e-05,
+      "loss": 0.057,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9856016239747754e-05,
+      "loss": 0.0506,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985424846587815e-05,
+      "loss": 0.0504,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985246998577486e-05,
+      "loss": 0.0542,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985068080137013e-05,
+      "loss": 0.0582,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848880914607834e-05,
+      "loss": 0.0473,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847070327443476e-05,
+      "loss": 0.0478,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984524904184418e-05,
+      "loss": 0.0482,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843417059788702e-05,
+      "loss": 0.0631,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9841574383267415e-05,
+      "loss": 0.0451,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9839721014282307e-05,
+      "loss": 0.0529,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983785695484699e-05,
+      "loss": 0.0527,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835982206986685e-05,
+      "loss": 0.0556,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834096772738233e-05,
+      "loss": 0.0553,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832200654150077e-05,
+      "loss": 0.0535,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983029385328227e-05,
+      "loss": 0.0636,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982837637220647e-05,
+      "loss": 0.0523,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982644821300595e-05,
+      "loss": 0.0613,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824509377775567e-05,
+      "loss": 0.0639,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982255986862179e-05,
+      "loss": 0.0436,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9820599687662677e-05,
+      "loss": 0.0474,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981862883702789e-05,
+      "loss": 0.0491,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816647318858666e-05,
+      "loss": 0.0523,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981465513530785e-05,
+      "loss": 0.0465,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981265228853987e-05,
+      "loss": 0.0501,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981063878073073e-05,
+      "loss": 0.0497,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980861461406802e-05,
+      "loss": 0.0511,
+      "step": 282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806579790750922e-05,
+      "loss": 0.0588,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9804534312990183e-05,
+      "loss": 0.0585,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980247818300813e-05,
+      "loss": 0.052,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9800411403038664e-05,
+      "loss": 0.0529,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9798333975327258e-05,
+      "loss": 0.0504,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9796245902130948e-05,
+      "loss": 0.0533,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979414718571834e-05,
+      "loss": 0.0552,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9792037828369607e-05,
+      "loss": 0.0483,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978991783237647e-05,
+      "loss": 0.0524,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787787200042224e-05,
+      "loss": 0.052,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9785645933681713e-05,
+      "loss": 0.0469,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978349403562133e-05,
+      "loss": 0.0566,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978133150819902e-05,
+      "loss": 0.0469,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9779158353764293e-05,
+      "loss": 0.0552,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977697457467818e-05,
+      "loss": 0.0568,
+      "step": 297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9774780173313273e-05,
+      "loss": 0.0421,
+      "step": 298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9772575152053696e-05,
+      "loss": 0.0376,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977035951329511e-05,
+      "loss": 0.0486,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976813325944472e-05,
+      "loss": 0.04,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9765896392921253e-05,
+      "loss": 0.0425,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9763648916154982e-05,
+      "loss": 0.0563,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9761390831587687e-05,
+      "loss": 0.0585,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975912214167269e-05,
+      "loss": 0.0588,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975684284887483e-05,
+      "loss": 0.0516,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975455295567046e-05,
+      "loss": 0.0442,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752252464547454e-05,
+      "loss": 0.0469,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97499413780052e-05,
+      "loss": 0.0598,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9747619698554606e-05,
+      "loss": 0.0468,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745287428718072e-05,
+      "loss": 0.0432,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9742944571029515e-05,
+      "loss": 0.0546,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9740591128034357e-05,
+      "loss": 0.072,
+      "step": 313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973822710228951e-05,
+      "loss": 0.0522,
+      "step": 314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9735852496363395e-05,
+      "loss": 0.0473,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973346731283592e-05,
+      "loss": 0.0534,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973107155429849e-05,
+      "loss": 0.0476,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728665223354003e-05,
+      "loss": 0.0606,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972624832261683e-05,
+      "loss": 0.0422,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9723820854712836e-05,
+      "loss": 0.0428,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972138282227937e-05,
+      "loss": 0.047,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971893422796524e-05,
+      "loss": 0.0648,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9716475074430762e-05,
+      "loss": 0.0541,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714005364347687e-05,
+      "loss": 0.0458,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971152510039926e-05,
+      "loss": 0.0495,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9709034285280185e-05,
+      "loss": 0.0525,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9706532921696634e-05,
+      "loss": 0.0526,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9704021012366228e-05,
+      "loss": 0.0424,
+      "step": 328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9701498560018055e-05,
+      "loss": 0.056,
+      "step": 329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9698965567392656e-05,
+      "loss": 0.0499,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9696422037242023e-05,
+      "loss": 0.0453,
+      "step": 331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96938679723296e-05,
+      "loss": 0.0553,
+      "step": 332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9691303375430265e-05,
+      "loss": 0.0532,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.968872824933035e-05,
+      "loss": 0.0561,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9686142596827623e-05,
+      "loss": 0.0434,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9683546420731292e-05,
+      "loss": 0.0465,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9680939723861993e-05,
+      "loss": 0.0453,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967832250905179e-05,
+      "loss": 0.0517,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9675694779144184e-05,
+      "loss": 0.0492,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9673056536994094e-05,
+      "loss": 0.0581,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967040778546786e-05,
+      "loss": 0.0427,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9667748527443248e-05,
+      "loss": 0.0427,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966507876580942e-05,
+      "loss": 0.057,
+      "step": 343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9662398503466963e-05,
+      "loss": 0.0556,
+      "step": 344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659707743327878e-05,
+      "loss": 0.0509,
+      "step": 345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9657006488315563e-05,
+      "loss": 0.0515,
+      "step": 346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965429474136482e-05,
+      "loss": 0.0544,
+      "step": 347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965157250542185e-05,
+      "loss": 0.0487,
+      "step": 348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9648839783444243e-05,
+      "loss": 0.0513,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9646096578401e-05,
+      "loss": 0.0544,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643342893272493e-05,
+      "loss": 0.0704,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9640578731050487e-05,
+      "loss": 0.0496,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9637804094738133e-05,
+      "loss": 0.0442,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635018987349955e-05,
+      "loss": 0.0488,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963222341191186e-05,
+      "loss": 0.0502,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9629417371461124e-05,
+      "loss": 0.0479,
+      "step": 356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962660086904639e-05,
+      "loss": 0.048,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623773907727682e-05,
+      "loss": 0.0566,
+      "step": 358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962093649057637e-05,
+      "loss": 0.0531,
+      "step": 359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9618088620675188e-05,
+      "loss": 0.0482,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9615230301118234e-05,
+      "loss": 0.059,
+      "step": 361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961236153501095e-05,
+      "loss": 0.0593,
+      "step": 362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9609482325470137e-05,
+      "loss": 0.0394,
+      "step": 363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960659267562393e-05,
+      "loss": 0.0393,
+      "step": 364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9603692588611816e-05,
+      "loss": 0.0476,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9600782067584625e-05,
+      "loss": 0.0515,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9597861115704512e-05,
+      "loss": 0.0571,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.0438,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591987932090836e-05,
+      "loss": 0.0412,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9589035706738237e-05,
+      "loss": 0.054,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958607306329466e-05,
+      "loss": 0.0499,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9583100004978886e-05,
+      "loss": 0.0488,
+      "step": 372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9580116535021022e-05,
+      "loss": 0.0494,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9577122656662485e-05,
+      "loss": 0.0497,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9574118373156e-05,
+      "loss": 0.0458,
+      "step": 375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95711036877656e-05,
+      "loss": 0.052,
+      "step": 376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9568078603766613e-05,
+      "loss": 0.0419,
+      "step": 377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9565043124445665e-05,
+      "loss": 0.0579,
+      "step": 378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9561997253100684e-05,
+      "loss": 0.0469,
+      "step": 379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9558940993040885e-05,
+      "loss": 0.0646,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955587434758676e-05,
+      "loss": 0.0504,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9552797320070097e-05,
+      "loss": 0.0527,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954970991383396e-05,
+      "loss": 0.0391,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546612132232688e-05,
+      "loss": 0.0513,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9543503978631894e-05,
+      "loss": 0.0532,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540385456408456e-05,
+      "loss": 0.0596,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953725656895052e-05,
+      "loss": 0.0457,
+      "step": 387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953411731965749e-05,
+      "loss": 0.0528,
+      "step": 388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530967711940038e-05,
+      "loss": 0.0481,
+      "step": 389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952780774922008e-05,
+      "loss": 0.0586,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952463743493078e-05,
+      "loss": 0.0528,
+      "step": 391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521456772516554e-05,
+      "loss": 0.0466,
+      "step": 392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518265765433066e-05,
+      "loss": 0.0443,
+      "step": 393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951506441714721e-05,
+      "loss": 0.0448,
+      "step": 394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951185273113712e-05,
+      "loss": 0.0454,
+      "step": 395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508630710892157e-05,
+      "loss": 0.0426,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9505398359912914e-05,
+      "loss": 0.0437,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950215568171121e-05,
+      "loss": 0.0469,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9498902679810083e-05,
+      "loss": 0.0451,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9495639357743773e-05,
+      "loss": 0.0471,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9492365719057752e-05,
+      "loss": 0.0445,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489081767308696e-05,
+      "loss": 0.0526,
+      "step": 402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485787506064483e-05,
+      "loss": 0.0517,
+      "step": 403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9482482938904183e-05,
+      "loss": 0.0521,
+      "step": 404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947916806941808e-05,
+      "loss": 0.053,
+      "step": 405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947584290120764e-05,
+      "loss": 0.0417,
+      "step": 406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472507437885523e-05,
+      "loss": 0.0578,
+      "step": 407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9469161683075568e-05,
+      "loss": 0.0465,
+      "step": 408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94658056404128e-05,
+      "loss": 0.0405,
+      "step": 409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9462439313543423e-05,
+      "loss": 0.0522,
+      "step": 410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459062706124813e-05,
+      "loss": 0.0432,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945567582182551e-05,
+      "loss": 0.0407,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9452278664325227e-05,
+      "loss": 0.0475,
+      "step": 413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9448871237314834e-05,
+      "loss": 0.045,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9445453544496363e-05,
+      "loss": 0.0478,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9442025589582988e-05,
+      "loss": 0.0414,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438587376299047e-05,
+      "loss": 0.0513,
+      "step": 417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9435138908380017e-05,
+      "loss": 0.04,
+      "step": 418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431680189572514e-05,
+      "loss": 0.0435,
+      "step": 419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942821122363429e-05,
+      "loss": 0.0424,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942473201433424e-05,
+      "loss": 0.0511,
+      "step": 421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9421242565452373e-05,
+      "loss": 0.0477,
+      "step": 422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9417742880779835e-05,
+      "loss": 0.0406,
+      "step": 423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9414232964118893e-05,
+      "loss": 0.0572,
+      "step": 424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941071281928292e-05,
+      "loss": 0.0509,
+      "step": 425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940718245009641e-05,
+      "loss": 0.0519,
+      "step": 426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940364186039496e-05,
+      "loss": 0.0446,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9400091054025286e-05,
+      "loss": 0.0469,
+      "step": 428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939653003484518e-05,
+      "loss": 0.0492,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9392958806723544e-05,
+      "loss": 0.0395,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9389377373540373e-05,
+      "loss": 0.0408,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9385785739186746e-05,
+      "loss": 0.0522,
+      "step": 432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9382183907564824e-05,
+      "loss": 0.044,
+      "step": 433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378571882587848e-05,
+      "loss": 0.045,
+      "step": 434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937494966818014e-05,
+      "loss": 0.0514,
+      "step": 435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9371317268277075e-05,
+      "loss": 0.0423,
+      "step": 436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936767468682511e-05,
+      "loss": 0.0464,
+      "step": 437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9364021927781764e-05,
+      "loss": 0.0506,
+      "step": 438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9360358995115605e-05,
+      "loss": 0.0528,
+      "step": 439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9356685892806255e-05,
+      "loss": 0.051,
+      "step": 440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9353002624844392e-05,
+      "loss": 0.0426,
+      "step": 441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9349309195231732e-05,
+      "loss": 0.0478,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934560560798104e-05,
+      "loss": 0.0485,
+      "step": 443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9341891867116102e-05,
+      "loss": 0.0517,
+      "step": 444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9338167976671747e-05,
+      "loss": 0.0481,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933443394069383e-05,
+      "loss": 0.0318,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9330689763239225e-05,
+      "loss": 0.0493,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9326935448375822e-05,
+      "loss": 0.0416,
+      "step": 448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9323171000182538e-05,
+      "loss": 0.0424,
+      "step": 449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9319396422749284e-05,
+      "loss": 0.0507,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9315611720176982e-05,
+      "loss": 0.0459,
+      "step": 451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931181689657756e-05,
+      "loss": 0.0526,
+      "step": 452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9308011956073933e-05,
+      "loss": 0.0468,
+      "step": 453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9304196902800014e-05,
+      "loss": 0.0507,
+      "step": 454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9300371740900698e-05,
+      "loss": 0.0517,
+      "step": 455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9296536474531864e-05,
+      "loss": 0.0402,
+      "step": 456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9292691107860374e-05,
+      "loss": 0.0546,
+      "step": 457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928883564506406e-05,
+      "loss": 0.0464,
+      "step": 458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928497009033172e-05,
+      "loss": 0.0497,
+      "step": 459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9281094447863126e-05,
+      "loss": 0.0504,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927720872186899e-05,
+      "loss": 0.0379,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9273312916571008e-05,
+      "loss": 0.0557,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9269407036201803e-05,
+      "loss": 0.0482,
+      "step": 463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265491085004956e-05,
+      "loss": 0.0475,
+      "step": 464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926156506723498e-05,
+      "loss": 0.0494,
+      "step": 465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9257628987157343e-05,
+      "loss": 0.0441,
+      "step": 466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9253682849048417e-05,
+      "loss": 0.0513,
+      "step": 467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9249726657195534e-05,
+      "loss": 0.0455,
+      "step": 468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9245760415896923e-05,
+      "loss": 0.0356,
+      "step": 469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924178412946174e-05,
+      "loss": 0.0433,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9237797802210067e-05,
+      "loss": 0.0418,
+      "step": 471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9233801438472875e-05,
+      "loss": 0.0485,
+      "step": 472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922979504259205e-05,
+      "loss": 0.0618,
+      "step": 473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922577861892037e-05,
+      "loss": 0.0522,
+      "step": 474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9221752171821523e-05,
+      "loss": 0.0497,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9217715705670072e-05,
+      "loss": 0.0515,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9213669224851473e-05,
+      "loss": 0.0447,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9209612733762056e-05,
+      "loss": 0.0468,
+      "step": 478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9205546236809037e-05,
+      "loss": 0.0538,
+      "step": 479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9201469738410493e-05,
+      "loss": 0.052,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919738324299537e-05,
+      "loss": 0.0452,
+      "step": 481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193286755003475e-05,
+      "loss": 0.0452,
+      "step": 482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918918027888548e-05,
+      "loss": 0.0544,
+      "step": 483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9185063819102896e-05,
+      "loss": 0.0493,
+      "step": 484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918093738012809e-05,
+      "loss": 0.0516,
+      "step": 485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9176800966444263e-05,
+      "loss": 0.042,
+      "step": 486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917265458254546e-05,
+      "loss": 0.041,
+      "step": 487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9168498232936557e-05,
+      "loss": 0.0397,
+      "step": 488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9164331922133254e-05,
+      "loss": 0.048,
+      "step": 489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9160155654662075e-05,
+      "loss": 0.0328,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9155969435060363e-05,
+      "loss": 0.037,
+      "step": 491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915177326787627e-05,
+      "loss": 0.0447,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914756715766877e-05,
+      "loss": 0.0529,
+      "step": 493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914335110900761e-05,
+      "loss": 0.0375,
+      "step": 494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9139125126473365e-05,
+      "loss": 0.043,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9134889214657385e-05,
+      "loss": 0.0441,
+      "step": 496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9130643378161817e-05,
+      "loss": 0.0512,
+      "step": 497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9126387621599585e-05,
+      "loss": 0.0452,
+      "step": 498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9122121949594392e-05,
+      "loss": 0.0536,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9117846366780714e-05,
+      "loss": 0.0417,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9113560877803798e-05,
+      "loss": 0.0508,
+      "step": 501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9109265487319645e-05,
+      "loss": 0.0344,
+      "step": 502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910496019999502e-05,
+      "loss": 0.0388,
+      "step": 503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9100645020507448e-05,
+      "loss": 0.0466,
+      "step": 504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.0451,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9091985003807242e-05,
+      "loss": 0.0484,
+      "step": 506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9087640176003356e-05,
+      "loss": 0.0435,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9083285474854012e-05,
+      "loss": 0.0351,
+      "step": 508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9078920905090408e-05,
+      "loss": 0.0434,
+      "step": 509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907454647145447e-05,
+      "loss": 0.0385,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9070162178698838e-05,
+      "loss": 0.0376,
+      "step": 511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9065768031586864e-05,
+      "loss": 0.0473,
+      "step": 512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906136403489261e-05,
+      "loss": 0.0516,
+      "step": 513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905695019340083e-05,
+      "loss": 0.0438,
+      "step": 514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9052526511906993e-05,
+      "loss": 0.0399,
+      "step": 515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9048092995217234e-05,
+      "loss": 0.0472,
+      "step": 516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904364964814839e-05,
+      "loss": 0.0405,
+      "step": 517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9039196475527974e-05,
+      "loss": 0.0452,
+      "step": 518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9034733482194176e-05,
+      "loss": 0.0531,
+      "step": 519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903026067299585e-05,
+      "loss": 0.0514,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9025778052792518e-05,
+      "loss": 0.0503,
+      "step": 521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902128562645437e-05,
+      "loss": 0.0562,
+      "step": 522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901678339886223e-05,
+      "loss": 0.0464,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901227137490759e-05,
+      "loss": 0.043,
+      "step": 524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9007749559492575e-05,
+      "loss": 0.043,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9003217957529948e-05,
+      "loss": 0.0481,
+      "step": 526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899867657394311e-05,
+      "loss": 0.0369,
+      "step": 527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899412541366609e-05,
+      "loss": 0.0448,
+      "step": 528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898956448164353e-05,
+      "loss": 0.0481,
+      "step": 529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8984993782830695e-05,
+      "loss": 0.0501,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8980413322193462e-05,
+      "loss": 0.0507,
+      "step": 531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8975823104708314e-05,
+      "loss": 0.0427,
+      "step": 532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8971223135362328e-05,
+      "loss": 0.0463,
+      "step": 533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896661341915318e-05,
+      "loss": 0.0423,
+      "step": 534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8961993961089145e-05,
+      "loss": 0.0406,
+      "step": 535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895736476618906e-05,
+      "loss": 0.037,
+      "step": 536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8952725839482368e-05,
+      "loss": 0.0439,
+      "step": 537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894807718600906e-05,
+      "loss": 0.046,
+      "step": 538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8943418810819715e-05,
+      "loss": 0.046,
+      "step": 539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8938750718975456e-05,
+      "loss": 0.0416,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8934072915547978e-05,
+      "loss": 0.0468,
+      "step": 541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892938540561952e-05,
+      "loss": 0.0461,
+      "step": 542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8924688194282862e-05,
+      "loss": 0.0416,
+      "step": 543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8919981286641336e-05,
+      "loss": 0.0448,
+      "step": 544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891526468780881e-05,
+      "loss": 0.0417,
+      "step": 545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8910538402909658e-05,
+      "loss": 0.0438,
+      "step": 546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.89058024370788e-05,
+      "loss": 0.0446,
+      "step": 547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890105679546167e-05,
+      "loss": 0.0426,
+      "step": 548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8896301483214207e-05,
+      "loss": 0.0413,
+      "step": 549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8891536505502865e-05,
+      "loss": 0.0546,
+      "step": 550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8886761867504595e-05,
+      "loss": 0.043,
+      "step": 551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8881977574406842e-05,
+      "loss": 0.0505,
+      "step": 552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887718363140754e-05,
+      "loss": 0.0439,
+      "step": 553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8872380043715117e-05,
+      "loss": 0.037,
+      "step": 554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8867566816548468e-05,
+      "loss": 0.0423,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8862743955136966e-05,
+      "loss": 0.0482,
+      "step": 556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885791146472045e-05,
+      "loss": 0.0442,
+      "step": 557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8853069350549223e-05,
+      "loss": 0.0487,
+      "step": 558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884821761788404e-05,
+      "loss": 0.0397,
+      "step": 559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8843356271996106e-05,
+      "loss": 0.0469,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8838485318167078e-05,
+      "loss": 0.0402,
+      "step": 561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8833604761689045e-05,
+      "loss": 0.042,
+      "step": 562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8828714607864524e-05,
+      "loss": 0.0497,
+      "step": 563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882381486200647e-05,
+      "loss": 0.0489,
+      "step": 564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881890552943826e-05,
+      "loss": 0.0421,
+      "step": 565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8813986615493664e-05,
+      "loss": 0.0475,
+      "step": 566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8809058125516894e-05,
+      "loss": 0.0385,
+      "step": 567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8804120064862547e-05,
+      "loss": 0.0487,
+      "step": 568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879917243889562e-05,
+      "loss": 0.0466,
+      "step": 569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8794215252991504e-05,
+      "loss": 0.0429,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8789248512535978e-05,
+      "loss": 0.0441,
+      "step": 571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87842722229252e-05,
+      "loss": 0.0594,
+      "step": 572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87792863895657e-05,
+      "loss": 0.0425,
+      "step": 573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8774291017874384e-05,
+      "loss": 0.0373,
+      "step": 574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8769286113278515e-05,
+      "loss": 0.0415,
+      "step": 575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8764271681215714e-05,
+      "loss": 0.04,
+      "step": 576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8759247727133956e-05,
+      "loss": 0.0459,
+      "step": 577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8754214256491564e-05,
+      "loss": 0.0406,
+      "step": 578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8749171274757184e-05,
+      "loss": 0.0409,
+      "step": 579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874411878740982e-05,
+      "loss": 0.0445,
+      "step": 580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873905679993878e-05,
+      "loss": 0.0396,
+      "step": 581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873398531784371e-05,
+      "loss": 0.0454,
+      "step": 582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8728904346634563e-05,
+      "loss": 0.0423,
+      "step": 583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872381389183161e-05,
+      "loss": 0.0359,
+      "step": 584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8718713958965413e-05,
+      "loss": 0.0397,
+      "step": 585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8713604553576842e-05,
+      "loss": 0.0433,
+      "step": 586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870848568121705e-05,
+      "loss": 0.044,
+      "step": 587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8703357347447487e-05,
+      "loss": 0.0372,
+      "step": 588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8698219557839875e-05,
+      "loss": 0.0405,
+      "step": 589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8693072317976205e-05,
+      "loss": 0.0479,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868791563344874e-05,
+      "loss": 0.0448,
+      "step": 591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868274950986001e-05,
+      "loss": 0.0522,
+      "step": 592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8677573952822793e-05,
+      "loss": 0.0414,
+      "step": 593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867238896796012e-05,
+      "loss": 0.033,
+      "step": 594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8667194560905255e-05,
+      "loss": 0.0415,
+      "step": 595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8661990737301714e-05,
+      "loss": 0.0485,
+      "step": 596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865677750280323e-05,
+      "loss": 0.049,
+      "step": 597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8651554863073776e-05,
+      "loss": 0.0461,
+      "step": 598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864632282378753e-05,
+      "loss": 0.0359,
+      "step": 599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864108139062888e-05,
+      "loss": 0.053,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8635830569292434e-05,
+      "loss": 0.042,
+      "step": 601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863057036548299e-05,
+      "loss": 0.0436,
+      "step": 602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862530078491554e-05,
+      "loss": 0.0458,
+      "step": 603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862002183331527e-05,
+      "loss": 0.0373,
+      "step": 604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8614733516417543e-05,
+      "loss": 0.0406,
+      "step": 605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860943583996789e-05,
+      "loss": 0.0462,
+      "step": 606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8604128809722024e-05,
+      "loss": 0.0398,
+      "step": 607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859881243144581e-05,
+      "loss": 0.0406,
+      "step": 608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8593486710915277e-05,
+      "loss": 0.0479,
+      "step": 609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8588151653916596e-05,
+      "loss": 0.0433,
+      "step": 610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858280726624609e-05,
+      "loss": 0.0444,
+      "step": 611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8577453553710213e-05,
+      "loss": 0.039,
+      "step": 612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8572090522125553e-05,
+      "loss": 0.0479,
+      "step": 613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8566718177318825e-05,
+      "loss": 0.04,
+      "step": 614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.856133652512685e-05,
+      "loss": 0.0516,
+      "step": 615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8555945571396584e-05,
+      "loss": 0.0596,
+      "step": 616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8550545321985065e-05,
+      "loss": 0.0491,
+      "step": 617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8545135782759446e-05,
+      "loss": 0.0437,
+      "step": 618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8539716959596967e-05,
+      "loss": 0.0363,
+      "step": 619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853428885838495e-05,
+      "loss": 0.0453,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852885148502081e-05,
+      "loss": 0.039,
+      "step": 621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8523404845412028e-05,
+      "loss": 0.0449,
+      "step": 622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851794894547615e-05,
+      "loss": 0.0403,
+      "step": 623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8512483791140787e-05,
+      "loss": 0.0438,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85070093883436e-05,
+      "loss": 0.041,
+      "step": 625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8501525743032305e-05,
+      "loss": 0.0446,
+      "step": 626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8496032861164657e-05,
+      "loss": 0.0442,
+      "step": 627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8490530748708442e-05,
+      "loss": 0.0404,
+      "step": 628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8485019411641477e-05,
+      "loss": 0.0373,
+      "step": 629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8479498855951607e-05,
+      "loss": 0.0387,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8473969087636683e-05,
+      "loss": 0.0412,
+      "step": 631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8468430112704574e-05,
+      "loss": 0.0367,
+      "step": 632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8462881937173144e-05,
+      "loss": 0.0485,
+      "step": 633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8457324567070255e-05,
+      "loss": 0.041,
+      "step": 634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845175800843376e-05,
+      "loss": 0.0447,
+      "step": 635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84461822673115e-05,
+      "loss": 0.0449,
+      "step": 636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8440597349761286e-05,
+      "loss": 0.0464,
+      "step": 637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8435003261850895e-05,
+      "loss": 0.0451,
+      "step": 638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8429400009658076e-05,
+      "loss": 0.0437,
+      "step": 639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842378759927053e-05,
+      "loss": 0.0468,
+      "step": 640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8418166036785912e-05,
+      "loss": 0.0375,
+      "step": 641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.0459,
+      "step": 642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8406895479965766e-05,
+      "loss": 0.051,
+      "step": 643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8401246497875238e-05,
+      "loss": 0.0364,
+      "step": 644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8395588388177606e-05,
+      "loss": 0.0579,
+      "step": 645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.838992115702018e-05,
+      "loss": 0.0409,
+      "step": 646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8384244810560166e-05,
+      "loss": 0.0458,
+      "step": 647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8378559354964686e-05,
+      "loss": 0.04,
+      "step": 648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8372864796410748e-05,
+      "loss": 0.0439,
+      "step": 649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8367161141085256e-05,
+      "loss": 0.0424,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8361448395184995e-05,
+      "loss": 0.0462,
+      "step": 651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835572656491663e-05,
+      "loss": 0.0393,
+      "step": 652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8349995656496692e-05,
+      "loss": 0.0401,
+      "step": 653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8344255676151572e-05,
+      "loss": 0.0455,
+      "step": 654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8338506630117527e-05,
+      "loss": 0.0386,
+      "step": 655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8332748524640657e-05,
+      "loss": 0.0347,
+      "step": 656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8326981365976903e-05,
+      "loss": 0.0463,
+      "step": 657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832120516039205e-05,
+      "loss": 0.0459,
+      "step": 658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8315419914161697e-05,
+      "loss": 0.0347,
+      "step": 659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8309625633571283e-05,
+      "loss": 0.0503,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.830382232491605e-05,
+      "loss": 0.0398,
+      "step": 661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829800999450106e-05,
+      "loss": 0.0588,
+      "step": 662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829218864864116e-05,
+      "loss": 0.0437,
+      "step": 663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8286358293661007e-05,
+      "loss": 0.0505,
+      "step": 664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828051893589504e-05,
+      "loss": 0.03,
+      "step": 665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827467058168748e-05,
+      "loss": 0.0431,
+      "step": 666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8268813237392323e-05,
+      "loss": 0.0617,
+      "step": 667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826294690937333e-05,
+      "loss": 0.0325,
+      "step": 668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8257071604004025e-05,
+      "loss": 0.0367,
+      "step": 669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8251187327667682e-05,
+      "loss": 0.0485,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.824529408675733e-05,
+      "loss": 0.0426,
+      "step": 671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8239391887675725e-05,
+      "loss": 0.036,
+      "step": 672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8233480736835366e-05,
+      "loss": 0.0613,
+      "step": 673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822756064065847e-05,
+      "loss": 0.037,
+      "step": 674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8221631605576983e-05,
+      "loss": 0.0358,
+      "step": 675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8215693638032555e-05,
+      "loss": 0.0406,
+      "step": 676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8209746744476538e-05,
+      "loss": 0.0413,
+      "step": 677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820379093136999e-05,
+      "loss": 0.0468,
+      "step": 678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8197826205183657e-05,
+      "loss": 0.0416,
+      "step": 679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819185257239796e-05,
+      "loss": 0.0378,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8185870039503014e-05,
+      "loss": 0.0543,
+      "step": 681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8179878612998586e-05,
+      "loss": 0.0332,
+      "step": 682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8173878299394118e-05,
+      "loss": 0.0399,
+      "step": 683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8167869105208703e-05,
+      "loss": 0.0491,
+      "step": 684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816185103697108e-05,
+      "loss": 0.045,
+      "step": 685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8155824101219626e-05,
+      "loss": 0.0404,
+      "step": 686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814978830450237e-05,
+      "loss": 0.0388,
+      "step": 687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8143743653376944e-05,
+      "loss": 0.0361,
+      "step": 688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8137690154410617e-05,
+      "loss": 0.0452,
+      "step": 689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8131627814180264e-05,
+      "loss": 0.0355,
+      "step": 690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8125556639272375e-05,
+      "loss": 0.0376,
+      "step": 691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8119476636283018e-05,
+      "loss": 0.0482,
+      "step": 692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811338781181787e-05,
+      "loss": 0.0425,
+      "step": 693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8107290172492193e-05,
+      "loss": 0.0379,
+      "step": 694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8101183724930812e-05,
+      "loss": 0.0347,
+      "step": 695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809506847576813e-05,
+      "loss": 0.0429,
+      "step": 696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8088944431648118e-05,
+      "loss": 0.0421,
+      "step": 697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8082811599224293e-05,
+      "loss": 0.0337,
+      "step": 698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8076669985159726e-05,
+      "loss": 0.0347,
+      "step": 699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8070519596127027e-05,
+      "loss": 0.0458,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8064360438808335e-05,
+      "loss": 0.0515,
+      "step": 701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805819251989533e-05,
+      "loss": 0.0435,
+      "step": 702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8052015846089187e-05,
+      "loss": 0.0387,
+      "step": 703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804583042410062e-05,
+      "loss": 0.0452,
+      "step": 704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8039636260649823e-05,
+      "loss": 0.0461,
+      "step": 705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8033433362466506e-05,
+      "loss": 0.0417,
+      "step": 706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.802722173628986e-05,
+      "loss": 0.0363,
+      "step": 707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.802100138886856e-05,
+      "loss": 0.0362,
+      "step": 708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8014772326960758e-05,
+      "loss": 0.0357,
+      "step": 709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8008534557334064e-05,
+      "loss": 0.0367,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800228808676557e-05,
+      "loss": 0.0404,
+      "step": 711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7996032922041797e-05,
+      "loss": 0.044,
+      "step": 712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7989769069958728e-05,
+      "loss": 0.0404,
+      "step": 713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798349653732178e-05,
+      "loss": 0.0558,
+      "step": 714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79772153309458e-05,
+      "loss": 0.0535,
+      "step": 715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7970925457655055e-05,
+      "loss": 0.0514,
+      "step": 716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7964626924283238e-05,
+      "loss": 0.0541,
+      "step": 717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7958319737673444e-05,
+      "loss": 0.0461,
+      "step": 718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795200390467817e-05,
+      "loss": 0.0387,
+      "step": 719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7945679432159302e-05,
+      "loss": 0.0401,
+      "step": 720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7939346326988127e-05,
+      "loss": 0.0457,
+      "step": 721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7933004596045294e-05,
+      "loss": 0.0455,
+      "step": 722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7926654246220843e-05,
+      "loss": 0.0364,
+      "step": 723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7920295284414153e-05,
+      "loss": 0.0485,
+      "step": 724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791392771753398e-05,
+      "loss": 0.0411,
+      "step": 725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7907551552498423e-05,
+      "loss": 0.0401,
+      "step": 726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7901166796234922e-05,
+      "loss": 0.0366,
+      "step": 727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789477345568025e-05,
+      "loss": 0.0332,
+      "step": 728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78883715377805e-05,
+      "loss": 0.0443,
+      "step": 729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78819610494911e-05,
+      "loss": 0.0433,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7875541997776783e-05,
+      "loss": 0.0388,
+      "step": 731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7869114389611574e-05,
+      "loss": 0.0383,
+      "step": 732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786267823197881e-05,
+      "loss": 0.0428,
+      "step": 733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785623353187111e-05,
+      "loss": 0.0415,
+      "step": 734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7849780296290368e-05,
+      "loss": 0.0362,
+      "step": 735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7843318532247767e-05,
+      "loss": 0.0409,
+      "step": 736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783684824676374e-05,
+      "loss": 0.0368,
+      "step": 737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783036944686799e-05,
+      "loss": 0.0368,
+      "step": 738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782388213959946e-05,
+      "loss": 0.0386,
+      "step": 739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7817386332006346e-05,
+      "loss": 0.0452,
+      "step": 740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.781088203114607e-05,
+      "loss": 0.0543,
+      "step": 741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7804369244085283e-05,
+      "loss": 0.0426,
+      "step": 742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7797847977899873e-05,
+      "loss": 0.0376,
+      "step": 743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7791318239674914e-05,
+      "loss": 0.0389,
+      "step": 744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7784780036504696e-05,
+      "loss": 0.0314,
+      "step": 745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7778233375492716e-05,
+      "loss": 0.0474,
+      "step": 746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777167826375164e-05,
+      "loss": 0.0433,
+      "step": 747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7765114708403332e-05,
+      "loss": 0.0472,
+      "step": 748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7758542716578826e-05,
+      "loss": 0.0387,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7751962295418314e-05,
+      "loss": 0.0436,
+      "step": 750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7745373452071154e-05,
+      "loss": 0.0552,
+      "step": 751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7738776193695853e-05,
+      "loss": 0.0429,
+      "step": 752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7732170527460058e-05,
+      "loss": 0.0374,
+      "step": 753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7725556460540553e-05,
+      "loss": 0.0393,
+      "step": 754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7718934000123247e-05,
+      "loss": 0.0418,
+      "step": 755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7712303153403167e-05,
+      "loss": 0.0438,
+      "step": 756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7705663927584464e-05,
+      "loss": 0.0363,
+      "step": 757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769901632988037e-05,
+      "loss": 0.0488,
+      "step": 758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7692360367513238e-05,
+      "loss": 0.0428,
+      "step": 759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7685696047714486e-05,
+      "loss": 0.0448,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7679023377724623e-05,
+      "loss": 0.044,
+      "step": 761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7672342364793233e-05,
+      "loss": 0.0383,
+      "step": 762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7665653016178957e-05,
+      "loss": 0.0387,
+      "step": 763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7658955339149497e-05,
+      "loss": 0.0404,
+      "step": 764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7652249340981608e-05,
+      "loss": 0.0349,
+      "step": 765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.764553502896107e-05,
+      "loss": 0.0494,
+      "step": 766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7638812410382714e-05,
+      "loss": 0.0411,
+      "step": 767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763208149255038e-05,
+      "loss": 0.0462,
+      "step": 768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7625342282776935e-05,
+      "loss": 0.0425,
+      "step": 769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7618594788384252e-05,
+      "loss": 0.0413,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7611839016703208e-05,
+      "loss": 0.0385,
+      "step": 771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7605074975073667e-05,
+      "loss": 0.0358,
+      "step": 772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.759830267084448e-05,
+      "loss": 0.0402,
+      "step": 773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7591522111373475e-05,
+      "loss": 0.0529,
+      "step": 774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758473330402745e-05,
+      "loss": 0.0355,
+      "step": 775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757793625618217e-05,
+      "loss": 0.0505,
+      "step": 776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7571130975222335e-05,
+      "loss": 0.0409,
+      "step": 777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7564317468541608e-05,
+      "loss": 0.0387,
+      "step": 778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.0431,
+      "step": 779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7550665807636783e-05,
+      "loss": 0.0456,
+      "step": 780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7543827668244648e-05,
+      "loss": 0.0375,
+      "step": 781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7536981332795535e-05,
+      "loss": 0.0419,
+      "step": 782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7530126808727703e-05,
+      "loss": 0.0397,
+      "step": 783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7523264103488314e-05,
+      "loss": 0.0348,
+      "step": 784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7516393224533406e-05,
+      "loss": 0.0485,
+      "step": 785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7509514179327914e-05,
+      "loss": 0.046,
+      "step": 786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.750262697534563e-05,
+      "loss": 0.0338,
+      "step": 787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7495731620069216e-05,
+      "loss": 0.0404,
+      "step": 788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7488828120990194e-05,
+      "loss": 0.0445,
+      "step": 789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748191648560893e-05,
+      "loss": 0.0503,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7474996721434628e-05,
+      "loss": 0.0317,
+      "step": 791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7468068835985326e-05,
+      "loss": 0.0425,
+      "step": 792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746113283678789e-05,
+      "loss": 0.042,
+      "step": 793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7454188731377988e-05,
+      "loss": 0.0363,
+      "step": 794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744723652730011e-05,
+      "loss": 0.0408,
+      "step": 795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7440276232107537e-05,
+      "loss": 0.0372,
+      "step": 796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7433307853362344e-05,
+      "loss": 0.0367,
+      "step": 797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742633139863538e-05,
+      "loss": 0.0371,
+      "step": 798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7419346875506282e-05,
+      "loss": 0.0451,
+      "step": 799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7412354291563442e-05,
+      "loss": 0.0496,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7405353654404017e-05,
+      "loss": 0.0274,
+      "step": 801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7398344971633902e-05,
+      "loss": 0.0385,
+      "step": 802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7391328250867748e-05,
+      "loss": 0.052,
+      "step": 803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738430349972893e-05,
+      "loss": 0.0369,
+      "step": 804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737727072584955e-05,
+      "loss": 0.0354,
+      "step": 805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7370229936870425e-05,
+      "loss": 0.043,
+      "step": 806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7363181140441086e-05,
+      "loss": 0.0345,
+      "step": 807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7356124344219757e-05,
+      "loss": 0.039,
+      "step": 808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7349059555873348e-05,
+      "loss": 0.0416,
+      "step": 809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7341986783077466e-05,
+      "loss": 0.0414,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7334906033516388e-05,
+      "loss": 0.0318,
+      "step": 811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7327817314883055e-05,
+      "loss": 0.035,
+      "step": 812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.732072063487906e-05,
+      "loss": 0.0426,
+      "step": 813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7313616001214666e-05,
+      "loss": 0.0397,
+      "step": 814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730650342160875e-05,
+      "loss": 0.0474,
+      "step": 815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7299382903788844e-05,
+      "loss": 0.0411,
+      "step": 816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7292254455491095e-05,
+      "loss": 0.0317,
+      "step": 817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7285118084460265e-05,
+      "loss": 0.0511,
+      "step": 818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727797379844972e-05,
+      "loss": 0.0358,
+      "step": 819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7270821605221448e-05,
+      "loss": 0.0346,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7263661512546e-05,
+      "loss": 0.0336,
+      "step": 821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7256493528202523e-05,
+      "loss": 0.0419,
+      "step": 822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7249317659978736e-05,
+      "loss": 0.0447,
+      "step": 823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7242133915670923e-05,
+      "loss": 0.0414,
+      "step": 824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7234942303083925e-05,
+      "loss": 0.0433,
+      "step": 825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7227742830031132e-05,
+      "loss": 0.0472,
+      "step": 826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7220535504334477e-05,
+      "loss": 0.0384,
+      "step": 827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7213320333824417e-05,
+      "loss": 0.0402,
+      "step": 828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7206097326339937e-05,
+      "loss": 0.0488,
+      "step": 829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7198866489728537e-05,
+      "loss": 0.0374,
+      "step": 830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7191627831846226e-05,
+      "loss": 0.0364,
+      "step": 831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71843813605575e-05,
+      "loss": 0.0381,
+      "step": 832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7177127083735352e-05,
+      "loss": 0.0362,
+      "step": 833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716986500926126e-05,
+      "loss": 0.038,
+      "step": 834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716259514502516e-05,
+      "loss": 0.046,
+      "step": 835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.715531749892546e-05,
+      "loss": 0.0434,
+      "step": 836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7148032078869027e-05,
+      "loss": 0.0474,
+      "step": 837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714073889277117e-05,
+      "loss": 0.0348,
+      "step": 838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.713343794855562e-05,
+      "loss": 0.0338,
+      "step": 839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7126129254154568e-05,
+      "loss": 0.0307,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7118812817508603e-05,
+      "loss": 0.0458,
+      "step": 841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7111488646566728e-05,
+      "loss": 0.045,
+      "step": 842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7104156749286352e-05,
+      "loss": 0.0367,
+      "step": 843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7096817133633283e-05,
+      "loss": 0.0465,
+      "step": 844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7089469807581705e-05,
+      "loss": 0.0377,
+      "step": 845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708211477911419e-05,
+      "loss": 0.0413,
+      "step": 846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7074752056221663e-05,
+      "loss": 0.0444,
+      "step": 847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7067381646903426e-05,
+      "loss": 0.0372,
+      "step": 848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706000355916712e-05,
+      "loss": 0.0395,
+      "step": 849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705261780102874e-05,
+      "loss": 0.0456,
+      "step": 850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7045224380512597e-05,
+      "loss": 0.0484,
+      "step": 851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7037823305651345e-05,
+      "loss": 0.0391,
+      "step": 852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7030414584485938e-05,
+      "loss": 0.0439,
+      "step": 853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.702299822506565e-05,
+      "loss": 0.0376,
+      "step": 854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701557423544805e-05,
+      "loss": 0.0432,
+      "step": 855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700814262369899e-05,
+      "loss": 0.0436,
+      "step": 856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700070339789262e-05,
+      "loss": 0.0357,
+      "step": 857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699325656611134e-05,
+      "loss": 0.0346,
+      "step": 858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6985802136445832e-05,
+      "loss": 0.0342,
+      "step": 859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6978340116995022e-05,
+      "loss": 0.0298,
+      "step": 860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6970870515866086e-05,
+      "loss": 0.045,
+      "step": 861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6963393341174438e-05,
+      "loss": 0.0357,
+      "step": 862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6955908601043724e-05,
+      "loss": 0.0437,
+      "step": 863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6948416303605796e-05,
+      "loss": 0.0377,
+      "step": 864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694091645700073e-05,
+      "loss": 0.0331,
+      "step": 865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6933409069376797e-05,
+      "loss": 0.0359,
+      "step": 866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6925894148890473e-05,
+      "loss": 0.0436,
+      "step": 867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691837170370639e-05,
+      "loss": 0.0412,
+      "step": 868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6910841741997395e-05,
+      "loss": 0.039,
+      "step": 869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6903304271944467e-05,
+      "loss": 0.0363,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6895759301736757e-05,
+      "loss": 0.0516,
+      "step": 871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688820683957156e-05,
+      "loss": 0.0418,
+      "step": 872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6880646893654327e-05,
+      "loss": 0.0447,
+      "step": 873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6873079472198613e-05,
+      "loss": 0.032,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6865504583426117e-05,
+      "loss": 0.0401,
+      "step": 875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6857922235566642e-05,
+      "loss": 0.0356,
+      "step": 876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6850332436858087e-05,
+      "loss": 0.0452,
+      "step": 877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6842735195546466e-05,
+      "loss": 0.0381,
+      "step": 878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683513051988586e-05,
+      "loss": 0.0392,
+      "step": 879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682751841813844e-05,
+      "loss": 0.0374,
+      "step": 880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681989889857444e-05,
+      "loss": 0.033,
+      "step": 881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6812271969472145e-05,
+      "loss": 0.0312,
+      "step": 882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680463763911791e-05,
+      "loss": 0.0442,
+      "step": 883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.679699591580611e-05,
+      "loss": 0.0389,
+      "step": 884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6789346807839165e-05,
+      "loss": 0.0374,
+      "step": 885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6781690323527512e-05,
+      "loss": 0.0442,
+      "step": 886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677402647118961e-05,
+      "loss": 0.0431,
+      "step": 887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676635525915191e-05,
+      "loss": 0.0432,
+      "step": 888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675867669574888e-05,
+      "loss": 0.0372,
+      "step": 889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6750990789322948e-05,
+      "loss": 0.0389,
+      "step": 890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6743297548224543e-05,
+      "loss": 0.0346,
+      "step": 891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673559698081205e-05,
+      "loss": 0.0434,
+      "step": 892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6727889095451814e-05,
+      "loss": 0.0439,
+      "step": 893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6720173900518137e-05,
+      "loss": 0.041,
+      "step": 894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671245140439326e-05,
+      "loss": 0.0461,
+      "step": 895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670472161546736e-05,
+      "loss": 0.0407,
+      "step": 896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669698454213852e-05,
+      "loss": 0.0405,
+      "step": 897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6689240192812765e-05,
+      "loss": 0.0433,
+      "step": 898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6681488575904003e-05,
+      "loss": 0.039,
+      "step": 899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6673729699834046e-05,
+      "loss": 0.0408,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6665963573032593e-05,
+      "loss": 0.044,
+      "step": 901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6658190203937218e-05,
+      "loss": 0.0331,
+      "step": 902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650409600993373e-05,
+      "loss": 0.0485,
+      "step": 903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6642621772654347e-05,
+      "loss": 0.0409,
+      "step": 904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663482672738131e-05,
+      "loss": 0.0422,
+      "step": 905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6627024473643244e-05,
+      "loss": 0.0423,
+      "step": 906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6619215019916983e-05,
+      "loss": 0.0404,
+      "step": 907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6611398374687172e-05,
+      "loss": 0.0349,
+      "step": 908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6603574546446274e-05,
+      "loss": 0.0422,
+      "step": 909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659574354369455e-05,
+      "loss": 0.0371,
+      "step": 910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6587905374940067e-05,
+      "loss": 0.0353,
+      "step": 911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6580060048698673e-05,
+      "loss": 0.0426,
+      "step": 912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6572207573493987e-05,
+      "loss": 0.0451,
+      "step": 913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6564347957857395e-05,
+      "loss": 0.0408,
+      "step": 914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6556481210328053e-05,
+      "loss": 0.0421,
+      "step": 915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.0319,
+      "step": 916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6540726353786428e-05,
+      "loss": 0.0369,
+      "step": 917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6532838261891145e-05,
+      "loss": 0.0361,
+      "step": 918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6524943072337094e-05,
+      "loss": 0.0453,
+      "step": 919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6517040793702066e-05,
+      "loss": 0.0381,
+      "step": 920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6509131434571564e-05,
+      "loss": 0.0423,
+      "step": 921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6501215003538776e-05,
+      "loss": 0.0382,
+      "step": 922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6493291509204582e-05,
+      "loss": 0.0419,
+      "step": 923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6485360960177525e-05,
+      "loss": 0.0336,
+      "step": 924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6477423365073825e-05,
+      "loss": 0.042,
+      "step": 925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646947873251735e-05,
+      "loss": 0.048,
+      "step": 926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6461527071139613e-05,
+      "loss": 0.0372,
+      "step": 927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645356838957976e-05,
+      "loss": 0.0381,
+      "step": 928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644560269648458e-05,
+      "loss": 0.0376,
+      "step": 929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6437630000508466e-05,
+      "loss": 0.0483,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642965031031342e-05,
+      "loss": 0.0391,
+      "step": 931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6421663634569045e-05,
+      "loss": 0.032,
+      "step": 932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641366998195254e-05,
+      "loss": 0.0383,
+      "step": 933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6405669361148668e-05,
+      "loss": 0.0366,
+      "step": 934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6397661780849785e-05,
+      "loss": 0.0391,
+      "step": 935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6389647249755788e-05,
+      "loss": 0.0356,
+      "step": 936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6381625776574135e-05,
+      "loss": 0.0405,
+      "step": 937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6373597370019825e-05,
+      "loss": 0.0444,
+      "step": 938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6365562038815394e-05,
+      "loss": 0.0457,
+      "step": 939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6357519791690893e-05,
+      "loss": 0.035,
+      "step": 940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634947063738389e-05,
+      "loss": 0.0344,
+      "step": 941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6341414584639464e-05,
+      "loss": 0.049,
+      "step": 942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633335164221018e-05,
+      "loss": 0.0423,
+      "step": 943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632528181885609e-05,
+      "loss": 0.0453,
+      "step": 944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6317205123344725e-05,
+      "loss": 0.0329,
+      "step": 945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6309121564451084e-05,
+      "loss": 0.0377,
+      "step": 946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6301031150957616e-05,
+      "loss": 0.0352,
+      "step": 947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6292933891654217e-05,
+      "loss": 0.0328,
+      "step": 948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6284829795338235e-05,
+      "loss": 0.0456,
+      "step": 949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627671887081443e-05,
+      "loss": 0.0386,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6268601126894986e-05,
+      "loss": 0.0357,
+      "step": 951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6260476572399494e-05,
+      "loss": 0.0416,
+      "step": 952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625234521615495e-05,
+      "loss": 0.0408,
+      "step": 953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6244207066995737e-05,
+      "loss": 0.0312,
+      "step": 954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6236062133763617e-05,
+      "loss": 0.0306,
+      "step": 955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6227910425307723e-05,
+      "loss": 0.036,
+      "step": 956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6219751950484547e-05,
+      "loss": 0.0256,
+      "step": 957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6211586718157935e-05,
+      "loss": 0.0351,
+      "step": 958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6203414737199083e-05,
+      "loss": 0.03,
+      "step": 959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61952360164865e-05,
+      "loss": 0.0302,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618705056490603e-05,
+      "loss": 0.0369,
+      "step": 961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6178858391350835e-05,
+      "loss": 0.0402,
+      "step": 962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6170659504721365e-05,
+      "loss": 0.0407,
+      "step": 963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6162453913925373e-05,
+      "loss": 0.0395,
+      "step": 964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6154241627877894e-05,
+      "loss": 0.0389,
+      "step": 965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614602265550124e-05,
+      "loss": 0.0327,
+      "step": 966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.613779700572499e-05,
+      "loss": 0.0384,
+      "step": 967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.612956468748596e-05,
+      "loss": 0.0488,
+      "step": 968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6121325709728228e-05,
+      "loss": 0.0415,
+      "step": 969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.611308008140311e-05,
+      "loss": 0.0417,
+      "step": 970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.610482781146913e-05,
+      "loss": 0.0376,
+      "step": 971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6096568908892048e-05,
+      "loss": 0.0357,
+      "step": 972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608830338264481e-05,
+      "loss": 0.0279,
+      "step": 973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608003124170758e-05,
+      "loss": 0.0421,
+      "step": 974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6071752495067684e-05,
+      "loss": 0.0484,
+      "step": 975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6063467151719653e-05,
+      "loss": 0.0417,
+      "step": 976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6055175220665158e-05,
+      "loss": 0.033,
+      "step": 977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6046876710913043e-05,
+      "loss": 0.0478,
+      "step": 978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60385716314793e-05,
+      "loss": 0.0422,
+      "step": 979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603025999138705e-05,
+      "loss": 0.0366,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.602194179966655e-05,
+      "loss": 0.0372,
+      "step": 981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6013617065355167e-05,
+      "loss": 0.0459,
+      "step": 982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6005285797497383e-05,
+      "loss": 0.0413,
+      "step": 983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5996948005144774e-05,
+      "loss": 0.0495,
+      "step": 984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.598860369735601e-05,
+      "loss": 0.0335,
+      "step": 985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.598025288319684e-05,
+      "loss": 0.0385,
+      "step": 986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5971895571740072e-05,
+      "loss": 0.0402,
+      "step": 987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.596353177206558e-05,
+      "loss": 0.0396,
+      "step": 988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5955161493260288e-05,
+      "loss": 0.0365,
+      "step": 989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5946784744418162e-05,
+      "loss": 0.039,
+      "step": 990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5938401534640185e-05,
+      "loss": 0.0434,
+      "step": 991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5930011873034377e-05,
+      "loss": 0.0398,
+      "step": 992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.592161576871575e-05,
+      "loss": 0.037,
+      "step": 993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591321323080633e-05,
+      "loss": 0.0325,
+      "step": 994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5904804268435127e-05,
+      "loss": 0.0418,
+      "step": 995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.589638889073813e-05,
+      "loss": 0.0327,
+      "step": 996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5887967106858296e-05,
+      "loss": 0.0393,
+      "step": 997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587953892594555e-05,
+      "loss": 0.0332,
+      "step": 998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587110435715676e-05,
+      "loss": 0.0416,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5862663409655733e-05,
+      "loss": 0.0339,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585421609261322e-05,
+      "loss": 0.0344,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.584576241520687e-05,
+      "loss": 0.0434,
+      "step": 1002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583730238662126e-05,
+      "loss": 0.0399,
+      "step": 1003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.582883601604786e-05,
+      "loss": 0.0346,
+      "step": 1004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5820363312685032e-05,
+      "loss": 0.038,
+      "step": 1005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581188428573802e-05,
+      "loss": 0.039,
+      "step": 1006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5803398944418934e-05,
+      "loss": 0.0428,
+      "step": 1007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.579490729794675e-05,
+      "loss": 0.0451,
+      "step": 1008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578640935554728e-05,
+      "loss": 0.0378,
+      "step": 1009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5777905126453204e-05,
+      "loss": 0.0449,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5769394619904002e-05,
+      "loss": 0.043,
+      "step": 1011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5760877845145996e-05,
+      "loss": 0.0338,
+      "step": 1012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.57523548114323e-05,
+      "loss": 0.0354,
+      "step": 1013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5743825528022848e-05,
+      "loss": 0.0346,
+      "step": 1014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5735290004184354e-05,
+      "loss": 0.0395,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.57267482491903e-05,
+      "loss": 0.0466,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5718200272320965e-05,
+      "loss": 0.0374,
+      "step": 1017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570964608286336e-05,
+      "loss": 0.0381,
+      "step": 1018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5701085690111268e-05,
+      "loss": 0.043,
+      "step": 1019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5692519103365194e-05,
+      "loss": 0.0396,
+      "step": 1020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5683946331932385e-05,
+      "loss": 0.0388,
+      "step": 1021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56753673851268e-05,
+      "loss": 0.0321,
+      "step": 1022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5666782272269113e-05,
+      "loss": 0.0407,
+      "step": 1023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5658191002686694e-05,
+      "loss": 0.0345,
+      "step": 1024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5649593585713604e-05,
+      "loss": 0.0432,
+      "step": 1025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5640990030690576e-05,
+      "loss": 0.0492,
+      "step": 1026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563238034696502e-05,
+      "loss": 0.0363,
+      "step": 1027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5623764543891e-05,
+      "loss": 0.0336,
+      "step": 1028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.561514263082923e-05,
+      "loss": 0.0343,
+      "step": 1029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560651461714707e-05,
+      "loss": 0.0297,
+      "step": 1030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5597880512218486e-05,
+      "loss": 0.0409,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5589240325424088e-05,
+      "loss": 0.0351,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5580594066151076e-05,
+      "loss": 0.0325,
+      "step": 1033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5571941743793256e-05,
+      "loss": 0.0354,
+      "step": 1034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556328336775102e-05,
+      "loss": 0.0306,
+      "step": 1035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5554618947431326e-05,
+      "loss": 0.0417,
+      "step": 1036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5545948492247723e-05,
+      "loss": 0.0332,
+      "step": 1037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5537272011620294e-05,
+      "loss": 0.0304,
+      "step": 1038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5528589514975675e-05,
+      "loss": 0.0336,
+      "step": 1039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5519901011747046e-05,
+      "loss": 0.0368,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.55112065113741e-05,
+      "loss": 0.0367,
+      "step": 1041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502506023303058e-05,
+      "loss": 0.0474,
+      "step": 1042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5493799556986634e-05,
+      "loss": 0.0364,
+      "step": 1043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5485087121884056e-05,
+      "loss": 0.0342,
+      "step": 1044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547636872746101e-05,
+      "loss": 0.0443,
+      "step": 1045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546764438318968e-05,
+      "loss": 0.0372,
+      "step": 1046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5458914098548702e-05,
+      "loss": 0.043,
+      "step": 1047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5450177883023174e-05,
+      "loss": 0.0327,
+      "step": 1048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5441435746104628e-05,
+      "loss": 0.0393,
+      "step": 1049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5432687697291037e-05,
+      "loss": 0.0433,
+      "step": 1050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5423933746086793e-05,
+      "loss": 0.0366,
+      "step": 1051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5415173902002703e-05,
+      "loss": 0.0368,
+      "step": 1052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.032,
+      "step": 1053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5397636573270213e-05,
+      "loss": 0.0361,
+      "step": 1054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5388859107675396e-05,
+      "loss": 0.0406,
+      "step": 1055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5380075787307875e-05,
+      "loss": 0.0364,
+      "step": 1056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5371286621710374e-05,
+      "loss": 0.0348,
+      "step": 1057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536249162043194e-05,
+      "loss": 0.0331,
+      "step": 1058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5353690793028e-05,
+      "loss": 0.0438,
+      "step": 1059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534488414906028e-05,
+      "loss": 0.0309,
+      "step": 1060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.533607169809683e-05,
+      "loss": 0.0454,
+      "step": 1061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.532725344971202e-05,
+      "loss": 0.0389,
+      "step": 1062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5318429413486507e-05,
+      "loss": 0.0376,
+      "step": 1063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530959959900725e-05,
+      "loss": 0.0411,
+      "step": 1064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5300764015867467e-05,
+      "loss": 0.0391,
+      "step": 1065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5291922673666664e-05,
+      "loss": 0.0342,
+      "step": 1066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5283075582010586e-05,
+      "loss": 0.0318,
+      "step": 1067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527422275051124e-05,
+      "loss": 0.0377,
+      "step": 1068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5265364188786852e-05,
+      "loss": 0.0448,
+      "step": 1069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5256499906461893e-05,
+      "loss": 0.0328,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5247629913167034e-05,
+      "loss": 0.04,
+      "step": 1071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5238754218539155e-05,
+      "loss": 0.0479,
+      "step": 1072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5229872832221336e-05,
+      "loss": 0.0382,
+      "step": 1073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.522098576386283e-05,
+      "loss": 0.0386,
+      "step": 1074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5212093023119075e-05,
+      "loss": 0.0457,
+      "step": 1075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5203194619651664e-05,
+      "loss": 0.0353,
+      "step": 1076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5194290563128346e-05,
+      "loss": 0.0373,
+      "step": 1077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5185380863223007e-05,
+      "loss": 0.0368,
+      "step": 1078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5176465529615672e-05,
+      "loss": 0.0364,
+      "step": 1079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5167544571992478e-05,
+      "loss": 0.038,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5158618000045679e-05,
+      "loss": 0.0376,
+      "step": 1081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5149685823473633e-05,
+      "loss": 0.0296,
+      "step": 1082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5140748051980767e-05,
+      "loss": 0.0389,
+      "step": 1083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5131804695277612e-05,
+      "loss": 0.0352,
+      "step": 1084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5122855763080748e-05,
+      "loss": 0.0406,
+      "step": 1085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5113901265112829e-05,
+      "loss": 0.0338,
+      "step": 1086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5104941211102536e-05,
+      "loss": 0.0458,
+      "step": 1087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5095975610784603e-05,
+      "loss": 0.0373,
+      "step": 1088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5087004473899783e-05,
+      "loss": 0.0412,
+      "step": 1089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5078027810194847e-05,
+      "loss": 0.041,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5069045629422569e-05,
+      "loss": 0.0387,
+      "step": 1091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5060057941341717e-05,
+      "loss": 0.0317,
+      "step": 1092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5051064755717041e-05,
+      "loss": 0.0394,
+      "step": 1093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5042066082319266e-05,
+      "loss": 0.0462,
+      "step": 1094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5033061930925081e-05,
+      "loss": 0.0391,
+      "step": 1095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5024052311317122e-05,
+      "loss": 0.039,
+      "step": 1096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5015037233283969e-05,
+      "loss": 0.0331,
+      "step": 1097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5006016706620132e-05,
+      "loss": 0.0367,
+      "step": 1098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.499699074112604e-05,
+      "loss": 0.0405,
+      "step": 1099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4987959346608029e-05,
+      "loss": 0.0335,
+      "step": 1100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.497892253287834e-05,
+      "loss": 0.0331,
+      "step": 1101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4969880309755098e-05,
+      "loss": 0.0398,
+      "step": 1102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4960832687062303e-05,
+      "loss": 0.0325,
+      "step": 1103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.495177967462982e-05,
+      "loss": 0.0372,
+      "step": 1104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4942721282293378e-05,
+      "loss": 0.0442,
+      "step": 1105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4933657519894542e-05,
+      "loss": 0.0349,
+      "step": 1106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4924588397280717e-05,
+      "loss": 0.0368,
+      "step": 1107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4915513924305132e-05,
+      "loss": 0.0409,
+      "step": 1108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4906434110826822e-05,
+      "loss": 0.031,
+      "step": 1109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4897348966710632e-05,
+      "loss": 0.031,
+      "step": 1110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4888258501827197e-05,
+      "loss": 0.0359,
+      "step": 1111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4879162726052926e-05,
+      "loss": 0.036,
+      "step": 1112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4870061649270014e-05,
+      "loss": 0.0387,
+      "step": 1113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4860955281366392e-05,
+      "loss": 0.0273,
+      "step": 1114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4851843632235761e-05,
+      "loss": 0.0468,
+      "step": 1115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4842726711777546e-05,
+      "loss": 0.0419,
+      "step": 1116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.483360452989691e-05,
+      "loss": 0.0401,
+      "step": 1117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.482447709650472e-05,
+      "loss": 0.0354,
+      "step": 1118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4815344421517561e-05,
+      "loss": 0.0365,
+      "step": 1119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4806206514857703e-05,
+      "loss": 0.0299,
+      "step": 1120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4797063386453107e-05,
+      "loss": 0.0385,
+      "step": 1121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4787915046237401e-05,
+      "loss": 0.0338,
+      "step": 1122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4778761504149882e-05,
+      "loss": 0.0388,
+      "step": 1123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4769602770135492e-05,
+      "loss": 0.0409,
+      "step": 1124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4760438854144816e-05,
+      "loss": 0.0302,
+      "step": 1125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4751269766134075e-05,
+      "loss": 0.0325,
+      "step": 1126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.47420955160651e-05,
+      "loss": 0.0372,
+      "step": 1127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4732916113905336e-05,
+      "loss": 0.0411,
+      "step": 1128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4723731569627827e-05,
+      "loss": 0.0333,
+      "step": 1129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4714541893211195e-05,
+      "loss": 0.0325,
+      "step": 1130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4705347094639643e-05,
+      "loss": 0.0376,
+      "step": 1131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4696147183902949e-05,
+      "loss": 0.0364,
+      "step": 1132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4686942170996426e-05,
+      "loss": 0.0419,
+      "step": 1133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4677732065920945e-05,
+      "loss": 0.042,
+      "step": 1134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4668516878682902e-05,
+      "loss": 0.0462,
+      "step": 1135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.465929661929422e-05,
+      "loss": 0.0358,
+      "step": 1136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4650071297772326e-05,
+      "loss": 0.0271,
+      "step": 1137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4640840924140155e-05,
+      "loss": 0.0283,
+      "step": 1138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4631605508426124e-05,
+      "loss": 0.0451,
+      "step": 1139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4622365060664132e-05,
+      "loss": 0.0357,
+      "step": 1140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4613119590893536e-05,
+      "loss": 0.036,
+      "step": 1141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4603869109159172e-05,
+      "loss": 0.0455,
+      "step": 1142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4594613625511288e-05,
+      "loss": 0.0403,
+      "step": 1143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.45853531500056e-05,
+      "loss": 0.0455,
+      "step": 1144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4576087692703223e-05,
+      "loss": 0.0378,
+      "step": 1145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4566817263670698e-05,
+      "loss": 0.037,
+      "step": 1146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.455754187297996e-05,
+      "loss": 0.0363,
+      "step": 1147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4548261530708336e-05,
+      "loss": 0.0387,
+      "step": 1148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4538976246938541e-05,
+      "loss": 0.0384,
+      "step": 1149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4529686031758642e-05,
+      "loss": 0.0354,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4520390895262089e-05,
+      "loss": 0.0309,
+      "step": 1151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4511090847547643e-05,
+      "loss": 0.0321,
+      "step": 1152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4501785898719443e-05,
+      "loss": 0.0292,
+      "step": 1153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4492476058886915e-05,
+      "loss": 0.0329,
+      "step": 1154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4483161338164827e-05,
+      "loss": 0.037,
+      "step": 1155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4473841746673225e-05,
+      "loss": 0.0345,
+      "step": 1156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4464517294537472e-05,
+      "loss": 0.0374,
+      "step": 1157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4455187991888195e-05,
+      "loss": 0.0419,
+      "step": 1158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4445853848861295e-05,
+      "loss": 0.0277,
+      "step": 1159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4436514875597935e-05,
+      "loss": 0.0382,
+      "step": 1160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4427171082244523e-05,
+      "loss": 0.0467,
+      "step": 1161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4417822478952701e-05,
+      "loss": 0.0373,
+      "step": 1162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4408469075879344e-05,
+      "loss": 0.0325,
+      "step": 1163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4399110883186538e-05,
+      "loss": 0.0431,
+      "step": 1164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4389747911041568e-05,
+      "loss": 0.0383,
+      "step": 1165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.438038016961692e-05,
+      "loss": 0.036,
+      "step": 1166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4371007669090257e-05,
+      "loss": 0.0361,
+      "step": 1167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4361630419644417e-05,
+      "loss": 0.0348,
+      "step": 1168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4352248431467386e-05,
+      "loss": 0.0412,
+      "step": 1169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4342861714752315e-05,
+      "loss": 0.0401,
+      "step": 1170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.433347027969748e-05,
+      "loss": 0.0378,
+      "step": 1171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4324074136506283e-05,
+      "loss": 0.0424,
+      "step": 1172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4314673295387251e-05,
+      "loss": 0.0337,
+      "step": 1173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4305267766554007e-05,
+      "loss": 0.0412,
+      "step": 1174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4295857560225268e-05,
+      "loss": 0.0316,
+      "step": 1175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4286442686624842e-05,
+      "loss": 0.0401,
+      "step": 1176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4277023155981587e-05,
+      "loss": 0.0276,
+      "step": 1177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4267598978529447e-05,
+      "loss": 0.0447,
+      "step": 1178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.425817016450739e-05,
+      "loss": 0.0352,
+      "step": 1179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4248736724159442e-05,
+      "loss": 0.0364,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4239298667734638e-05,
+      "loss": 0.0417,
+      "step": 1181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4229856005487044e-05,
+      "loss": 0.0362,
+      "step": 1182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4220408747675714e-05,
+      "loss": 0.0377,
+      "step": 1183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.421095690456471e-05,
+      "loss": 0.0358,
+      "step": 1184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4201500486423067e-05,
+      "loss": 0.0364,
+      "step": 1185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4192039503524792e-05,
+      "loss": 0.0321,
+      "step": 1186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4182573966148851e-05,
+      "loss": 0.0451,
+      "step": 1187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4173103884579163e-05,
+      "loss": 0.0389,
+      "step": 1188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4163629269104573e-05,
+      "loss": 0.0377,
+      "step": 1189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.0348,
+      "step": 1190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.414466647762073e-05,
+      "loss": 0.0357,
+      "step": 1191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4135178322213765e-05,
+      "loss": 0.0314,
+      "step": 1192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.412568567410646e-05,
+      "loss": 0.037,
+      "step": 1193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4116188543612182e-05,
+      "loss": 0.0381,
+      "step": 1194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4106686941049171e-05,
+      "loss": 0.0341,
+      "step": 1195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4097180876740525e-05,
+      "loss": 0.0458,
+      "step": 1196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4087670361014194e-05,
+      "loss": 0.0343,
+      "step": 1197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4078155404202957e-05,
+      "loss": 0.0398,
+      "step": 1198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4068636016644424e-05,
+      "loss": 0.0302,
+      "step": 1199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4059112208681012e-05,
+      "loss": 0.028,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4049583990659947e-05,
+      "loss": 0.0287,
+      "step": 1201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4040051372933244e-05,
+      "loss": 0.0305,
+      "step": 1202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4030514365857696e-05,
+      "loss": 0.0289,
+      "step": 1203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.402097297979487e-05,
+      "loss": 0.0278,
+      "step": 1204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4011427225111091e-05,
+      "loss": 0.038,
+      "step": 1205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4001877112177418e-05,
+      "loss": 0.0401,
+      "step": 1206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3992322651369663e-05,
+      "loss": 0.0307,
+      "step": 1207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3982763853068345e-05,
+      "loss": 0.0381,
+      "step": 1208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3973200727658709e-05,
+      "loss": 0.0343,
+      "step": 1209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3963633285530688e-05,
+      "loss": 0.0314,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3954061537078927e-05,
+      "loss": 0.0298,
+      "step": 1211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3944485492702717e-05,
+      "loss": 0.0368,
+      "step": 1212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.393490516280605e-05,
+      "loss": 0.0274,
+      "step": 1213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3925320557797549e-05,
+      "loss": 0.0312,
+      "step": 1214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3915731688090492e-05,
+      "loss": 0.0365,
+      "step": 1215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3906138564102794e-05,
+      "loss": 0.0321,
+      "step": 1216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3896541196256985e-05,
+      "loss": 0.0436,
+      "step": 1217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3886939594980209e-05,
+      "loss": 0.0402,
+      "step": 1218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3877333770704208e-05,
+      "loss": 0.0336,
+      "step": 1219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3867723733865314e-05,
+      "loss": 0.0288,
+      "step": 1220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3858109494904437e-05,
+      "loss": 0.0338,
+      "step": 1221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3848491064267046e-05,
+      "loss": 0.0398,
+      "step": 1222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3838868452403172e-05,
+      "loss": 0.036,
+      "step": 1223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3829241669767381e-05,
+      "loss": 0.0359,
+      "step": 1224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.381961072681878e-05,
+      "loss": 0.0308,
+      "step": 1225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3809975634020985e-05,
+      "loss": 0.0332,
+      "step": 1226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.380033640184213e-05,
+      "loss": 0.0342,
+      "step": 1227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3790693040754839e-05,
+      "loss": 0.0383,
+      "step": 1228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3781045561236228e-05,
+      "loss": 0.0351,
+      "step": 1229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3771393973767886e-05,
+      "loss": 0.031,
+      "step": 1230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.376173828883586e-05,
+      "loss": 0.041,
+      "step": 1231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3752078516930653e-05,
+      "loss": 0.0346,
+      "step": 1232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.374241466854721e-05,
+      "loss": 0.0288,
+      "step": 1233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3732746754184905e-05,
+      "loss": 0.0355,
+      "step": 1234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3723074784347521e-05,
+      "loss": 0.0392,
+      "step": 1235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.371339876954326e-05,
+      "loss": 0.0313,
+      "step": 1236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3703718720284707e-05,
+      "loss": 0.0404,
+      "step": 1237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.369403464708884e-05,
+      "loss": 0.038,
+      "step": 1238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3684346560477e-05,
+      "loss": 0.0335,
+      "step": 1239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3674654470974897e-05,
+      "loss": 0.0361,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3664958389112581e-05,
+      "loss": 0.0327,
+      "step": 1241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3655258325424447e-05,
+      "loss": 0.029,
+      "step": 1242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3645554290449215e-05,
+      "loss": 0.0342,
+      "step": 1243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3635846294729915e-05,
+      "loss": 0.0405,
+      "step": 1244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3626134348813885e-05,
+      "loss": 0.0308,
+      "step": 1245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3616418463252747e-05,
+      "loss": 0.0226,
+      "step": 1246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.360669864860242e-05,
+      "loss": 0.0413,
+      "step": 1247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3596974915423071e-05,
+      "loss": 0.0333,
+      "step": 1248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.358724727427914e-05,
+      "loss": 0.0398,
+      "step": 1249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3577515735739302e-05,
+      "loss": 0.0371,
+      "step": 1250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3567780310376476e-05,
+      "loss": 0.0305,
+      "step": 1251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3558041008767799e-05,
+      "loss": 0.0384,
+      "step": 1252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.354829784149462e-05,
+      "loss": 0.0366,
+      "step": 1253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3538550819142484e-05,
+      "loss": 0.0321,
+      "step": 1254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3528799952301128e-05,
+      "loss": 0.0357,
+      "step": 1255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.351904525156447e-05,
+      "loss": 0.0367,
+      "step": 1256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3509286727530588e-05,
+      "loss": 0.0338,
+      "step": 1257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3499524390801716e-05,
+      "loss": 0.0337,
+      "step": 1258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3489758251984222e-05,
+      "loss": 0.031,
+      "step": 1259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3479988321688619e-05,
+      "loss": 0.0317,
+      "step": 1260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3470214610529536e-05,
+      "loss": 0.0278,
+      "step": 1261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3460437129125694e-05,
+      "loss": 0.0387,
+      "step": 1262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3450655888099935e-05,
+      "loss": 0.0333,
+      "step": 1263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3440870898079166e-05,
+      "loss": 0.0281,
+      "step": 1264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3431082169694376e-05,
+      "loss": 0.0291,
+      "step": 1265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3421289713580616e-05,
+      "loss": 0.0366,
+      "step": 1266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.341149354037698e-05,
+      "loss": 0.0371,
+      "step": 1267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.340169366072661e-05,
+      "loss": 0.0336,
+      "step": 1268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3391890085276669e-05,
+      "loss": 0.0316,
+      "step": 1269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3382082824678336e-05,
+      "loss": 0.0308,
+      "step": 1270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.337227188958679e-05,
+      "loss": 0.0272,
+      "step": 1271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3362457290661215e-05,
+      "loss": 0.034,
+      "step": 1272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.335263903856476e-05,
+      "loss": 0.0384,
+      "step": 1273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3342817143964557e-05,
+      "loss": 0.0364,
+      "step": 1274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.333299161753168e-05,
+      "loss": 0.0337,
+      "step": 1275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3323162469941164e-05,
+      "loss": 0.0441,
+      "step": 1276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3313329711871969e-05,
+      "loss": 0.0306,
+      "step": 1277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3303493354006985e-05,
+      "loss": 0.0367,
+      "step": 1278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3293653407032997e-05,
+      "loss": 0.0265,
+      "step": 1279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.328380988164071e-05,
+      "loss": 0.0372,
+      "step": 1280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3273962788524705e-05,
+      "loss": 0.0342,
+      "step": 1281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3264112138383445e-05,
+      "loss": 0.0337,
+      "step": 1282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3254257941919249e-05,
+      "loss": 0.0361,
+      "step": 1283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3244400209838297e-05,
+      "loss": 0.0416,
+      "step": 1284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3234538952850606e-05,
+      "loss": 0.039,
+      "step": 1285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3224674181670025e-05,
+      "loss": 0.0545,
+      "step": 1286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3214805907014223e-05,
+      "loss": 0.0397,
+      "step": 1287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3204934139604668e-05,
+      "loss": 0.0339,
+      "step": 1288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3195058890166628e-05,
+      "loss": 0.0334,
+      "step": 1289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3185180169429155e-05,
+      "loss": 0.0402,
+      "step": 1290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3175297988125071e-05,
+      "loss": 0.0423,
+      "step": 1291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3165412356990954e-05,
+      "loss": 0.041,
+      "step": 1292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.315552328676714e-05,
+      "loss": 0.0306,
+      "step": 1293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.314563078819769e-05,
+      "loss": 0.0332,
+      "step": 1294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3135734872030397e-05,
+      "loss": 0.0364,
+      "step": 1295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3125835549016763e-05,
+      "loss": 0.0375,
+      "step": 1296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3115932829911997e-05,
+      "loss": 0.0385,
+      "step": 1297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3106026725474987e-05,
+      "loss": 0.0309,
+      "step": 1298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3096117246468317e-05,
+      "loss": 0.0354,
+      "step": 1299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3086204403658216e-05,
+      "loss": 0.0297,
+      "step": 1300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3076288207814585e-05,
+      "loss": 0.0348,
+      "step": 1301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066368669710953e-05,
+      "loss": 0.0363,
+      "step": 1302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056445800124497e-05,
+      "loss": 0.0332,
+      "step": 1303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3046519609836002e-05,
+      "loss": 0.0364,
+      "step": 1304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303659010962986e-05,
+      "loss": 0.0396,
+      "step": 1305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3026657310294067e-05,
+      "loss": 0.0302,
+      "step": 1306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3016721222620197e-05,
+      "loss": 0.0333,
+      "step": 1307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3006781857403394e-05,
+      "loss": 0.0359,
+      "step": 1308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2996839225442377e-05,
+      "loss": 0.0407,
+      "step": 1309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2986893337539397e-05,
+      "loss": 0.0226,
+      "step": 1310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976944204500255e-05,
+      "loss": 0.0384,
+      "step": 1311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2966991837134271e-05,
+      "loss": 0.0523,
+      "step": 1312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295703624625428e-05,
+      "loss": 0.0303,
+      "step": 1313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.294707744267662e-05,
+      "loss": 0.0322,
+      "step": 1314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2937115437221119e-05,
+      "loss": 0.0247,
+      "step": 1315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2927150240711089e-05,
+      "loss": 0.03,
+      "step": 1316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2917181863973298e-05,
+      "loss": 0.0359,
+      "step": 1317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.290721031783798e-05,
+      "loss": 0.0432,
+      "step": 1318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2897235613138804e-05,
+      "loss": 0.0299,
+      "step": 1319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2887257760712875e-05,
+      "loss": 0.0391,
+      "step": 1320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.287727677140072e-05,
+      "loss": 0.0328,
+      "step": 1321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2867292656046268e-05,
+      "loss": 0.0394,
+      "step": 1322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.285730542549685e-05,
+      "loss": 0.0332,
+      "step": 1323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.284731509060318e-05,
+      "loss": 0.0447,
+      "step": 1324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2837321662219342e-05,
+      "loss": 0.0402,
+      "step": 1325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2827325151202783e-05,
+      "loss": 0.0305,
+      "step": 1326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.028,
+      "step": 1327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2807322924718024e-05,
+      "loss": 0.0481,
+      "step": 1328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2797317230981416e-05,
+      "loss": 0.0349,
+      "step": 1329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2787308498075253e-05,
+      "loss": 0.0287,
+      "step": 1330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.27772967368736e-05,
+      "loss": 0.0309,
+      "step": 1331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.276728195825383e-05,
+      "loss": 0.029,
+      "step": 1332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2757264173096582e-05,
+      "loss": 0.0343,
+      "step": 1333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2747243392285769e-05,
+      "loss": 0.0261,
+      "step": 1334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2737219626708549e-05,
+      "loss": 0.0357,
+      "step": 1335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2727192887255332e-05,
+      "loss": 0.0303,
+      "step": 1336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2717163184819761e-05,
+      "loss": 0.029,
+      "step": 1337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2707130530298688e-05,
+      "loss": 0.0416,
+      "step": 1338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2697094934592177e-05,
+      "loss": 0.0447,
+      "step": 1339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2687056408603493e-05,
+      "loss": 0.0333,
+      "step": 1340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2677014963239075e-05,
+      "loss": 0.0278,
+      "step": 1341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2666970609408535e-05,
+      "loss": 0.0461,
+      "step": 1342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2656923358024659e-05,
+      "loss": 0.042,
+      "step": 1343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.264687322000336e-05,
+      "loss": 0.0408,
+      "step": 1344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2636820206263704e-05,
+      "loss": 0.0305,
+      "step": 1345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2626764327727871e-05,
+      "loss": 0.0354,
+      "step": 1346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.261670559532116e-05,
+      "loss": 0.0313,
+      "step": 1347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2606644019971967e-05,
+      "loss": 0.0367,
+      "step": 1348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.259657961261178e-05,
+      "loss": 0.0318,
+      "step": 1349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2586512384175156e-05,
+      "loss": 0.0363,
+      "step": 1350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2576442345599729e-05,
+      "loss": 0.0394,
+      "step": 1351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2566369507826175e-05,
+      "loss": 0.037,
+      "step": 1352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2556293881798218e-05,
+      "loss": 0.032,
+      "step": 1353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2546215478462611e-05,
+      "loss": 0.0328,
+      "step": 1354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2536134308769118e-05,
+      "loss": 0.0342,
+      "step": 1355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2526050383670516e-05,
+      "loss": 0.0321,
+      "step": 1356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2515963714122577e-05,
+      "loss": 0.0302,
+      "step": 1357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2505874311084041e-05,
+      "loss": 0.0357,
+      "step": 1358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2495782185516638e-05,
+      "loss": 0.0323,
+      "step": 1359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2485687348385038e-05,
+      "loss": 0.0309,
+      "step": 1360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2475589810656868e-05,
+      "loss": 0.0395,
+      "step": 1361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.246548958330268e-05,
+      "loss": 0.0271,
+      "step": 1362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2455386677295962e-05,
+      "loss": 0.0272,
+      "step": 1363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2445281103613096e-05,
+      "loss": 0.043,
+      "step": 1364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2435172873233371e-05,
+      "loss": 0.03,
+      "step": 1365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2425061997138972e-05,
+      "loss": 0.0309,
+      "step": 1366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2414948486314932e-05,
+      "loss": 0.0442,
+      "step": 1367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2404832351749179e-05,
+      "loss": 0.0322,
+      "step": 1368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.239471360443246e-05,
+      "loss": 0.0395,
+      "step": 1369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2384592255358385e-05,
+      "loss": 0.03,
+      "step": 1370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2374468315523375e-05,
+      "loss": 0.032,
+      "step": 1371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2364341795926684e-05,
+      "loss": 0.0353,
+      "step": 1372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2354212707570344e-05,
+      "loss": 0.0366,
+      "step": 1373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2344081061459195e-05,
+      "loss": 0.0372,
+      "step": 1374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2333946868600854e-05,
+      "loss": 0.0356,
+      "step": 1375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2323810140005701e-05,
+      "loss": 0.0356,
+      "step": 1376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2313670886686874e-05,
+      "loss": 0.0336,
+      "step": 1377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.230352911966025e-05,
+      "loss": 0.0387,
+      "step": 1378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2293384849944445e-05,
+      "loss": 0.0297,
+      "step": 1379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2283238088560783e-05,
+      "loss": 0.0351,
+      "step": 1380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2273088846533303e-05,
+      "loss": 0.0327,
+      "step": 1381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.226293713488874e-05,
+      "loss": 0.0328,
+      "step": 1382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2252782964656502e-05,
+      "loss": 0.0302,
+      "step": 1383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2242626346868682e-05,
+      "loss": 0.0358,
+      "step": 1384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2232467292560025e-05,
+      "loss": 0.0392,
+      "step": 1385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2222305812767919e-05,
+      "loss": 0.0319,
+      "step": 1386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2212141918532394e-05,
+      "loss": 0.0325,
+      "step": 1387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2201975620896102e-05,
+      "loss": 0.0279,
+      "step": 1388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2191806930904304e-05,
+      "loss": 0.0372,
+      "step": 1389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2181635859604864e-05,
+      "loss": 0.0302,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2171462418048228e-05,
+      "loss": 0.0286,
+      "step": 1391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.216128661728742e-05,
+      "loss": 0.0311,
+      "step": 1392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2151108468378028e-05,
+      "loss": 0.0235,
+      "step": 1393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2140927982378189e-05,
+      "loss": 0.0262,
+      "step": 1394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2130745170348584e-05,
+      "loss": 0.034,
+      "step": 1395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.212056004335241e-05,
+      "loss": 0.0285,
+      "step": 1396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2110372612455394e-05,
+      "loss": 0.0314,
+      "step": 1397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2100182888725751e-05,
+      "loss": 0.0382,
+      "step": 1398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2089990883234206e-05,
+      "loss": 0.039,
+      "step": 1399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2079796607053939e-05,
+      "loss": 0.0325,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2069600071260621e-05,
+      "loss": 0.0338,
+      "step": 1401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2059401286932354e-05,
+      "loss": 0.0326,
+      "step": 1402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.204920026514971e-05,
+      "loss": 0.027,
+      "step": 1403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2038997016995668e-05,
+      "loss": 0.0345,
+      "step": 1404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.202879155355564e-05,
+      "loss": 0.0307,
+      "step": 1405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2018583885917434e-05,
+      "loss": 0.0454,
+      "step": 1406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2008374025171265e-05,
+      "loss": 0.0279,
+      "step": 1407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1998161982409724e-05,
+      "loss": 0.0384,
+      "step": 1408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1987947768727773e-05,
+      "loss": 0.0355,
+      "step": 1409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1977731395222731e-05,
+      "loss": 0.0332,
+      "step": 1410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1967512872994266e-05,
+      "loss": 0.0343,
+      "step": 1411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1957292213144386e-05,
+      "loss": 0.0299,
+      "step": 1412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1947069426777404e-05,
+      "loss": 0.0293,
+      "step": 1413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1936844524999966e-05,
+      "loss": 0.0357,
+      "step": 1414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1926617518921e-05,
+      "loss": 0.0375,
+      "step": 1415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1916388419651725e-05,
+      "loss": 0.0355,
+      "step": 1416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1906157238305635e-05,
+      "loss": 0.0398,
+      "step": 1417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.189592398599849e-05,
+      "loss": 0.036,
+      "step": 1418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1885688673848289e-05,
+      "loss": 0.0348,
+      "step": 1419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1875451312975278e-05,
+      "loss": 0.0441,
+      "step": 1420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1865211914501926e-05,
+      "loss": 0.0317,
+      "step": 1421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1854970489552924e-05,
+      "loss": 0.0429,
+      "step": 1422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1844727049255143e-05,
+      "loss": 0.0273,
+      "step": 1423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1834481604737671e-05,
+      "loss": 0.0306,
+      "step": 1424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1824234167131748e-05,
+      "loss": 0.0299,
+      "step": 1425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1813984747570802e-05,
+      "loss": 0.0335,
+      "step": 1426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1803733357190396e-05,
+      "loss": 0.0363,
+      "step": 1427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1793480007128248e-05,
+      "loss": 0.0348,
+      "step": 1428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1783224708524192e-05,
+      "loss": 0.0373,
+      "step": 1429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1772967472520196e-05,
+      "loss": 0.0346,
+      "step": 1430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1762708310260312e-05,
+      "loss": 0.0323,
+      "step": 1431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1752447232890702e-05,
+      "loss": 0.0274,
+      "step": 1432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1742184251559602e-05,
+      "loss": 0.0251,
+      "step": 1433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1731919377417318e-05,
+      "loss": 0.0362,
+      "step": 1434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1721652621616208e-05,
+      "loss": 0.0342,
+      "step": 1435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.171138399531068e-05,
+      "loss": 0.0357,
+      "step": 1436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1701113509657174e-05,
+      "loss": 0.0308,
+      "step": 1437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1690841175814145e-05,
+      "loss": 0.0372,
+      "step": 1438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1680567004942062e-05,
+      "loss": 0.0306,
+      "step": 1439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1670291008203383e-05,
+      "loss": 0.0297,
+      "step": 1440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1660013196762556e-05,
+      "loss": 0.0327,
+      "step": 1441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1649733581786001e-05,
+      "loss": 0.0353,
+      "step": 1442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.163945217444209e-05,
+      "loss": 0.031,
+      "step": 1443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1629168985901149e-05,
+      "loss": 0.0344,
+      "step": 1444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1618884027335438e-05,
+      "loss": 0.0334,
+      "step": 1445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1608597309919141e-05,
+      "loss": 0.0279,
+      "step": 1446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1598308844828348e-05,
+      "loss": 0.0374,
+      "step": 1447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1588018643241053e-05,
+      "loss": 0.0311,
+      "step": 1448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1577726716337134e-05,
+      "loss": 0.0304,
+      "step": 1449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1567433075298341e-05,
+      "loss": 0.0342,
+      "step": 1450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1557137731308302e-05,
+      "loss": 0.0457,
+      "step": 1451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1546840695552465e-05,
+      "loss": 0.0292,
+      "step": 1452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.153654197921815e-05,
+      "loss": 0.0312,
+      "step": 1453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1526241593494474e-05,
+      "loss": 0.0324,
+      "step": 1454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1515939549572389e-05,
+      "loss": 0.0339,
+      "step": 1455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1505635858644635e-05,
+      "loss": 0.032,
+      "step": 1456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1495330531905749e-05,
+      "loss": 0.0291,
+      "step": 1457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1485023580552039e-05,
+      "loss": 0.0345,
+      "step": 1458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1474715015781584e-05,
+      "loss": 0.0266,
+      "step": 1459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1464404848794218e-05,
+      "loss": 0.0325,
+      "step": 1460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1454093090791503e-05,
+      "loss": 0.0366,
+      "step": 1461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1443779752976745e-05,
+      "loss": 0.0344,
+      "step": 1462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1433464846554953e-05,
+      "loss": 0.0337,
+      "step": 1463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.0352,
+      "step": 1464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1412830372718851e-05,
+      "loss": 0.0384,
+      "step": 1465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1402510827723043e-05,
+      "loss": 0.0325,
+      "step": 1466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1392189758957183e-05,
+      "loss": 0.0263,
+      "step": 1467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1381867177634693e-05,
+      "loss": 0.0383,
+      "step": 1468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1371543094970624e-05,
+      "loss": 0.0354,
+      "step": 1469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1361217522181667e-05,
+      "loss": 0.0284,
+      "step": 1470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.135089047048613e-05,
+      "loss": 0.0376,
+      "step": 1471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.134056195110393e-05,
+      "loss": 0.0349,
+      "step": 1472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1330231975256576e-05,
+      "loss": 0.0282,
+      "step": 1473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1319900554167158e-05,
+      "loss": 0.0327,
+      "step": 1474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.130956769906034e-05,
+      "loss": 0.0331,
+      "step": 1475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1299233421162339e-05,
+      "loss": 0.0322,
+      "step": 1476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1288897731700922e-05,
+      "loss": 0.032,
+      "step": 1477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1278560641905392e-05,
+      "loss": 0.0334,
+      "step": 1478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1268222163006566e-05,
+      "loss": 0.03,
+      "step": 1479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1257882306236776e-05,
+      "loss": 0.0306,
+      "step": 1480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1247541082829847e-05,
+      "loss": 0.0232,
+      "step": 1481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1237198504021091e-05,
+      "loss": 0.0295,
+      "step": 1482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1226854581047295e-05,
+      "loss": 0.0425,
+      "step": 1483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.12165093251467e-05,
+      "loss": 0.0315,
+      "step": 1484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1206162747559001e-05,
+      "loss": 0.0328,
+      "step": 1485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1195814859525332e-05,
+      "loss": 0.036,
+      "step": 1486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1185465672288237e-05,
+      "loss": 0.0338,
+      "step": 1487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1175115197091682e-05,
+      "loss": 0.0345,
+      "step": 1488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1164763445181034e-05,
+      "loss": 0.0346,
+      "step": 1489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1154410427803042e-05,
+      "loss": 0.0353,
+      "step": 1490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1144056156205834e-05,
+      "loss": 0.031,
+      "step": 1491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1133700641638892e-05,
+      "loss": 0.0356,
+      "step": 1492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1123343895353063e-05,
+      "loss": 0.0279,
+      "step": 1493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1112985928600513e-05,
+      "loss": 0.0362,
+      "step": 1494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1102626752634758e-05,
+      "loss": 0.0277,
+      "step": 1495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1092266378710602e-05,
+      "loss": 0.0442,
+      "step": 1496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1081904818084169e-05,
+      "loss": 0.0365,
+      "step": 1497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1071542082012865e-05,
+      "loss": 0.029,
+      "step": 1498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.106117818175538e-05,
+      "loss": 0.0384,
+      "step": 1499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1050813128571654e-05,
+      "loss": 0.0303,
+      "step": 1500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1040446933722894e-05,
+      "loss": 0.0304,
+      "step": 1501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1030079608471544e-05,
+      "loss": 0.0329,
+      "step": 1502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1019711164081272e-05,
+      "loss": 0.0316,
+      "step": 1503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1009341611816963e-05,
+      "loss": 0.0345,
+      "step": 1504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.099897096294471e-05,
+      "loss": 0.0298,
+      "step": 1505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0988599228731794e-05,
+      "loss": 0.0343,
+      "step": 1506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0978226420446671e-05,
+      "loss": 0.0302,
+      "step": 1507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0967852549358975e-05,
+      "loss": 0.0321,
+      "step": 1508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0957477626739483e-05,
+      "loss": 0.0284,
+      "step": 1509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0947101663860123e-05,
+      "loss": 0.0302,
+      "step": 1510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0936724671993948e-05,
+      "loss": 0.0378,
+      "step": 1511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.092634666241513e-05,
+      "loss": 0.0373,
+      "step": 1512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.091596764639895e-05,
+      "loss": 0.0286,
+      "step": 1513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0905587635221782e-05,
+      "loss": 0.0284,
+      "step": 1514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.089520664016107e-05,
+      "loss": 0.0283,
+      "step": 1515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0884824672495346e-05,
+      "loss": 0.0352,
+      "step": 1516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0874441743504182e-05,
+      "loss": 0.0304,
+      "step": 1517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.086405786446821e-05,
+      "loss": 0.0313,
+      "step": 1518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0853673046669071e-05,
+      "loss": 0.0337,
+      "step": 1519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0843287301389455e-05,
+      "loss": 0.0392,
+      "step": 1520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0832900639913035e-05,
+      "loss": 0.0334,
+      "step": 1521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0822513073524498e-05,
+      "loss": 0.0305,
+      "step": 1522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0812124613509496e-05,
+      "loss": 0.0379,
+      "step": 1523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.080173527115467e-05,
+      "loss": 0.0338,
+      "step": 1524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0791345057747604e-05,
+      "loss": 0.0399,
+      "step": 1525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0780953984576841e-05,
+      "loss": 0.0276,
+      "step": 1526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0770562062931849e-05,
+      "loss": 0.0363,
+      "step": 1527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0760169304103023e-05,
+      "loss": 0.0325,
+      "step": 1528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0749775719381665e-05,
+      "loss": 0.0276,
+      "step": 1529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0739381320059978e-05,
+      "loss": 0.0288,
+      "step": 1530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0728986117431044e-05,
+      "loss": 0.0309,
+      "step": 1531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0718590122788822e-05,
+      "loss": 0.0331,
+      "step": 1532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0708193347428128e-05,
+      "loss": 0.0282,
+      "step": 1533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0697795802644635e-05,
+      "loss": 0.0321,
+      "step": 1534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0687397499734842e-05,
+      "loss": 0.0315,
+      "step": 1535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0676998449996075e-05,
+      "loss": 0.0313,
+      "step": 1536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0666598664726471e-05,
+      "loss": 0.032,
+      "step": 1537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0656198155224968e-05,
+      "loss": 0.0291,
+      "step": 1538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.064579693279129e-05,
+      "loss": 0.031,
+      "step": 1539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0635395008725936e-05,
+      "loss": 0.0257,
+      "step": 1540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0624992394330165e-05,
+      "loss": 0.0318,
+      "step": 1541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.061458910090599e-05,
+      "loss": 0.0392,
+      "step": 1542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0604185139756155e-05,
+      "loss": 0.0289,
+      "step": 1543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0593780522184139e-05,
+      "loss": 0.0329,
+      "step": 1544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0583375259494125e-05,
+      "loss": 0.0271,
+      "step": 1545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0572969362991e-05,
+      "loss": 0.0316,
+      "step": 1546
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0562562843980346e-05,
+      "loss": 0.0327,
+      "step": 1547
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0552155713768409e-05,
+      "loss": 0.0274,
+      "step": 1548
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0541747983662114e-05,
+      "loss": 0.0269,
+      "step": 1549
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0531339664969021e-05,
+      "loss": 0.0345,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0520930768997347e-05,
+      "loss": 0.0337,
+      "step": 1551
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0510521307055914e-05,
+      "loss": 0.0324,
+      "step": 1552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0500111290454189e-05,
+      "loss": 0.0401,
+      "step": 1553
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0489700730502212e-05,
+      "loss": 0.0299,
+      "step": 1554
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.047928963851063e-05,
+      "loss": 0.0269,
+      "step": 1555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.046887802579067e-05,
+      "loss": 0.0225,
+      "step": 1556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0458465903654107e-05,
+      "loss": 0.0232,
+      "step": 1557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0448053283413293e-05,
+      "loss": 0.0195,
+      "step": 1558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0437640176381099e-05,
+      "loss": 0.0386,
+      "step": 1559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0427226593870947e-05,
+      "loss": 0.0208,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0416812547196753e-05,
+      "loss": 0.0277,
+      "step": 1561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0406398047672956e-05,
+      "loss": 0.0246,
+      "step": 1562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0395983106614474e-05,
+      "loss": 0.031,
+      "step": 1563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0385567735336712e-05,
+      "loss": 0.0197,
+      "step": 1564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0375151945155544e-05,
+      "loss": 0.0283,
+      "step": 1565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0364735747387292e-05,
+      "loss": 0.0253,
+      "step": 1566
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0354319153348724e-05,
+      "loss": 0.025,
+      "step": 1567
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.034390217435704e-05,
+      "loss": 0.024,
+      "step": 1568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0333484821729857e-05,
+      "loss": 0.0275,
+      "step": 1569
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0323067106785199e-05,
+      "loss": 0.0248,
+      "step": 1570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.031264904084148e-05,
+      "loss": 0.0336,
+      "step": 1571
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.03022306352175e-05,
+      "loss": 0.0217,
+      "step": 1572
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0291811901232428e-05,
+      "loss": 0.0337,
+      "step": 1573
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0281392850205783e-05,
+      "loss": 0.0262,
+      "step": 1574
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0270973493457437e-05,
+      "loss": 0.0274,
+      "step": 1575
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.026055384230759e-05,
+      "loss": 0.0251,
+      "step": 1576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.025013390807676e-05,
+      "loss": 0.0264,
+      "step": 1577
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0239713702085781e-05,
+      "loss": 0.0254,
+      "step": 1578
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0229293235655768e-05,
+      "loss": 0.0299,
+      "step": 1579
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0218872520108135e-05,
+      "loss": 0.0323,
+      "step": 1580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.020845156676455e-05,
+      "loss": 0.0231,
+      "step": 1581
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0198030386946956e-05,
+      "loss": 0.0288,
+      "step": 1582
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0187608991977531e-05,
+      "loss": 0.0259,
+      "step": 1583
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0177187393178687e-05,
+      "loss": 0.0329,
+      "step": 1584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0166765601873067e-05,
+      "loss": 0.0288,
+      "step": 1585
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0156343629383509e-05,
+      "loss": 0.0312,
+      "step": 1586
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0145921487033065e-05,
+      "loss": 0.0265,
+      "step": 1587
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0135499186144952e-05,
+      "loss": 0.0284,
+      "step": 1588
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0125076738042577e-05,
+      "loss": 0.0281,
+      "step": 1589
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.011465415404949e-05,
+      "loss": 0.0333,
+      "step": 1590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.010423144548941e-05,
+      "loss": 0.0254,
+      "step": 1591
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0093808623686165e-05,
+      "loss": 0.0303,
+      "step": 1592
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0083385699963733e-05,
+      "loss": 0.0263,
+      "step": 1593
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0072962685646182e-05,
+      "loss": 0.0281,
+      "step": 1594
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0062539592057689e-05,
+      "loss": 0.0222,
+      "step": 1595
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0052116430522514e-05,
+      "loss": 0.0237,
+      "step": 1596
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0041693212364992e-05,
+      "loss": 0.0269,
+      "step": 1597
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0031269948909517e-05,
+      "loss": 0.0291,
+      "step": 1598
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0020846651480534e-05,
+      "loss": 0.0217,
+      "step": 1599
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0010423331402528e-05,
+      "loss": 0.0274,
+      "step": 1600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.0286,
+      "step": 1601
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.989576668597476e-06,
+      "loss": 0.0288,
+      "step": 1602
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.979153348519466e-06,
+      "loss": 0.029,
+      "step": 1603
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.968730051090487e-06,
+      "loss": 0.0229,
+      "step": 1604
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.958306787635011e-06,
+      "loss": 0.0274,
+      "step": 1605
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.94788356947749e-06,
+      "loss": 0.0309,
+      "step": 1606
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.937460407942312e-06,
+      "loss": 0.0303,
+      "step": 1607
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.927037314353821e-06,
+      "loss": 0.0308,
+      "step": 1608
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.91661430003627e-06,
+      "loss": 0.0261,
+      "step": 1609
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.906191376313837e-06,
+      "loss": 0.0274,
+      "step": 1610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.895768554510593e-06,
+      "loss": 0.0207,
+      "step": 1611
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.88534584595051e-06,
+      "loss": 0.0234,
+      "step": 1612
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.874923261957428e-06,
+      "loss": 0.0283,
+      "step": 1613
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.864500813855051e-06,
+      "loss": 0.0319,
+      "step": 1614
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.854078512966937e-06,
+      "loss": 0.029,
+      "step": 1615
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.84365637061649e-06,
+      "loss": 0.0259,
+      "step": 1616
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.833234398126938e-06,
+      "loss": 0.0324,
+      "step": 1617
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.822812606821317e-06,
+      "loss": 0.0321,
+      "step": 1618
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.812391008022472e-06,
+      "loss": 0.0334,
+      "step": 1619
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.801969613053047e-06,
+      "loss": 0.0242,
+      "step": 1620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791548433235452e-06,
+      "loss": 0.0342,
+      "step": 1621
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.781127479891872e-06,
+      "loss": 0.0304,
+      "step": 1622
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.770706764344235e-06,
+      "loss": 0.0316,
+      "step": 1623
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.760286297914224e-06,
+      "loss": 0.0268,
+      "step": 1624
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.749866091923241e-06,
+      "loss": 0.0235,
+      "step": 1625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.739446157692416e-06,
+      "loss": 0.0306,
+      "step": 1626
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.729026506542566e-06,
+      "loss": 0.0276,
+      "step": 1627
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.71860714979422e-06,
+      "loss": 0.0294,
+      "step": 1628
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.708188098767577e-06,
+      "loss": 0.0224,
+      "step": 1629
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.697769364782502e-06,
+      "loss": 0.031,
+      "step": 1630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.687350959158525e-06,
+      "loss": 0.0315,
+      "step": 1631
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.676932893214806e-06,
+      "loss": 0.0341,
+      "step": 1632
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.666515178270146e-06,
+      "loss": 0.0272,
+      "step": 1633
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.65609782564296e-06,
+      "loss": 0.0317,
+      "step": 1634
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.645680846651281e-06,
+      "loss": 0.0299,
+      "step": 1635
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.635264252612712e-06,
+      "loss": 0.0306,
+      "step": 1636
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.62484805484446e-06,
+      "loss": 0.0334,
+      "step": 1637
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.614432264663288e-06,
+      "loss": 0.0301,
+      "step": 1638
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.604016893385528e-06,
+      "loss": 0.0295,
+      "step": 1639
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.593601952327049e-06,
+      "loss": 0.0257,
+      "step": 1640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.58318745280325e-06,
+      "loss": 0.0239,
+      "step": 1641
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.572773406129056e-06,
+      "loss": 0.0307,
+      "step": 1642
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.562359823618901e-06,
+      "loss": 0.0269,
+      "step": 1643
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.551946716586712e-06,
+      "loss": 0.029,
+      "step": 1644
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.541534096345896e-06,
+      "loss": 0.0259,
+      "step": 1645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.531121974209334e-06,
+      "loss": 0.0253,
+      "step": 1646
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.52071036148937e-06,
+      "loss": 0.0261,
+      "step": 1647
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.510299269497791e-06,
+      "loss": 0.0294,
+      "step": 1648
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.499888709545816e-06,
+      "loss": 0.0222,
+      "step": 1649
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.489478692944087e-06,
+      "loss": 0.0153,
+      "step": 1650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.479069231002658e-06,
+      "loss": 0.0332,
+      "step": 1651
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.46866033503098e-06,
+      "loss": 0.0235,
+      "step": 1652
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.458252016337891e-06,
+      "loss": 0.0232,
+      "step": 1653
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.447844286231594e-06,
+      "loss": 0.0222,
+      "step": 1654
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.437437156019657e-06,
+      "loss": 0.0258,
+      "step": 1655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.427030637009002e-06,
+      "loss": 0.0292,
+      "step": 1656
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.416624740505882e-06,
+      "loss": 0.0279,
+      "step": 1657
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.406219477815868e-06,
+      "loss": 0.0379,
+      "step": 1658
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.395814860243848e-06,
+      "loss": 0.0295,
+      "step": 1659
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.385410899094014e-06,
+      "loss": 0.0285,
+      "step": 1660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.375007605669835e-06,
+      "loss": 0.03,
+      "step": 1661
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.364604991274069e-06,
+      "loss": 0.0237,
+      "step": 1662
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.354203067208711e-06,
+      "loss": 0.0177,
+      "step": 1663
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.343801844775035e-06,
+      "loss": 0.027,
+      "step": 1664
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.33340133527353e-06,
+      "loss": 0.028,
+      "step": 1665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32300155000393e-06,
+      "loss": 0.0298,
+      "step": 1666
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.312602500265162e-06,
+      "loss": 0.0258,
+      "step": 1667
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.302204197355368e-06,
+      "loss": 0.025,
+      "step": 1668
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.291806652571873e-06,
+      "loss": 0.0287,
+      "step": 1669
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.28140987721118e-06,
+      "loss": 0.0228,
+      "step": 1670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.27101388256896e-06,
+      "loss": 0.0217,
+      "step": 1671
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.260618679940025e-06,
+      "loss": 0.0247,
+      "step": 1672
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.250224280618337e-06,
+      "loss": 0.021,
+      "step": 1673
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.239830695896978e-06,
+      "loss": 0.0257,
+      "step": 1674
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.229437937068155e-06,
+      "loss": 0.0221,
+      "step": 1675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.219046015423162e-06,
+      "loss": 0.0241,
+      "step": 1676
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.2086549422524e-06,
+      "loss": 0.0217,
+      "step": 1677
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.198264728845332e-06,
+      "loss": 0.0244,
+      "step": 1678
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.187875386490504e-06,
+      "loss": 0.0325,
+      "step": 1679
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.177486926475507e-06,
+      "loss": 0.0215,
+      "step": 1680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.167099360086967e-06,
+      "loss": 0.0197,
+      "step": 1681
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.156712698610546e-06,
+      "loss": 0.0268,
+      "step": 1682
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.146326953330929e-06,
+      "loss": 0.0286,
+      "step": 1683
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.135942135531797e-06,
+      "loss": 0.0265,
+      "step": 1684
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12555825649582e-06,
+      "loss": 0.0264,
+      "step": 1685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.115175327504656e-06,
+      "loss": 0.0239,
+      "step": 1686
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.104793359838931e-06,
+      "loss": 0.0213,
+      "step": 1687
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.094412364778222e-06,
+      "loss": 0.03,
+      "step": 1688
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084032353601053e-06,
+      "loss": 0.0334,
+      "step": 1689
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.073653337584872e-06,
+      "loss": 0.0281,
+      "step": 1690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.063275328006055e-06,
+      "loss": 0.0264,
+      "step": 1691
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.05289833613988e-06,
+      "loss": 0.0309,
+      "step": 1692
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.042522373260522e-06,
+      "loss": 0.0299,
+      "step": 1693
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.03214745064103e-06,
+      "loss": 0.0315,
+      "step": 1694
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.02177357955333e-06,
+      "loss": 0.0371,
+      "step": 1695
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.01140077126821e-06,
+      "loss": 0.0362,
+      "step": 1696
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00102903705529e-06,
+      "loss": 0.0326,
+      "step": 1697
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.990658388183038e-06,
+      "loss": 0.0275,
+      "step": 1698
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.98028883591873e-06,
+      "loss": 0.0286,
+      "step": 1699
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.969920391528459e-06,
+      "loss": 0.0228,
+      "step": 1700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.959553066277104e-06,
+      "loss": 0.0274,
+      "step": 1701
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.949186871428347e-06,
+      "loss": 0.0282,
+      "step": 1702
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.938821818244626e-06,
+      "loss": 0.032,
+      "step": 1703
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.928457917987137e-06,
+      "loss": 0.025,
+      "step": 1704
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.918095181915833e-06,
+      "loss": 0.0259,
+      "step": 1705
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.9077336212894e-06,
+      "loss": 0.027,
+      "step": 1706
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.897373247365249e-06,
+      "loss": 0.0292,
+      "step": 1707
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.887014071399488e-06,
+      "loss": 0.0294,
+      "step": 1708
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.876656104646942e-06,
+      "loss": 0.0266,
+      "step": 1709
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.86629935836111e-06,
+      "loss": 0.0272,
+      "step": 1710
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.855943843794171e-06,
+      "loss": 0.032,
+      "step": 1711
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.84558957219696e-06,
+      "loss": 0.0236,
+      "step": 1712
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.835236554818967e-06,
+      "loss": 0.0288,
+      "step": 1713
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.82488480290832e-06,
+      "loss": 0.0238,
+      "step": 1714
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.814534327711768e-06,
+      "loss": 0.0288,
+      "step": 1715
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.804185140474675e-06,
+      "loss": 0.0255,
+      "step": 1716
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.793837252441e-06,
+      "loss": 0.0311,
+      "step": 1717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.783490674853302e-06,
+      "loss": 0.0379,
+      "step": 1718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.773145418952708e-06,
+      "loss": 0.0385,
+      "step": 1719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.762801495978914e-06,
+      "loss": 0.0234,
+      "step": 1720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.752458917170157e-06,
+      "loss": 0.0321,
+      "step": 1721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.742117693763229e-06,
+      "loss": 0.0241,
+      "step": 1722
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.731777836993437e-06,
+      "loss": 0.0281,
+      "step": 1723
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.721439358094613e-06,
+      "loss": 0.0238,
+      "step": 1724
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.71110226829908e-06,
+      "loss": 0.0262,
+      "step": 1725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.700766578837664e-06,
+      "loss": 0.0263,
+      "step": 1726
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.690432300939664e-06,
+      "loss": 0.0303,
+      "step": 1727
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.680099445832843e-06,
+      "loss": 0.0298,
+      "step": 1728
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.669768024743429e-06,
+      "loss": 0.023,
+      "step": 1729
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.659438048896072e-06,
+      "loss": 0.0179,
+      "step": 1730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.649109529513873e-06,
+      "loss": 0.0358,
+      "step": 1731
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.638782477818335e-06,
+      "loss": 0.0249,
+      "step": 1732
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.628456905029383e-06,
+      "loss": 0.034,
+      "step": 1733
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.618132822365312e-06,
+      "loss": 0.0262,
+      "step": 1734
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.607810241042819e-06,
+      "loss": 0.022,
+      "step": 1735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.597489172276959e-06,
+      "loss": 0.0306,
+      "step": 1736
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.587169627281149e-06,
+      "loss": 0.0262,
+      "step": 1737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.025,
+      "step": 1738
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.566535153445049e-06,
+      "loss": 0.025,
+      "step": 1739
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.556220247023257e-06,
+      "loss": 0.0301,
+      "step": 1740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.545906909208498e-06,
+      "loss": 0.0393,
+      "step": 1741
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.535595151205787e-06,
+      "loss": 0.0232,
+      "step": 1742
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.525284984218418e-06,
+      "loss": 0.0298,
+      "step": 1743
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.514976419447963e-06,
+      "loss": 0.029,
+      "step": 1744
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.504669468094256e-06,
+      "loss": 0.0384,
+      "step": 1745
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.494364141355368e-06,
+      "loss": 0.0237,
+      "step": 1746
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.484060450427616e-06,
+      "loss": 0.0234,
+      "step": 1747
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.47375840650553e-06,
+      "loss": 0.0317,
+      "step": 1748
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.463458020781855e-06,
+      "loss": 0.0297,
+      "step": 1749
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.453159304447535e-06,
+      "loss": 0.0314,
+      "step": 1750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.442862268691703e-06,
+      "loss": 0.0263,
+      "step": 1751
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.43256692470166e-06,
+      "loss": 0.0358,
+      "step": 1752
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.422273283662869e-06,
+      "loss": 0.026,
+      "step": 1753
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.41198135675895e-06,
+      "loss": 0.0201,
+      "step": 1754
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.401691155171654e-06,
+      "loss": 0.0318,
+      "step": 1755
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.391402690080864e-06,
+      "loss": 0.0236,
+      "step": 1756
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.381115972664563e-06,
+      "loss": 0.0228,
+      "step": 1757
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.370831014098854e-06,
+      "loss": 0.0208,
+      "step": 1758
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.360547825557912e-06,
+      "loss": 0.0324,
+      "step": 1759
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.350266418214006e-06,
+      "loss": 0.027,
+      "step": 1760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.339986803237446e-06,
+      "loss": 0.0324,
+      "step": 1761
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.32970899179662e-06,
+      "loss": 0.03,
+      "step": 1762
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.31943299505794e-06,
+      "loss": 0.0273,
+      "step": 1763
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.309158824185855e-06,
+      "loss": 0.0249,
+      "step": 1764
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.298886490342827e-06,
+      "loss": 0.027,
+      "step": 1765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.288616004689321e-06,
+      "loss": 0.0248,
+      "step": 1766
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.278347378383794e-06,
+      "loss": 0.0365,
+      "step": 1767
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.268080622582682e-06,
+      "loss": 0.0208,
+      "step": 1768
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.2578157484404e-06,
+      "loss": 0.0263,
+      "step": 1769
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.247552767109301e-06,
+      "loss": 0.0311,
+      "step": 1770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.237291689739692e-06,
+      "loss": 0.0236,
+      "step": 1771
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.227032527479807e-06,
+      "loss": 0.0287,
+      "step": 1772
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.216775291475808e-06,
+      "loss": 0.0258,
+      "step": 1773
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.206519992871757e-06,
+      "loss": 0.024,
+      "step": 1774
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.196266642809607e-06,
+      "loss": 0.0244,
+      "step": 1775
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.1860152524292e-06,
+      "loss": 0.0258,
+      "step": 1776
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.175765832868252e-06,
+      "loss": 0.0314,
+      "step": 1777
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.165518395262335e-06,
+      "loss": 0.0283,
+      "step": 1778
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.155272950744859e-06,
+      "loss": 0.0285,
+      "step": 1779
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.14502951044708e-06,
+      "loss": 0.0236,
+      "step": 1780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.134788085498073e-06,
+      "loss": 0.0349,
+      "step": 1781
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.124548687024723e-06,
+      "loss": 0.0321,
+      "step": 1782
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.114311326151716e-06,
+      "loss": 0.0328,
+      "step": 1783
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.104076014001514e-06,
+      "loss": 0.0228,
+      "step": 1784
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.093842761694366e-06,
+      "loss": 0.0355,
+      "step": 1785
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.083611580348278e-06,
+      "loss": 0.031,
+      "step": 1786
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.073382481079005e-06,
+      "loss": 0.0284,
+      "step": 1787
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.063155475000037e-06,
+      "loss": 0.0293,
+      "step": 1788
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.052930573222597e-06,
+      "loss": 0.0295,
+      "step": 1789
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.042707786855618e-06,
+      "loss": 0.0258,
+      "step": 1790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.032487127005732e-06,
+      "loss": 0.0245,
+      "step": 1791
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.02226860477727e-06,
+      "loss": 0.0233,
+      "step": 1792
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.01205223127223e-06,
+      "loss": 0.0255,
+      "step": 1793
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.001838017590278e-06,
+      "loss": 0.0302,
+      "step": 1794
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.991625974828735e-06,
+      "loss": 0.0283,
+      "step": 1795
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.981416114082568e-06,
+      "loss": 0.0399,
+      "step": 1796
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.971208446444366e-06,
+      "loss": 0.0261,
+      "step": 1797
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.961002983004335e-06,
+      "loss": 0.0296,
+      "step": 1798
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.950799734850292e-06,
+      "loss": 0.0319,
+      "step": 1799
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.940598713067647e-06,
+      "loss": 0.0159,
+      "step": 1800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.930399928739385e-06,
+      "loss": 0.0302,
+      "step": 1801
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.920203392946064e-06,
+      "loss": 0.0246,
+      "step": 1802
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.910009116765796e-06,
+      "loss": 0.0273,
+      "step": 1803
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.899817111274249e-06,
+      "loss": 0.031,
+      "step": 1804
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.889627387544611e-06,
+      "loss": 0.0247,
+      "step": 1805
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.879439956647594e-06,
+      "loss": 0.0252,
+      "step": 1806
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.86925482965142e-06,
+      "loss": 0.0258,
+      "step": 1807
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.859072017621811e-06,
+      "loss": 0.032,
+      "step": 1808
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.848891531621977e-06,
+      "loss": 0.03,
+      "step": 1809
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.838713382712583e-06,
+      "loss": 0.0305,
+      "step": 1810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.828537581951773e-06,
+      "loss": 0.0239,
+      "step": 1811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.818364140395138e-06,
+      "loss": 0.0247,
+      "step": 1812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.808193069095698e-06,
+      "loss": 0.0234,
+      "step": 1813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.798024379103903e-06,
+      "loss": 0.0352,
+      "step": 1814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.787858081467609e-06,
+      "loss": 0.0246,
+      "step": 1815
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.777694187232085e-06,
+      "loss": 0.0233,
+      "step": 1816
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.76753270743998e-06,
+      "loss": 0.0286,
+      "step": 1817
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.757373653131323e-06,
+      "loss": 0.0289,
+      "step": 1818
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.7472170353435e-06,
+      "loss": 0.0248,
+      "step": 1819
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.737062865111265e-06,
+      "loss": 0.0212,
+      "step": 1820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.726911153466699e-06,
+      "loss": 0.0271,
+      "step": 1821
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.716761911439217e-06,
+      "loss": 0.025,
+      "step": 1822
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.706615150055557e-06,
+      "loss": 0.0275,
+      "step": 1823
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.696470880339751e-06,
+      "loss": 0.0211,
+      "step": 1824
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.686329113313129e-06,
+      "loss": 0.0279,
+      "step": 1825
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.676189859994299e-06,
+      "loss": 0.0301,
+      "step": 1826
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.666053131399151e-06,
+      "loss": 0.0308,
+      "step": 1827
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.655918938540808e-06,
+      "loss": 0.036,
+      "step": 1828
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.64578729242966e-06,
+      "loss": 0.0319,
+      "step": 1829
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.63565820407332e-06,
+      "loss": 0.022,
+      "step": 1830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.625531684476624e-06,
+      "loss": 0.0327,
+      "step": 1831
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.615407744641618e-06,
+      "loss": 0.0219,
+      "step": 1832
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.605286395567544e-06,
+      "loss": 0.0344,
+      "step": 1833
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.595167648250825e-06,
+      "loss": 0.0224,
+      "step": 1834
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.585051513685067e-06,
+      "loss": 0.0292,
+      "step": 1835
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.574938002861033e-06,
+      "loss": 0.0253,
+      "step": 1836
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5648271267666295e-06,
+      "loss": 0.0268,
+      "step": 1837
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.554718896386907e-06,
+      "loss": 0.0258,
+      "step": 1838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5446133227040415e-06,
+      "loss": 0.0299,
+      "step": 1839
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.534510416697322e-06,
+      "loss": 0.0255,
+      "step": 1840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.524410189343139e-06,
+      "loss": 0.0188,
+      "step": 1841
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.514312651614964e-06,
+      "loss": 0.0375,
+      "step": 1842
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.504217814483364e-06,
+      "loss": 0.0289,
+      "step": 1843
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.494125688915959e-06,
+      "loss": 0.0232,
+      "step": 1844
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.484036285877429e-06,
+      "loss": 0.025,
+      "step": 1845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4739496163294875e-06,
+      "loss": 0.0168,
+      "step": 1846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.463865691230885e-06,
+      "loss": 0.022,
+      "step": 1847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.453784521537393e-06,
+      "loss": 0.0401,
+      "step": 1848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.443706118201782e-06,
+      "loss": 0.0222,
+      "step": 1849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.433630492173831e-06,
+      "loss": 0.0212,
+      "step": 1850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.423557654400275e-06,
+      "loss": 0.0394,
+      "step": 1851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.413487615824847e-06,
+      "loss": 0.0282,
+      "step": 1852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.403420387388223e-06,
+      "loss": 0.0357,
+      "step": 1853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.393355980028039e-06,
+      "loss": 0.0278,
+      "step": 1854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.383294404678842e-06,
+      "loss": 0.0299,
+      "step": 1855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.373235672272132e-06,
+      "loss": 0.0262,
+      "step": 1856
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.363179793736297e-06,
+      "loss": 0.0212,
+      "step": 1857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.35312677999664e-06,
+      "loss": 0.0309,
+      "step": 1858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.343076641975345e-06,
+      "loss": 0.025,
+      "step": 1859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.333029390591466e-06,
+      "loss": 0.0346,
+      "step": 1860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.322985036760928e-06,
+      "loss": 0.0269,
+      "step": 1861
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.312943591396508e-06,
+      "loss": 0.0168,
+      "step": 1862
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3029050654078235e-06,
+      "loss": 0.0249,
+      "step": 1863
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.292869469701315e-06,
+      "loss": 0.023,
+      "step": 1864
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.282836815180241e-06,
+      "loss": 0.0244,
+      "step": 1865
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.272807112744666e-06,
+      "loss": 0.0194,
+      "step": 1866
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.262780373291452e-06,
+      "loss": 0.0238,
+      "step": 1867
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.252756607714238e-06,
+      "loss": 0.023,
+      "step": 1868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.242735826903422e-06,
+      "loss": 0.023,
+      "step": 1869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2327180417461716e-06,
+      "loss": 0.0269,
+      "step": 1870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.222703263126402e-06,
+      "loss": 0.0315,
+      "step": 1871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.212691501924753e-06,
+      "loss": 0.0243,
+      "step": 1872
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.2026827690185874e-06,
+      "loss": 0.0274,
+      "step": 1873
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.192677075281978e-06,
+      "loss": 0.0279,
+      "step": 1874
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.0285,
+      "step": 1875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.172674848797218e-06,
+      "loss": 0.0282,
+      "step": 1876
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.162678337780664e-06,
+      "loss": 0.0282,
+      "step": 1877
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1526849093968234e-06,
+      "loss": 0.0271,
+      "step": 1878
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.142694574503151e-06,
+      "loss": 0.0205,
+      "step": 1879
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.132707343953732e-06,
+      "loss": 0.0199,
+      "step": 1880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1227232285992855e-06,
+      "loss": 0.0254,
+      "step": 1881
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.112742239287127e-06,
+      "loss": 0.0167,
+      "step": 1882
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1027643868612e-06,
+      "loss": 0.0327,
+      "step": 1883
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.092789682162024e-06,
+      "loss": 0.0299,
+      "step": 1884
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.082818136026706e-06,
+      "loss": 0.0206,
+      "step": 1885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.072849759288914e-06,
+      "loss": 0.0212,
+      "step": 1886
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.062884562778883e-06,
+      "loss": 0.0236,
+      "step": 1887
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.052922557323383e-06,
+      "loss": 0.0298,
+      "step": 1888
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.042963753745722e-06,
+      "loss": 0.0307,
+      "step": 1889
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.033008162865732e-06,
+      "loss": 0.0357,
+      "step": 1890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.023055795499747e-06,
+      "loss": 0.0243,
+      "step": 1891
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.013106662460605e-06,
+      "loss": 0.0251,
+      "step": 1892
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.003160774557625e-06,
+      "loss": 0.0246,
+      "step": 1893
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.993218142596609e-06,
+      "loss": 0.0236,
+      "step": 1894
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.98327877737981e-06,
+      "loss": 0.0347,
+      "step": 1895
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9733426897059375e-06,
+      "loss": 0.0341,
+      "step": 1896
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.963409890370141e-06,
+      "loss": 0.0313,
+      "step": 1897
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.953480390164001e-06,
+      "loss": 0.034,
+      "step": 1898
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.943554199875506e-06,
+      "loss": 0.0257,
+      "step": 1899
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.93363133028905e-06,
+      "loss": 0.0339,
+      "step": 1900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.923711792185419e-06,
+      "loss": 0.0286,
+      "step": 1901
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.913795596341786e-06,
+      "loss": 0.0376,
+      "step": 1902
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.903882753531687e-06,
+      "loss": 0.0354,
+      "step": 1903
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8939732745250145e-06,
+      "loss": 0.0231,
+      "step": 1904
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.884067170088007e-06,
+      "loss": 0.0282,
+      "step": 1905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.874164450983238e-06,
+      "loss": 0.0225,
+      "step": 1906
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.864265127969606e-06,
+      "loss": 0.0242,
+      "step": 1907
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.854369211802314e-06,
+      "loss": 0.0259,
+      "step": 1908
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.844476713232863e-06,
+      "loss": 0.0367,
+      "step": 1909
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.834587643009047e-06,
+      "loss": 0.031,
+      "step": 1910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.824702011874933e-06,
+      "loss": 0.0183,
+      "step": 1911
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.81481983057085e-06,
+      "loss": 0.0225,
+      "step": 1912
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.804941109833375e-06,
+      "loss": 0.029,
+      "step": 1913
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7950658603953355e-06,
+      "loss": 0.0219,
+      "step": 1914
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.78519409298578e-06,
+      "loss": 0.0203,
+      "step": 1915
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.775325818329974e-06,
+      "loss": 0.0344,
+      "step": 1916
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.765461047149397e-06,
+      "loss": 0.0292,
+      "step": 1917
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.755599790161707e-06,
+      "loss": 0.0235,
+      "step": 1918
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.745742058080753e-06,
+      "loss": 0.0265,
+      "step": 1919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.735887861616555e-06,
+      "loss": 0.0343,
+      "step": 1920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.726037211475296e-06,
+      "loss": 0.0356,
+      "step": 1921
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.716190118359293e-06,
+      "loss": 0.024,
+      "step": 1922
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.706346592967006e-06,
+      "loss": 0.0255,
+      "step": 1923
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6965066459930195e-06,
+      "loss": 0.0269,
+      "step": 1924
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.686670288128032e-06,
+      "loss": 0.0275,
+      "step": 1925
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.67683753005884e-06,
+      "loss": 0.0315,
+      "step": 1926
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.667008382468323e-06,
+      "loss": 0.0233,
+      "step": 1927
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.657182856035447e-06,
+      "loss": 0.0321,
+      "step": 1928
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6473609614352405e-06,
+      "loss": 0.0229,
+      "step": 1929
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.637542709338789e-06,
+      "loss": 0.03,
+      "step": 1930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.627728110413214e-06,
+      "loss": 0.0254,
+      "step": 1931
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.617917175321668e-06,
+      "loss": 0.0237,
+      "step": 1932
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6081099147233344e-06,
+      "loss": 0.0188,
+      "step": 1933
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.598306339273393e-06,
+      "loss": 0.0282,
+      "step": 1934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.588506459623023e-06,
+      "loss": 0.0225,
+      "step": 1935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.578710286419388e-06,
+      "loss": 0.018,
+      "step": 1936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.568917830305625e-06,
+      "loss": 0.0209,
+      "step": 1937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.559129101920836e-06,
+      "loss": 0.0283,
+      "step": 1938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.54934411190007e-06,
+      "loss": 0.0276,
+      "step": 1939
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.539562870874309e-06,
+      "loss": 0.0344,
+      "step": 1940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.529785389470469e-06,
+      "loss": 0.0208,
+      "step": 1941
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.520011678311382e-06,
+      "loss": 0.0268,
+      "step": 1942
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.510241748015779e-06,
+      "loss": 0.0286,
+      "step": 1943
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.500475609198291e-06,
+      "loss": 0.0251,
+      "step": 1944
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.490713272469414e-06,
+      "loss": 0.0172,
+      "step": 1945
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.480954748435531e-06,
+      "loss": 0.0305,
+      "step": 1946
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4712000476988715e-06,
+      "loss": 0.0294,
+      "step": 1947
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4614491808575215e-06,
+      "loss": 0.026,
+      "step": 1948
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.451702158505384e-06,
+      "loss": 0.0183,
+      "step": 1949
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4419589912322025e-06,
+      "loss": 0.0297,
+      "step": 1950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4322196896235236e-06,
+      "loss": 0.0298,
+      "step": 1951
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.422484264260698e-06,
+      "loss": 0.0217,
+      "step": 1952
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.412752725720864e-06,
+      "loss": 0.0343,
+      "step": 1953
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.403025084576932e-06,
+      "loss": 0.0271,
+      "step": 1954
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.393301351397583e-06,
+      "loss": 0.0257,
+      "step": 1955
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.383581536747252e-06,
+      "loss": 0.0249,
+      "step": 1956
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3738656511861195e-06,
+      "loss": 0.0339,
+      "step": 1957
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.364153705270088e-06,
+      "loss": 0.0165,
+      "step": 1958
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3544457095507885e-06,
+      "loss": 0.0236,
+      "step": 1959
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.344741674575553e-06,
+      "loss": 0.028,
+      "step": 1960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.335041610887422e-06,
+      "loss": 0.0291,
+      "step": 1961
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.325345529025107e-06,
+      "loss": 0.0282,
+      "step": 1962
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.315653439523003e-06,
+      "loss": 0.0228,
+      "step": 1963
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.305965352911162e-06,
+      "loss": 0.0327,
+      "step": 1964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.296281279715294e-06,
+      "loss": 0.0283,
+      "step": 1965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2866012304567436e-06,
+      "loss": 0.0189,
+      "step": 1966
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.276925215652482e-06,
+      "loss": 0.0228,
+      "step": 1967
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.267253245815098e-06,
+      "loss": 0.0272,
+      "step": 1968
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.257585331452791e-06,
+      "loss": 0.0408,
+      "step": 1969
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.247921483069352e-06,
+      "loss": 0.0178,
+      "step": 1970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.238261711164146e-06,
+      "loss": 0.0219,
+      "step": 1971
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.228606026232118e-06,
+      "loss": 0.0304,
+      "step": 1972
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.218954438763773e-06,
+      "loss": 0.0209,
+      "step": 1973
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.209306959245161e-06,
+      "loss": 0.0206,
+      "step": 1974
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1996635981578755e-06,
+      "loss": 0.0193,
+      "step": 1975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.190024365979017e-06,
+      "loss": 0.0262,
+      "step": 1976
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.180389273181224e-06,
+      "loss": 0.0251,
+      "step": 1977
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.170758330232621e-06,
+      "loss": 0.0196,
+      "step": 1978
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.161131547596833e-06,
+      "loss": 0.0254,
+      "step": 1979
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1515089357329565e-06,
+      "loss": 0.0194,
+      "step": 1980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.141890505095566e-06,
+      "loss": 0.0257,
+      "step": 1981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1322762661346865e-06,
+      "loss": 0.0308,
+      "step": 1982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.122666229295792e-06,
+      "loss": 0.0254,
+      "step": 1983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.113060405019794e-06,
+      "loss": 0.0304,
+      "step": 1984
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.103458803743018e-06,
+      "loss": 0.0229,
+      "step": 1985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.093861435897208e-06,
+      "loss": 0.0249,
+      "step": 1986
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0842683119095095e-06,
+      "loss": 0.0199,
+      "step": 1987
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0746794422024555e-06,
+      "loss": 0.0252,
+      "step": 1988
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.065094837193954e-06,
+      "loss": 0.026,
+      "step": 1989
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.055514507297284e-06,
+      "loss": 0.0211,
+      "step": 1990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.045938462921077e-06,
+      "loss": 0.0246,
+      "step": 1991
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.03636671446931e-06,
+      "loss": 0.0349,
+      "step": 1992
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.026799272341296e-06,
+      "loss": 0.0272,
+      "step": 1993
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0172361469316585e-06,
+      "loss": 0.0189,
+      "step": 1994
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.00767734863034e-06,
+      "loss": 0.0271,
+      "step": 1995
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.998122887822582e-06,
+      "loss": 0.0243,
+      "step": 1996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.988572774888913e-06,
+      "loss": 0.019,
+      "step": 1997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.979027020205131e-06,
+      "loss": 0.0235,
+      "step": 1998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9694856341423045e-06,
+      "loss": 0.0225,
+      "step": 1999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9599486270667595e-06,
+      "loss": 0.0233,
+      "step": 2000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.950416009340056e-06,
+      "loss": 0.0252,
+      "step": 2001
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.940887791318993e-06,
+      "loss": 0.0243,
+      "step": 2002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.93136398335558e-06,
+      "loss": 0.0212,
+      "step": 2003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9218445957970435e-06,
+      "loss": 0.019,
+      "step": 2004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.912329638985808e-06,
+      "loss": 0.0198,
+      "step": 2005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.902819123259478e-06,
+      "loss": 0.0186,
+      "step": 2006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8933130589508314e-06,
+      "loss": 0.0175,
+      "step": 2007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.883811456387821e-06,
+      "loss": 0.0205,
+      "step": 2008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8743143258935435e-06,
+      "loss": 0.0269,
+      "step": 2009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8648216777862345e-06,
+      "loss": 0.0342,
+      "step": 2010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.855333522379272e-06,
+      "loss": 0.0214,
+      "step": 2011
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.0271,
+      "step": 2012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.836370730895427e-06,
+      "loss": 0.0217,
+      "step": 2013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8268961154208395e-06,
+      "loss": 0.0261,
+      "step": 2014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.817426033851152e-06,
+      "loss": 0.029,
+      "step": 2015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8079604964752126e-06,
+      "loss": 0.0254,
+      "step": 2016
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.798499513576934e-06,
+      "loss": 0.0322,
+      "step": 2017
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7890430954352895e-06,
+      "loss": 0.0185,
+      "step": 2018
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.779591252324286e-06,
+      "loss": 0.0177,
+      "step": 2019
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7701439945129625e-06,
+      "loss": 0.0273,
+      "step": 2020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7607013322653635e-06,
+      "loss": 0.0293,
+      "step": 2021
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.751263275840562e-06,
+      "loss": 0.024,
+      "step": 2022
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.741829835492611e-06,
+      "loss": 0.0186,
+      "step": 2023
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.73240102147056e-06,
+      "loss": 0.02,
+      "step": 2024
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.722976844018414e-06,
+      "loss": 0.0204,
+      "step": 2025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.713557313375162e-06,
+      "loss": 0.0253,
+      "step": 2026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.704142439774732e-06,
+      "loss": 0.0268,
+      "step": 2027
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.694732233445996e-06,
+      "loss": 0.0231,
+      "step": 2028
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.685326704612753e-06,
+      "loss": 0.034,
+      "step": 2029
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.675925863493721e-06,
+      "loss": 0.0296,
+      "step": 2030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.666529720302525e-06,
+      "loss": 0.029,
+      "step": 2031
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.657138285247687e-06,
+      "loss": 0.0303,
+      "step": 2032
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.647751568532616e-06,
+      "loss": 0.0165,
+      "step": 2033
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.638369580355587e-06,
+      "loss": 0.0267,
+      "step": 2034
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.628992330909744e-06,
+      "loss": 0.0262,
+      "step": 2035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.619619830383083e-06,
+      "loss": 0.0232,
+      "step": 2036
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.610252088958434e-06,
+      "loss": 0.0187,
+      "step": 2037
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.600889116813466e-06,
+      "loss": 0.0189,
+      "step": 2038
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.591530924120659e-06,
+      "loss": 0.0294,
+      "step": 2039
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5821775210473015e-06,
+      "loss": 0.0212,
+      "step": 2040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5728289177554805e-06,
+      "loss": 0.0343,
+      "step": 2041
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.563485124402068e-06,
+      "loss": 0.0213,
+      "step": 2042
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.554146151138706e-06,
+      "loss": 0.0195,
+      "step": 2043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5448120081118085e-06,
+      "loss": 0.0311,
+      "step": 2044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.53548270546253e-06,
+      "loss": 0.0233,
+      "step": 2045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.526158253326777e-06,
+      "loss": 0.0193,
+      "step": 2046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5168386618351785e-06,
+      "loss": 0.0202,
+      "step": 2047
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.507523941113086e-06,
+      "loss": 0.0233,
+      "step": 2048
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.498214101280561e-06,
+      "loss": 0.0201,
+      "step": 2049
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.488909152452357e-06,
+      "loss": 0.0272,
+      "step": 2050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.479609104737917e-06,
+      "loss": 0.0204,
+      "step": 2051
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4703139682413585e-06,
+      "loss": 0.0263,
+      "step": 2052
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.461023753061463e-06,
+      "loss": 0.0252,
+      "step": 2053
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.451738469291666e-06,
+      "loss": 0.0259,
+      "step": 2054
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.442458127020045e-06,
+      "loss": 0.0216,
+      "step": 2055
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.433182736329305e-06,
+      "loss": 0.0244,
+      "step": 2056
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.423912307296779e-06,
+      "loss": 0.0191,
+      "step": 2057
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.414646849994403e-06,
+      "loss": 0.0242,
+      "step": 2058
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.405386374488709e-06,
+      "loss": 0.0198,
+      "step": 2059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.396130890840834e-06,
+      "loss": 0.0159,
+      "step": 2060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.386880409106465e-06,
+      "loss": 0.0243,
+      "step": 2061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.377634939335874e-06,
+      "loss": 0.027,
+      "step": 2062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.368394491573876e-06,
+      "loss": 0.0239,
+      "step": 2063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.359159075859848e-06,
+      "loss": 0.0282,
+      "step": 2064
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.349928702227677e-06,
+      "loss": 0.0252,
+      "step": 2065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3407033807057855e-06,
+      "loss": 0.0244,
+      "step": 2066
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.331483121317098e-06,
+      "loss": 0.0249,
+      "step": 2067
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.322267934079056e-06,
+      "loss": 0.0239,
+      "step": 2068
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.313057829003578e-06,
+      "loss": 0.0265,
+      "step": 2069
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.303852816097057e-06,
+      "loss": 0.0406,
+      "step": 2070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.294652905360356e-06,
+      "loss": 0.0206,
+      "step": 2071
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.285458106788808e-06,
+      "loss": 0.019,
+      "step": 2072
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.27626843037218e-06,
+      "loss": 0.0176,
+      "step": 2073
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.267083886094668e-06,
+      "loss": 0.0205,
+      "step": 2074
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2579044839349e-06,
+      "loss": 0.0314,
+      "step": 2075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.248730233865926e-06,
+      "loss": 0.0355,
+      "step": 2076
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.239561145855183e-06,
+      "loss": 0.0208,
+      "step": 2077
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.230397229864515e-06,
+      "loss": 0.0253,
+      "step": 2078
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.221238495850121e-06,
+      "loss": 0.0212,
+      "step": 2079
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.212084953762599e-06,
+      "loss": 0.0213,
+      "step": 2080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.202936613546894e-06,
+      "loss": 0.0212,
+      "step": 2081
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.193793485142301e-06,
+      "loss": 0.023,
+      "step": 2082
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1846555784824404e-06,
+      "loss": 0.0246,
+      "step": 2083
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.17552290349528e-06,
+      "loss": 0.0247,
+      "step": 2084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.166395470103092e-06,
+      "loss": 0.0271,
+      "step": 2085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.157273288222455e-06,
+      "loss": 0.0364,
+      "step": 2086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1481563677642455e-06,
+      "loss": 0.0217,
+      "step": 2087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.139044718633609e-06,
+      "loss": 0.0242,
+      "step": 2088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.129938350729989e-06,
+      "loss": 0.0264,
+      "step": 2089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.120837273947073e-06,
+      "loss": 0.0235,
+      "step": 2090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.111741498172808e-06,
+      "loss": 0.0215,
+      "step": 2091
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.102651033289369e-06,
+      "loss": 0.0227,
+      "step": 2092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.09356588917318e-06,
+      "loss": 0.0235,
+      "step": 2093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.084486075694871e-06,
+      "loss": 0.0183,
+      "step": 2094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.075411602719284e-06,
+      "loss": 0.026,
+      "step": 2095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.06634248010546e-06,
+      "loss": 0.0185,
+      "step": 2096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.057278717706624e-06,
+      "loss": 0.0243,
+      "step": 2097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0482203253701815e-06,
+      "loss": 0.0242,
+      "step": 2098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0391673129377e-06,
+      "loss": 0.0237,
+      "step": 2099
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.030119690244903e-06,
+      "loss": 0.0209,
+      "step": 2100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.02107746712166e-06,
+      "loss": 0.0267,
+      "step": 2101
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.012040653391972e-06,
+      "loss": 0.0212,
+      "step": 2102
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.003009258873963e-06,
+      "loss": 0.0257,
+      "step": 2103
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.993983293379871e-06,
+      "loss": 0.0199,
+      "step": 2104
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.984962766716034e-06,
+      "loss": 0.0209,
+      "step": 2105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.975947688682881e-06,
+      "loss": 0.0239,
+      "step": 2106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9669380690749215e-06,
+      "loss": 0.0219,
+      "step": 2107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.957933917680735e-06,
+      "loss": 0.0187,
+      "step": 2108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.948935244282961e-06,
+      "loss": 0.0224,
+      "step": 2109
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.939942058658286e-06,
+      "loss": 0.0176,
+      "step": 2110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9309543705774335e-06,
+      "loss": 0.0255,
+      "step": 2111
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.921972189805154e-06,
+      "loss": 0.0257,
+      "step": 2112
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.912995526100216e-06,
+      "loss": 0.0223,
+      "step": 2113
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.904024389215401e-06,
+      "loss": 0.022,
+      "step": 2114
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.895058788897469e-06,
+      "loss": 0.0315,
+      "step": 2115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.886098734887177e-06,
+      "loss": 0.0206,
+      "step": 2116
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.87714423691925e-06,
+      "loss": 0.0179,
+      "step": 2117
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.868195304722391e-06,
+      "loss": 0.0175,
+      "step": 2118
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.859251948019236e-06,
+      "loss": 0.0189,
+      "step": 2119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.850314176526373e-06,
+      "loss": 0.0221,
+      "step": 2120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8413819999543185e-06,
+      "loss": 0.0325,
+      "step": 2121
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8324554280075255e-06,
+      "loss": 0.0327,
+      "step": 2122
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.823534470384333e-06,
+      "loss": 0.0213,
+      "step": 2123
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.814619136776998e-06,
+      "loss": 0.0291,
+      "step": 2124
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.805709436871659e-06,
+      "loss": 0.0214,
+      "step": 2125
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.796805380348336e-06,
+      "loss": 0.0238,
+      "step": 2126
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.78790697688093e-06,
+      "loss": 0.0231,
+      "step": 2127
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7790142361371735e-06,
+      "loss": 0.0268,
+      "step": 2128
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.77012716777867e-06,
+      "loss": 0.0309,
+      "step": 2129
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.761245781460846e-06,
+      "loss": 0.0168,
+      "step": 2130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.752370086832971e-06,
+      "loss": 0.0267,
+      "step": 2131
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.743500093538111e-06,
+      "loss": 0.024,
+      "step": 2132
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.73463581121315e-06,
+      "loss": 0.0228,
+      "step": 2133
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.725777249488761e-06,
+      "loss": 0.0261,
+      "step": 2134
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.716924417989413e-06,
+      "loss": 0.0197,
+      "step": 2135
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7080773263333405e-06,
+      "loss": 0.024,
+      "step": 2136
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.699235984132535e-06,
+      "loss": 0.0252,
+      "step": 2137
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.690400400992752e-06,
+      "loss": 0.0236,
+      "step": 2138
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.681570586513492e-06,
+      "loss": 0.0261,
+      "step": 2139
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.672746550287985e-06,
+      "loss": 0.0245,
+      "step": 2140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.663928301903175e-06,
+      "loss": 0.0256,
+      "step": 2141
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.655115850939722e-06,
+      "loss": 0.0197,
+      "step": 2142
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6463092069719996e-06,
+      "loss": 0.0289,
+      "step": 2143
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.637508379568057e-06,
+      "loss": 0.0262,
+      "step": 2144
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.628713378289634e-06,
+      "loss": 0.0284,
+      "step": 2145
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.619924212692125e-06,
+      "loss": 0.0179,
+      "step": 2146
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.611140892324606e-06,
+      "loss": 0.0204,
+      "step": 2147
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.602363426729788e-06,
+      "loss": 0.0265,
+      "step": 2148
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.0205,
+      "step": 2149
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.584826097997297e-06,
+      "loss": 0.0223,
+      "step": 2150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.576066253913209e-06,
+      "loss": 0.0336,
+      "step": 2151
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.567312302708965e-06,
+      "loss": 0.0263,
+      "step": 2152
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.558564253895374e-06,
+      "loss": 0.0297,
+      "step": 2153
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.549822116976827e-06,
+      "loss": 0.0174,
+      "step": 2154
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5410859014512975e-06,
+      "loss": 0.0242,
+      "step": 2155
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.532355616810321e-06,
+      "loss": 0.023,
+      "step": 2156
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.523631272538992e-06,
+      "loss": 0.0312,
+      "step": 2157
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.514912878115948e-06,
+      "loss": 0.0319,
+      "step": 2158
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5062004430133655e-06,
+      "loss": 0.0178,
+      "step": 2159
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.497493976696945e-06,
+      "loss": 0.0226,
+      "step": 2160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4887934886259025e-06,
+      "loss": 0.0214,
+      "step": 2161
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.480098988252958e-06,
+      "loss": 0.0247,
+      "step": 2162
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.471410485024327e-06,
+      "loss": 0.0292,
+      "step": 2163
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4627279883797096e-06,
+      "loss": 0.0266,
+      "step": 2164
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.454051507752279e-06,
+      "loss": 0.0269,
+      "step": 2165
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.445381052568674e-06,
+      "loss": 0.0266,
+      "step": 2166
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.436716632248985e-06,
+      "loss": 0.0203,
+      "step": 2167
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4280582562067465e-06,
+      "loss": 0.023,
+      "step": 2168
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.419405933848927e-06,
+      "loss": 0.0202,
+      "step": 2169
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.410759674575915e-06,
+      "loss": 0.0216,
+      "step": 2170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4021194877815165e-06,
+      "loss": 0.0241,
+      "step": 2171
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.393485382852936e-06,
+      "loss": 0.0249,
+      "step": 2172
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.384857369170772e-06,
+      "loss": 0.0221,
+      "step": 2173
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.376235456109003e-06,
+      "loss": 0.0241,
+      "step": 2174
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3676196530349845e-06,
+      "loss": 0.0182,
+      "step": 2175
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.359009969309429e-06,
+      "loss": 0.0289,
+      "step": 2176
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.350406414286401e-06,
+      "loss": 0.0255,
+      "step": 2177
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.341808997313308e-06,
+      "loss": 0.0332,
+      "step": 2178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3332177277308885e-06,
+      "loss": 0.0284,
+      "step": 2179
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.324632614873199e-06,
+      "loss": 0.0241,
+      "step": 2180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.316053668067619e-06,
+      "loss": 0.0249,
+      "step": 2181
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3074808966348105e-06,
+      "loss": 0.0187,
+      "step": 2182
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.298914309888736e-06,
+      "loss": 0.0217,
+      "step": 2183
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.290353917136639e-06,
+      "loss": 0.0225,
+      "step": 2184
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.281799727679039e-06,
+      "loss": 0.0206,
+      "step": 2185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2732517508097016e-06,
+      "loss": 0.0323,
+      "step": 2186
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.264709995815651e-06,
+      "loss": 0.0228,
+      "step": 2187
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.25617447197715e-06,
+      "loss": 0.0203,
+      "step": 2188
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.247645188567698e-06,
+      "loss": 0.022,
+      "step": 2189
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2391221548540094e-06,
+      "loss": 0.026,
+      "step": 2190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.230605380096001e-06,
+      "loss": 0.0244,
+      "step": 2191
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.222094873546797e-06,
+      "loss": 0.0131,
+      "step": 2192
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213590644452717e-06,
+      "loss": 0.0297,
+      "step": 2193
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.205092702053257e-06,
+      "loss": 0.0258,
+      "step": 2194
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1966010555810696e-06,
+      "loss": 0.0246,
+      "step": 2195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.18811571426198e-06,
+      "loss": 0.0192,
+      "step": 2196
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.179636687314967e-06,
+      "loss": 0.0189,
+      "step": 2197
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.171163983952144e-06,
+      "loss": 0.0319,
+      "step": 2198
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1626976133787445e-06,
+      "loss": 0.0221,
+      "step": 2199
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.154237584793132e-06,
+      "loss": 0.0235,
+      "step": 2200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.145783907386782e-06,
+      "loss": 0.0186,
+      "step": 2201
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1373365903442665e-06,
+      "loss": 0.0182,
+      "step": 2202
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1288956428432455e-06,
+      "loss": 0.0227,
+      "step": 2203
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.120461074054455e-06,
+      "loss": 0.019,
+      "step": 2204
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112032893141706e-06,
+      "loss": 0.019,
+      "step": 2205
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1036111092618725e-06,
+      "loss": 0.0179,
+      "step": 2206
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.095195731564878e-06,
+      "loss": 0.0168,
+      "step": 2207
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.086786769193673e-06,
+      "loss": 0.0197,
+      "step": 2208
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0783842312842506e-06,
+      "loss": 0.0286,
+      "step": 2209
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.069988126965625e-06,
+      "loss": 0.0223,
+      "step": 2210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.061598465359815e-06,
+      "loss": 0.0215,
+      "step": 2211
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.053215255581843e-06,
+      "loss": 0.0196,
+      "step": 2212
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.044838506739711e-06,
+      "loss": 0.0145,
+      "step": 2213
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.03646822793442e-06,
+      "loss": 0.0306,
+      "step": 2214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.028104428259929e-06,
+      "loss": 0.0224,
+      "step": 2215
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.019747116803164e-06,
+      "loss": 0.0193,
+      "step": 2216
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.011396302643989e-06,
+      "loss": 0.0271,
+      "step": 2217
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.003051994855226e-06,
+      "loss": 0.0193,
+      "step": 2218
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.99471420250262e-06,
+      "loss": 0.0225,
+      "step": 2219
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.986382934644836e-06,
+      "loss": 0.016,
+      "step": 2220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.978058200333453e-06,
+      "loss": 0.0249,
+      "step": 2221
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.969740008612951e-06,
+      "loss": 0.0261,
+      "step": 2222
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.961428368520702e-06,
+      "loss": 0.0231,
+      "step": 2223
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.953123289086958e-06,
+      "loss": 0.0242,
+      "step": 2224
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.944824779334845e-06,
+      "loss": 0.0181,
+      "step": 2225
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.936532848280351e-06,
+      "loss": 0.025,
+      "step": 2226
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.928247504932317e-06,
+      "loss": 0.0371,
+      "step": 2227
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.919968758292425e-06,
+      "loss": 0.029,
+      "step": 2228
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9116966173551915e-06,
+      "loss": 0.0186,
+      "step": 2229
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.903431091107957e-06,
+      "loss": 0.0232,
+      "step": 2230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.895172188530871e-06,
+      "loss": 0.0303,
+      "step": 2231
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8869199185968944e-06,
+      "loss": 0.0213,
+      "step": 2232
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.878674290271773e-06,
+      "loss": 0.0223,
+      "step": 2233
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8704353125140436e-06,
+      "loss": 0.0165,
+      "step": 2234
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.862202994275015e-06,
+      "loss": 0.0185,
+      "step": 2235
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.853977344498761e-06,
+      "loss": 0.0261,
+      "step": 2236
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.845758372122107e-06,
+      "loss": 0.0228,
+      "step": 2237
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.83754608607463e-06,
+      "loss": 0.0238,
+      "step": 2238
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.82934049527864e-06,
+      "loss": 0.0252,
+      "step": 2239
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.821141608649169e-06,
+      "loss": 0.0285,
+      "step": 2240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.812949435093971e-06,
+      "loss": 0.0238,
+      "step": 2241
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.804763983513503e-06,
+      "loss": 0.0205,
+      "step": 2242
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7965852628009204e-06,
+      "loss": 0.0187,
+      "step": 2243
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.788413281842065e-06,
+      "loss": 0.0223,
+      "step": 2244
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.780248049515456e-06,
+      "loss": 0.0255,
+      "step": 2245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7720895746922812e-06,
+      "loss": 0.0242,
+      "step": 2246
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7639378662363823e-06,
+      "loss": 0.022,
+      "step": 2247
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7557929330042654e-06,
+      "loss": 0.0213,
+      "step": 2248
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.747654783845052e-06,
+      "loss": 0.0327,
+      "step": 2249
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.739523427600509e-06,
+      "loss": 0.0214,
+      "step": 2250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7313988731050156e-06,
+      "loss": 0.0314,
+      "step": 2251
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.723281129185574e-06,
+      "loss": 0.0208,
+      "step": 2252
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7151702046617677e-06,
+      "loss": 0.0173,
+      "step": 2253
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7070661083457838e-06,
+      "loss": 0.0166,
+      "step": 2254
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.698968849042386e-06,
+      "loss": 0.0224,
+      "step": 2255
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.690878435548917e-06,
+      "loss": 0.0222,
+      "step": 2256
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.682794876655278e-06,
+      "loss": 0.0233,
+      "step": 2257
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6747181811439146e-06,
+      "loss": 0.0215,
+      "step": 2258
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6666483577898215e-06,
+      "loss": 0.0278,
+      "step": 2259
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.658585415360537e-06,
+      "loss": 0.0237,
+      "step": 2260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.650529362616113e-06,
+      "loss": 0.0218,
+      "step": 2261
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6424802083091124e-06,
+      "loss": 0.0259,
+      "step": 2262
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.634437961184608e-06,
+      "loss": 0.0161,
+      "step": 2263
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6264026299801758e-06,
+      "loss": 0.0274,
+      "step": 2264
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.618374223425868e-06,
+      "loss": 0.0251,
+      "step": 2265
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6103527502442183e-06,
+      "loss": 0.0323,
+      "step": 2266
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6023382191502175e-06,
+      "loss": 0.0248,
+      "step": 2267
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5943306388513333e-06,
+      "loss": 0.025,
+      "step": 2268
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.586330018047465e-06,
+      "loss": 0.0273,
+      "step": 2269
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.57833636543096e-06,
+      "loss": 0.0159,
+      "step": 2270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.570349689686583e-06,
+      "loss": 0.0208,
+      "step": 2271
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5623699994915363e-06,
+      "loss": 0.0271,
+      "step": 2272
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.554397303515421e-06,
+      "loss": 0.0306,
+      "step": 2273
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5464316104202403e-06,
+      "loss": 0.0257,
+      "step": 2274
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5384729288603912e-06,
+      "loss": 0.0306,
+      "step": 2275
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5305212674826526e-06,
+      "loss": 0.0275,
+      "step": 2276
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.522576634926176e-06,
+      "loss": 0.0203,
+      "step": 2277
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.514639039822475e-06,
+      "loss": 0.0232,
+      "step": 2278
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.506708490795424e-06,
+      "loss": 0.0224,
+      "step": 2279
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.498784996461225e-06,
+      "loss": 0.0278,
+      "step": 2280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.490868565428439e-06,
+      "loss": 0.0252,
+      "step": 2281
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4829592062979358e-06,
+      "loss": 0.0231,
+      "step": 2282
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.475056927662912e-06,
+      "loss": 0.0244,
+      "step": 2283
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.467161738108855e-06,
+      "loss": 0.0131,
+      "step": 2284
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4592736462135734e-06,
+      "loss": 0.03,
+      "step": 2285
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.0192,
+      "step": 2286
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.443518789671949e-06,
+      "loss": 0.029,
+      "step": 2287
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4356520421426063e-06,
+      "loss": 0.0261,
+      "step": 2288
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.427792426506017e-06,
+      "loss": 0.0233,
+      "step": 2289
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4199399513013297e-06,
+      "loss": 0.0229,
+      "step": 2290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.412094625059934e-06,
+      "loss": 0.0286,
+      "step": 2291
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4042564563054525e-06,
+      "loss": 0.0224,
+      "step": 2292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3964254535537322e-06,
+      "loss": 0.0208,
+      "step": 2293
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.388601625312833e-06,
+      "loss": 0.0171,
+      "step": 2294
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3807849800830207e-06,
+      "loss": 0.0294,
+      "step": 2295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3729755263567577e-06,
+      "loss": 0.0158,
+      "step": 2296
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3651732726186924e-06,
+      "loss": 0.0182,
+      "step": 2297
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3573782273456523e-06,
+      "loss": 0.0206,
+      "step": 2298
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3495903990066315e-06,
+      "loss": 0.0199,
+      "step": 2299
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.341809796062783e-06,
+      "loss": 0.024,
+      "step": 2300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.33403642696741e-06,
+      "loss": 0.0185,
+      "step": 2301
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3262703001659567e-06,
+      "loss": 0.0154,
+      "step": 2302
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.318511424096e-06,
+      "loss": 0.0234,
+      "step": 2303
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.310759807187238e-06,
+      "loss": 0.0168,
+      "step": 2304
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3030154578614783e-06,
+      "loss": 0.0171,
+      "step": 2305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2952783845326465e-06,
+      "loss": 0.0175,
+      "step": 2306
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.287548595606742e-06,
+      "loss": 0.0239,
+      "step": 2307
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2798260994818663e-06,
+      "loss": 0.0226,
+      "step": 2308
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2721109045481867e-06,
+      "loss": 0.0235,
+      "step": 2309
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2644030191879563e-06,
+      "loss": 0.0242,
+      "step": 2310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.256702451775461e-06,
+      "loss": 0.0203,
+      "step": 2311
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2490092106770545e-06,
+      "loss": 0.0205,
+      "step": 2312
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2413233042511203e-06,
+      "loss": 0.0258,
+      "step": 2313
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2336447408480874e-06,
+      "loss": 0.0233,
+      "step": 2314
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2259735288103956e-06,
+      "loss": 0.0187,
+      "step": 2315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.218309676472492e-06,
+      "loss": 0.0289,
+      "step": 2316
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.210653192160842e-06,
+      "loss": 0.0254,
+      "step": 2317
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2030040841938927e-06,
+      "loss": 0.0187,
+      "step": 2318
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.195362360882096e-06,
+      "loss": 0.0192,
+      "step": 2319
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1877280305278568e-06,
+      "loss": 0.0242,
+      "step": 2320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1801011014255645e-06,
+      "loss": 0.0213,
+      "step": 2321
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1724815818615584e-06,
+      "loss": 0.021,
+      "step": 2322
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.164869480114138e-06,
+      "loss": 0.0224,
+      "step": 2323
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.157264804453537e-06,
+      "loss": 0.0253,
+      "step": 2324
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1496675631419147e-06,
+      "loss": 0.0267,
+      "step": 2325
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1420777644333613e-06,
+      "loss": 0.0221,
+      "step": 2326
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.134495416573884e-06,
+      "loss": 0.0301,
+      "step": 2327
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1269205278013903e-06,
+      "loss": 0.0189,
+      "step": 2328
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1193531063456785e-06,
+      "loss": 0.0184,
+      "step": 2329
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.111793160428439e-06,
+      "loss": 0.0185,
+      "step": 2330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1042406982632467e-06,
+      "loss": 0.0265,
+      "step": 2331
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0966957280555367e-06,
+      "loss": 0.0248,
+      "step": 2332
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.08915825800261e-06,
+      "loss": 0.019,
+      "step": 2333
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0816282962936083e-06,
+      "loss": 0.0222,
+      "step": 2334
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.074105851109531e-06,
+      "loss": 0.0208,
+      "step": 2335
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0665909306232023e-06,
+      "loss": 0.0193,
+      "step": 2336
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0590835429992747e-06,
+      "loss": 0.0194,
+      "step": 2337
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0515836963942056e-06,
+      "loss": 0.0137,
+      "step": 2338
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.044091398956279e-06,
+      "loss": 0.0229,
+      "step": 2339
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0366066588255616e-06,
+      "loss": 0.0169,
+      "step": 2340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.029129484133915e-06,
+      "loss": 0.0257,
+      "step": 2341
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.02165988300498e-06,
+      "loss": 0.0272,
+      "step": 2342
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.014197863554171e-06,
+      "loss": 0.0273,
+      "step": 2343
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0067434338886623e-06,
+      "loss": 0.0219,
+      "step": 2344
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9992966021073834e-06,
+      "loss": 0.0236,
+      "step": 2345
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9918573763010106e-06,
+      "loss": 0.0201,
+      "step": 2346
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.984425764551954e-06,
+      "loss": 0.021,
+      "step": 2347
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9770017749343526e-06,
+      "loss": 0.0321,
+      "step": 2348
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9695854155140648e-06,
+      "loss": 0.0252,
+      "step": 2349
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9621766943486597e-06,
+      "loss": 0.0198,
+      "step": 2350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9547756194874046e-06,
+      "loss": 0.0216,
+      "step": 2351
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9473821989712626e-06,
+      "loss": 0.0195,
+      "step": 2352
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9399964408328795e-06,
+      "loss": 0.0261,
+      "step": 2353
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.932618353096576e-06,
+      "loss": 0.0202,
+      "step": 2354
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.925247943778339e-06,
+      "loss": 0.0286,
+      "step": 2355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9178852208858146e-06,
+      "loss": 0.0247,
+      "step": 2356
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9105301924182972e-06,
+      "loss": 0.0309,
+      "step": 2357
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9031828663667194e-06,
+      "loss": 0.0142,
+      "step": 2358
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8958432507136492e-06,
+      "loss": 0.0215,
+      "step": 2359
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8885113534332742e-06,
+      "loss": 0.0187,
+      "step": 2360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8811871824913993e-06,
+      "loss": 0.0215,
+      "step": 2361
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.873870745845433e-06,
+      "loss": 0.0232,
+      "step": 2362
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8665620514443805e-06,
+      "loss": 0.0125,
+      "step": 2363
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8592611072288356e-06,
+      "loss": 0.0273,
+      "step": 2364
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8519679211309746e-06,
+      "loss": 0.0207,
+      "step": 2365
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8446825010745427e-06,
+      "loss": 0.0239,
+      "step": 2366
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8374048549748457e-06,
+      "loss": 0.0194,
+      "step": 2367
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8301349907387464e-06,
+      "loss": 0.0225,
+      "step": 2368
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8228729162646518e-06,
+      "loss": 0.0306,
+      "step": 2369
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8156186394425054e-06,
+      "loss": 0.0269,
+      "step": 2370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.80837216815378e-06,
+      "loss": 0.0212,
+      "step": 2371
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.801133510271463e-06,
+      "loss": 0.0241,
+      "step": 2372
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7939026736600673e-06,
+      "loss": 0.0201,
+      "step": 2373
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7866796661755866e-06,
+      "loss": 0.0257,
+      "step": 2374
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7794644956655272e-06,
+      "loss": 0.0175,
+      "step": 2375
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.772257169968866e-06,
+      "loss": 0.0246,
+      "step": 2376
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.765057696916077e-06,
+      "loss": 0.0265,
+      "step": 2377
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7578660843290796e-06,
+      "loss": 0.0221,
+      "step": 2378
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7506823400212657e-06,
+      "loss": 0.0221,
+      "step": 2379
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.743506471797476e-06,
+      "loss": 0.021,
+      "step": 2380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.736338487453999e-06,
+      "loss": 0.0185,
+      "step": 2381
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7291783947785544e-06,
+      "loss": 0.0189,
+      "step": 2382
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7220262015502796e-06,
+      "loss": 0.0278,
+      "step": 2383
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.714881915539739e-06,
+      "loss": 0.0185,
+      "step": 2384
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7077455445089075e-06,
+      "loss": 0.0272,
+      "step": 2385
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.70061709621116e-06,
+      "loss": 0.022,
+      "step": 2386
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6934965783912524e-06,
+      "loss": 0.0195,
+      "step": 2387
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.686383998785336e-06,
+      "loss": 0.0224,
+      "step": 2388
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.679279365120937e-06,
+      "loss": 0.0196,
+      "step": 2389
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6721826851169462e-06,
+      "loss": 0.0216,
+      "step": 2390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.665093966483615e-06,
+      "loss": 0.0234,
+      "step": 2391
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6580132169225335e-06,
+      "loss": 0.0181,
+      "step": 2392
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.650940444126654e-06,
+      "loss": 0.0239,
+      "step": 2393
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.643875655780248e-06,
+      "loss": 0.0189,
+      "step": 2394
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6368188595589183e-06,
+      "loss": 0.0218,
+      "step": 2395
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6297700631295776e-06,
+      "loss": 0.0264,
+      "step": 2396
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6227292741504517e-06,
+      "loss": 0.0234,
+      "step": 2397
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.615696500271072e-06,
+      "loss": 0.0206,
+      "step": 2398
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6086717491322545e-06,
+      "loss": 0.0268,
+      "step": 2399
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6016550283661037e-06,
+      "loss": 0.021,
+      "step": 2400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5946463455959868e-06,
+      "loss": 0.0222,
+      "step": 2401
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5876457084365593e-06,
+      "loss": 0.0212,
+      "step": 2402
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5806531244937193e-06,
+      "loss": 0.0159,
+      "step": 2403
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5736686013646226e-06,
+      "loss": 0.0201,
+      "step": 2404
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5666921466376583e-06,
+      "loss": 0.0259,
+      "step": 2405
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5597237678924637e-06,
+      "loss": 0.019,
+      "step": 2406
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5527634726998906e-06,
+      "loss": 0.0233,
+      "step": 2407
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5458112686220136e-06,
+      "loss": 0.0162,
+      "step": 2408
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5388671632121144e-06,
+      "loss": 0.0191,
+      "step": 2409
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.531931164014676e-06,
+      "loss": 0.0213,
+      "step": 2410
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5250032785653744e-06,
+      "loss": 0.0195,
+      "step": 2411
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.518083514391073e-06,
+      "loss": 0.018,
+      "step": 2412
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.511171879009808e-06,
+      "loss": 0.0234,
+      "step": 2413
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5042683799307864e-06,
+      "loss": 0.0236,
+      "step": 2414
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4973730246543736e-06,
+      "loss": 0.0284,
+      "step": 2415
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4904858206720892e-06,
+      "loss": 0.0203,
+      "step": 2416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.483606775466595e-06,
+      "loss": 0.0196,
+      "step": 2417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4767358965116893e-06,
+      "loss": 0.0268,
+      "step": 2418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4698731912722983e-06,
+      "loss": 0.0183,
+      "step": 2419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4630186672044687e-06,
+      "loss": 0.021,
+      "step": 2420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4561723317553556e-06,
+      "loss": 0.0242,
+      "step": 2421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4493341923632195e-06,
+      "loss": 0.017,
+      "step": 2422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.0178,
+      "step": 2423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.435682531458393e-06,
+      "loss": 0.0209,
+      "step": 2424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.428869024777668e-06,
+      "loss": 0.028,
+      "step": 2425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.422063743817832e-06,
+      "loss": 0.0183,
+      "step": 2426
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4152666959725513e-06,
+      "loss": 0.0187,
+      "step": 2427
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.408477888626527e-06,
+      "loss": 0.0188,
+      "step": 2428
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4016973291555235e-06,
+      "loss": 0.0144,
+      "step": 2429
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3949250249263333e-06,
+      "loss": 0.0175,
+      "step": 2430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3881609832967934e-06,
+      "loss": 0.0239,
+      "step": 2431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3814052116157495e-06,
+      "loss": 0.032,
+      "step": 2432
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.374657717223068e-06,
+      "loss": 0.0143,
+      "step": 2433
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3679185074496247e-06,
+      "loss": 0.0177,
+      "step": 2434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.361187589617292e-06,
+      "loss": 0.0199,
+      "step": 2435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.354464971038933e-06,
+      "loss": 0.0141,
+      "step": 2436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.347750659018397e-06,
+      "loss": 0.0195,
+      "step": 2437
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3410446608505044e-06,
+      "loss": 0.0217,
+      "step": 2438
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3343469838210444e-06,
+      "loss": 0.0236,
+      "step": 2439
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3276576352067725e-06,
+      "loss": 0.0253,
+      "step": 2440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3209766222753814e-06,
+      "loss": 0.0199,
+      "step": 2441
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3143039522855204e-06,
+      "loss": 0.0245,
+      "step": 2442
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3076396324867643e-06,
+      "loss": 0.0262,
+      "step": 2443
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.300983670119632e-06,
+      "loss": 0.0273,
+      "step": 2444
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2943360724155406e-06,
+      "loss": 0.0168,
+      "step": 2445
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2876968465968354e-06,
+      "loss": 0.0236,
+      "step": 2446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.281065999876756e-06,
+      "loss": 0.0175,
+      "step": 2447
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.27444353945945e-06,
+      "loss": 0.0237,
+      "step": 2448
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2678294725399465e-06,
+      "loss": 0.0174,
+      "step": 2449
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2612238063041515e-06,
+      "loss": 0.0156,
+      "step": 2450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2546265479288464e-06,
+      "loss": 0.0189,
+      "step": 2451
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2480377045816858e-06,
+      "loss": 0.0186,
+      "step": 2452
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.241457283421178e-06,
+      "loss": 0.0218,
+      "step": 2453
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.23488529159667e-06,
+      "loss": 0.0206,
+      "step": 2454
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.22832173624836e-06,
+      "loss": 0.0211,
+      "step": 2455
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2217666245072864e-06,
+      "loss": 0.0202,
+      "step": 2456
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2152199634953043e-06,
+      "loss": 0.0241,
+      "step": 2457
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.208681760325092e-06,
+      "loss": 0.0237,
+      "step": 2458
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2021520221001304e-06,
+      "loss": 0.0193,
+      "step": 2459
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.195630755914716e-06,
+      "loss": 0.0248,
+      "step": 2460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.189117968853933e-06,
+      "loss": 0.0248,
+      "step": 2461
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1826136679936605e-06,
+      "loss": 0.0161,
+      "step": 2462
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1761178604005418e-06,
+      "loss": 0.0272,
+      "step": 2463
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1696305531320115e-06,
+      "loss": 0.0246,
+      "step": 2464
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.16315175323626e-06,
+      "loss": 0.0246,
+      "step": 2465
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.156681467752234e-06,
+      "loss": 0.0153,
+      "step": 2466
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1502197037096317e-06,
+      "loss": 0.0256,
+      "step": 2467
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1437664681288926e-06,
+      "loss": 0.015,
+      "step": 2468
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1373217680211924e-06,
+      "loss": 0.0237,
+      "step": 2469
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.130885610388428e-06,
+      "loss": 0.0236,
+      "step": 2470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.12445800222322e-06,
+      "loss": 0.0178,
+      "step": 2471
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1180389505089007e-06,
+      "loss": 0.0191,
+      "step": 2472
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1116284622195018e-06,
+      "loss": 0.0231,
+      "step": 2473
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.105226544319756e-06,
+      "loss": 0.0247,
+      "step": 2474
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0988332037650816e-06,
+      "loss": 0.0238,
+      "step": 2475
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0924484475015783e-06,
+      "loss": 0.02,
+      "step": 2476
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0860722824660217e-06,
+      "loss": 0.019,
+      "step": 2477
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0797047155858495e-06,
+      "loss": 0.0273,
+      "step": 2478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0733457537791614e-06,
+      "loss": 0.0266,
+      "step": 2479
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.066995403954706e-06,
+      "loss": 0.0205,
+      "step": 2480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0606536730118767e-06,
+      "loss": 0.0248,
+      "step": 2481
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0543205678407e-06,
+      "loss": 0.02,
+      "step": 2482
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.047996095321835e-06,
+      "loss": 0.0246,
+      "step": 2483
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0416802623265596e-06,
+      "loss": 0.0186,
+      "step": 2484
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0353730757167634e-06,
+      "loss": 0.0205,
+      "step": 2485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0290745423449465e-06,
+      "loss": 0.0175,
+      "step": 2486
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0227846690542043e-06,
+      "loss": 0.0162,
+      "step": 2487
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.016503462678222e-06,
+      "loss": 0.0201,
+      "step": 2488
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.010230930041274e-06,
+      "loss": 0.02,
+      "step": 2489
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0039670779582066e-06,
+      "loss": 0.0206,
+      "step": 2490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.997711913234435e-06,
+      "loss": 0.035,
+      "step": 2491
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9914654426659374e-06,
+      "loss": 0.0246,
+      "step": 2492
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.985227673039245e-06,
+      "loss": 0.018,
+      "step": 2493
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.978998611131443e-06,
+      "loss": 0.0229,
+      "step": 2494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9727782637101435e-06,
+      "loss": 0.0238,
+      "step": 2495
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.966566637533497e-06,
+      "loss": 0.0201,
+      "step": 2496
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.960363739350177e-06,
+      "loss": 0.0268,
+      "step": 2497
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.954169575899386e-06,
+      "loss": 0.0219,
+      "step": 2498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9479841539108156e-06,
+      "loss": 0.0199,
+      "step": 2499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.941807480104676e-06,
+      "loss": 0.0285,
+      "step": 2500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.935639561191663e-06,
+      "loss": 0.0202,
+      "step": 2501
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9294804038729743e-06,
+      "loss": 0.0239,
+      "step": 2502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9233300148402767e-06,
+      "loss": 0.0253,
+      "step": 2503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9171884007757093e-06,
+      "loss": 0.0284,
+      "step": 2504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.911055568351883e-06,
+      "loss": 0.0255,
+      "step": 2505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.90493152423187e-06,
+      "loss": 0.0139,
+      "step": 2506
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.898816275069193e-06,
+      "loss": 0.0209,
+      "step": 2507
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8927098275078116e-06,
+      "loss": 0.0215,
+      "step": 2508
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.886612188182132e-06,
+      "loss": 0.0261,
+      "step": 2509
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8805233637169829e-06,
+      "loss": 0.0225,
+      "step": 2510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8744433607276269e-06,
+      "loss": 0.0275,
+      "step": 2511
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8683721858197369e-06,
+      "loss": 0.0166,
+      "step": 2512
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8623098455893862e-06,
+      "loss": 0.0197,
+      "step": 2513
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8562563466230577e-06,
+      "loss": 0.0169,
+      "step": 2514
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8502116954976334e-06,
+      "loss": 0.0172,
+      "step": 2515
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.844175898780377e-06,
+      "loss": 0.0144,
+      "step": 2516
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8381489630289273e-06,
+      "loss": 0.0156,
+      "step": 2517
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8321308947912997e-06,
+      "loss": 0.0146,
+      "step": 2518
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8261217006058828e-06,
+      "loss": 0.0238,
+      "step": 2519
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8201213870014168e-06,
+      "loss": 0.0188,
+      "step": 2520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8141299604969908e-06,
+      "loss": 0.0166,
+      "step": 2521
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8081474276020406e-06,
+      "loss": 0.0243,
+      "step": 2522
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8021737948163465e-06,
+      "loss": 0.0231,
+      "step": 2523
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7962090686300116e-06,
+      "loss": 0.022,
+      "step": 2524
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7902532555234653e-06,
+      "loss": 0.0195,
+      "step": 2525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7843063619674473e-06,
+      "loss": 0.0156,
+      "step": 2526
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.778368394423018e-06,
+      "loss": 0.025,
+      "step": 2527
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7724393593415302e-06,
+      "loss": 0.0141,
+      "step": 2528
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.76651926316464e-06,
+      "loss": 0.0156,
+      "step": 2529
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.760608112324278e-06,
+      "loss": 0.02,
+      "step": 2530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7547059132426735e-06,
+      "loss": 0.0208,
+      "step": 2531
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7488126723323185e-06,
+      "loss": 0.0192,
+      "step": 2532
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7429283959959776e-06,
+      "loss": 0.0226,
+      "step": 2533
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7370530906266724e-06,
+      "loss": 0.0197,
+      "step": 2534
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7311867626076795e-06,
+      "loss": 0.0225,
+      "step": 2535
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7253294183125223e-06,
+      "loss": 0.0183,
+      "step": 2536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.719481064104962e-06,
+      "loss": 0.029,
+      "step": 2537
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7136417063389964e-06,
+      "loss": 0.015,
+      "step": 2538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7078113513588434e-06,
+      "loss": 0.0196,
+      "step": 2539
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7019900054989446e-06,
+      "loss": 0.0238,
+      "step": 2540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6961776750839509e-06,
+      "loss": 0.0284,
+      "step": 2541
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6903743664287197e-06,
+      "loss": 0.0204,
+      "step": 2542
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6845800858383055e-06,
+      "loss": 0.0145,
+      "step": 2543
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6787948396079544e-06,
+      "loss": 0.024,
+      "step": 2544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6730186340230981e-06,
+      "loss": 0.0277,
+      "step": 2545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.667251475359344e-06,
+      "loss": 0.0156,
+      "step": 2546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.661493369882473e-06,
+      "loss": 0.0167,
+      "step": 2547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6557443238484283e-06,
+      "loss": 0.0167,
+      "step": 2548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6500043435033109e-06,
+      "loss": 0.0175,
+      "step": 2549
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6442734350833721e-06,
+      "loss": 0.017,
+      "step": 2550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6385516048150063e-06,
+      "loss": 0.0217,
+      "step": 2551
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.632838858914747e-06,
+      "loss": 0.0205,
+      "step": 2552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6271352035892552e-06,
+      "loss": 0.0253,
+      "step": 2553
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.621440645035318e-06,
+      "loss": 0.0182,
+      "step": 2554
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6157551894398371e-06,
+      "loss": 0.0174,
+      "step": 2555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6100788429798242e-06,
+      "loss": 0.0198,
+      "step": 2556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6044116118223973e-06,
+      "loss": 0.0212,
+      "step": 2557
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5987535021247668e-06,
+      "loss": 0.0201,
+      "step": 2558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5931045200342355e-06,
+      "loss": 0.0242,
+      "step": 2559
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.0191,
+      "step": 2560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.581833963214091e-06,
+      "loss": 0.0233,
+      "step": 2561
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5762124007294721e-06,
+      "loss": 0.0247,
+      "step": 2562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5705999903419267e-06,
+      "loss": 0.0229,
+      "step": 2563
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564996738149106e-06,
+      "loss": 0.0186,
+      "step": 2564
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5594026502387184e-06,
+      "loss": 0.024,
+      "step": 2565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.553817732688503e-06,
+      "loss": 0.0273,
+      "step": 2566
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5482419915662418e-06,
+      "loss": 0.0199,
+      "step": 2567
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5426754329297477e-06,
+      "loss": 0.022,
+      "step": 2568
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.537118062826859e-06,
+      "loss": 0.0144,
+      "step": 2569
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5315698872954299e-06,
+      "loss": 0.0214,
+      "step": 2570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5260309123633189e-06,
+      "loss": 0.02,
+      "step": 2571
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.520501144048393e-06,
+      "loss": 0.0228,
+      "step": 2572
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5149805883585221e-06,
+      "loss": 0.0177,
+      "step": 2573
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5094692512915621e-06,
+      "loss": 0.0215,
+      "step": 2574
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5039671388353472e-06,
+      "loss": 0.0207,
+      "step": 2575
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4984742569676957e-06,
+      "loss": 0.02,
+      "step": 2576
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4929906116564008e-06,
+      "loss": 0.0181,
+      "step": 2577
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4875162088592155e-06,
+      "loss": 0.0221,
+      "step": 2578
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4820510545238542e-06,
+      "loss": 0.021,
+      "step": 2579
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4765951545879732e-06,
+      "loss": 0.0177,
+      "step": 2580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4711485149791894e-06,
+      "loss": 0.0204,
+      "step": 2581
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4657111416150504e-06,
+      "loss": 0.0195,
+      "step": 2582
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4602830404030387e-06,
+      "loss": 0.0273,
+      "step": 2583
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4548642172405569e-06,
+      "loss": 0.0295,
+      "step": 2584
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4494546780149366e-06,
+      "loss": 0.0232,
+      "step": 2585
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4440544286034186e-06,
+      "loss": 0.0272,
+      "step": 2586
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4386634748731499e-06,
+      "loss": 0.0187,
+      "step": 2587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4332818226811807e-06,
+      "loss": 0.0238,
+      "step": 2588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4279094778744473e-06,
+      "loss": 0.0212,
+      "step": 2589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4225464462897876e-06,
+      "loss": 0.0198,
+      "step": 2590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4171927337539103e-06,
+      "loss": 0.0236,
+      "step": 2591
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4118483460834065e-06,
+      "loss": 0.0204,
+      "step": 2592
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4065132890847245e-06,
+      "loss": 0.0249,
+      "step": 2593
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4011875685541898e-06,
+      "loss": 0.0238,
+      "step": 2594
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3958711902779775e-06,
+      "loss": 0.0187,
+      "step": 2595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.390564160032113e-06,
+      "loss": 0.0214,
+      "step": 2596
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3852664835824592e-06,
+      "loss": 0.0159,
+      "step": 2597
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3799781666847301e-06,
+      "loss": 0.0246,
+      "step": 2598
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3746992150844595e-06,
+      "loss": 0.0167,
+      "step": 2599
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.369429634517012e-06,
+      "loss": 0.0271,
+      "step": 2600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3641694307075682e-06,
+      "loss": 0.0322,
+      "step": 2601
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3589186093711227e-06,
+      "loss": 0.0216,
+      "step": 2602
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3536771762124745e-06,
+      "loss": 0.0205,
+      "step": 2603
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3484451369262251e-06,
+      "loss": 0.0227,
+      "step": 2604
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.34322249719677e-06,
+      "loss": 0.0221,
+      "step": 2605
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3380092626982887e-06,
+      "loss": 0.0202,
+      "step": 2606
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3328054390947475e-06,
+      "loss": 0.0196,
+      "step": 2607
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3276110320398849e-06,
+      "loss": 0.0212,
+      "step": 2608
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3224260471772088e-06,
+      "loss": 0.0165,
+      "step": 2609
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.317250490139992e-06,
+      "loss": 0.0221,
+      "step": 2610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3120843665512617e-06,
+      "loss": 0.0237,
+      "step": 2611
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3069276820237997e-06,
+      "loss": 0.0213,
+      "step": 2612
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3017804421601298e-06,
+      "loss": 0.0234,
+      "step": 2613
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2966426525525144e-06,
+      "loss": 0.018,
+      "step": 2614
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2915143187829516e-06,
+      "loss": 0.013,
+      "step": 2615
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.286395446423162e-06,
+      "loss": 0.0228,
+      "step": 2616
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2812860410345894e-06,
+      "loss": 0.0201,
+      "step": 2617
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2761861081683914e-06,
+      "loss": 0.0264,
+      "step": 2618
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2710956533654384e-06,
+      "loss": 0.0165,
+      "step": 2619
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2660146821562936e-06,
+      "loss": 0.0167,
+      "step": 2620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2609432000612232e-06,
+      "loss": 0.024,
+      "step": 2621
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.255881212590183e-06,
+      "loss": 0.0187,
+      "step": 2622
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2508287252428175e-06,
+      "loss": 0.022,
+      "step": 2623
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.245785743508441e-06,
+      "loss": 0.0262,
+      "step": 2624
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2407522728660447e-06,
+      "loss": 0.0251,
+      "step": 2625
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2357283187842872e-06,
+      "loss": 0.022,
+      "step": 2626
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2307138867214863e-06,
+      "loss": 0.0183,
+      "step": 2627
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2257089821256196e-06,
+      "loss": 0.0272,
+      "step": 2628
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.220713610434303e-06,
+      "loss": 0.0198,
+      "step": 2629
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2157277770748044e-06,
+      "loss": 0.0228,
+      "step": 2630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2107514874640235e-06,
+      "loss": 0.0233,
+      "step": 2631
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2057847470084993e-06,
+      "loss": 0.0187,
+      "step": 2632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.200827561104383e-06,
+      "loss": 0.0164,
+      "step": 2633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.195879935137455e-06,
+      "loss": 0.0256,
+      "step": 2634
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1909418744831048e-06,
+      "loss": 0.023,
+      "step": 2635
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.186013384506335e-06,
+      "loss": 0.03,
+      "step": 2636
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1810944705617466e-06,
+      "loss": 0.0235,
+      "step": 2637
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1761851379935307e-06,
+      "loss": 0.0143,
+      "step": 2638
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.171285392135476e-06,
+      "loss": 0.0259,
+      "step": 2639
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1663952383109566e-06,
+      "loss": 0.0143,
+      "step": 2640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1615146818329237e-06,
+      "loss": 0.0135,
+      "step": 2641
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1566437280038956e-06,
+      "loss": 0.0144,
+      "step": 2642
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1517823821159624e-06,
+      "loss": 0.0234,
+      "step": 2643
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1469306494507792e-06,
+      "loss": 0.0176,
+      "step": 2644
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1420885352795519e-06,
+      "loss": 0.0198,
+      "step": 2645
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1372560448630377e-06,
+      "loss": 0.0227,
+      "step": 2646
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1324331834515335e-06,
+      "loss": 0.0217,
+      "step": 2647
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1276199562848845e-06,
+      "loss": 0.0285,
+      "step": 2648
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1228163685924599e-06,
+      "loss": 0.0181,
+      "step": 2649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1180224255931626e-06,
+      "loss": 0.0164,
+      "step": 2650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1132381324954067e-06,
+      "loss": 0.0187,
+      "step": 2651
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.108463494497135e-06,
+      "loss": 0.0134,
+      "step": 2652
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1036985167857938e-06,
+      "loss": 0.0264,
+      "step": 2653
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.098943204538332e-06,
+      "loss": 0.0278,
+      "step": 2654
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.094197562921202e-06,
+      "loss": 0.0198,
+      "step": 2655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0894615970903454e-06,
+      "loss": 0.0197,
+      "step": 2656
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0847353121911952e-06,
+      "loss": 0.0269,
+      "step": 2657
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0800187133586626e-06,
+      "loss": 0.0169,
+      "step": 2658
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0753118057171396e-06,
+      "loss": 0.0272,
+      "step": 2659
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0706145943804846e-06,
+      "loss": 0.0175,
+      "step": 2660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0659270844520253e-06,
+      "loss": 0.0161,
+      "step": 2661
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0612492810245466e-06,
+      "loss": 0.0187,
+      "step": 2662
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.056581189180288e-06,
+      "loss": 0.0168,
+      "step": 2663
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.05192281399094e-06,
+      "loss": 0.0204,
+      "step": 2664
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0472741605176339e-06,
+      "loss": 0.0203,
+      "step": 2665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0426352338109392e-06,
+      "loss": 0.0191,
+      "step": 2666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.038006038910858e-06,
+      "loss": 0.0152,
+      "step": 2667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0333865808468203e-06,
+      "loss": 0.0191,
+      "step": 2668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0287768646376762e-06,
+      "loss": 0.0155,
+      "step": 2669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0241768952916898e-06,
+      "loss": 0.0195,
+      "step": 2670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.01958667780654e-06,
+      "loss": 0.0262,
+      "step": 2671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0150062171693075e-06,
+      "loss": 0.018,
+      "step": 2672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.010435518356474e-06,
+      "loss": 0.0226,
+      "step": 2673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0058745863339136e-06,
+      "loss": 0.0114,
+      "step": 2674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0013234260568915e-06,
+      "loss": 0.0239,
+      "step": 2675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.967820424700548e-07,
+      "loss": 0.018,
+      "step": 2676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.922504405074295e-07,
+      "loss": 0.0184,
+      "step": 2677
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.877286250924145e-07,
+      "loss": 0.0146,
+      "step": 2678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.83216601137773e-07,
+      "loss": 0.0179,
+      "step": 2679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.787143735456351e-07,
+      "loss": 0.0141,
+      "step": 2680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.742219472074833e-07,
+      "loss": 0.0211,
+      "step": 2681
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.697393270041533e-07,
+      "loss": 0.0265,
+      "step": 2682
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.652665178058285e-07,
+      "loss": 0.0317,
+      "step": 2683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.608035244720293e-07,
+      "loss": 0.0185,
+      "step": 2684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.563503518516114e-07,
+      "loss": 0.0185,
+      "step": 2685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.519070047827705e-07,
+      "loss": 0.0249,
+      "step": 2686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.474734880930114e-07,
+      "loss": 0.0213,
+      "step": 2687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.430498065991712e-07,
+      "loss": 0.0276,
+      "step": 2688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.386359651073929e-07,
+      "loss": 0.0216,
+      "step": 2689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.342319684131396e-07,
+      "loss": 0.0182,
+      "step": 2690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.298378213011661e-07,
+      "loss": 0.0173,
+      "step": 2691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.254535285455335e-07,
+      "loss": 0.0191,
+      "step": 2692
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.210790949095916e-07,
+      "loss": 0.0189,
+      "step": 2693
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.167145251459875e-07,
+      "loss": 0.0292,
+      "step": 2694
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.123598239966447e-07,
+      "loss": 0.0255,
+      "step": 2695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.08014996192762e-07,
+      "loss": 0.0143,
+      "step": 2696
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.0276,
+      "step": 2697
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.993549794925538e-07,
+      "loss": 0.0226,
+      "step": 2698
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.95039800004982e-07,
+      "loss": 0.0193,
+      "step": 2699
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.907345126803602e-07,
+      "loss": 0.0265,
+      "step": 2700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.864391221962065e-07,
+      "loss": 0.0133,
+      "step": 2701
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.821536332192882e-07,
+      "loss": 0.0267,
+      "step": 2702
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.778780504056106e-07,
+      "loss": 0.0178,
+      "step": 2703
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.73612378400418e-07,
+      "loss": 0.0199,
+      "step": 2704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.693566218381854e-07,
+      "loss": 0.018,
+      "step": 2705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.651107853426144e-07,
+      "loss": 0.0232,
+      "step": 2706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.608748735266359e-07,
+      "loss": 0.0195,
+      "step": 2707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.56648890992392e-07,
+      "loss": 0.029,
+      "step": 2708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.524328423312356e-07,
+      "loss": 0.0194,
+      "step": 2709
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.482267321237281e-07,
+      "loss": 0.028,
+      "step": 2710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.440305649396385e-07,
+      "loss": 0.015,
+      "step": 2711
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.398443453379268e-07,
+      "loss": 0.0167,
+      "step": 2712
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.356680778667503e-07,
+      "loss": 0.0168,
+      "step": 2713
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.31501767063445e-07,
+      "loss": 0.0205,
+      "step": 2714
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.273454174545415e-07,
+      "loss": 0.0245,
+      "step": 2715
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.231990335557383e-07,
+      "loss": 0.0209,
+      "step": 2716
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.190626198719143e-07,
+      "loss": 0.0259,
+      "step": 2717
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.149361808971045e-07,
+      "loss": 0.0132,
+      "step": 2718
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.108197211145219e-07,
+      "loss": 0.0243,
+      "step": 2719
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.067132449965254e-07,
+      "loss": 0.0194,
+      "step": 2720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.02616757004634e-07,
+      "loss": 0.0217,
+      "step": 2721
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.985302615895119e-07,
+      "loss": 0.0251,
+      "step": 2722
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.944537631909666e-07,
+      "loss": 0.0171,
+      "step": 2723
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.903872662379453e-07,
+      "loss": 0.0246,
+      "step": 2724
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.863307751485294e-07,
+      "loss": 0.0288,
+      "step": 2725
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822842943299292e-07,
+      "loss": 0.0236,
+      "step": 2726
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.782478281784789e-07,
+      "loss": 0.0225,
+      "step": 2727
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742213810796306e-07,
+      "loss": 0.0221,
+      "step": 2728
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70204957407954e-07,
+      "loss": 0.0225,
+      "step": 2729
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661985615271283e-07,
+      "loss": 0.0171,
+      "step": 2730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.62202197789934e-07,
+      "loss": 0.0158,
+      "step": 2731
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.582158705382581e-07,
+      "loss": 0.0193,
+      "step": 2732
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.542395841030792e-07,
+      "loss": 0.0254,
+      "step": 2733
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.502733428044684e-07,
+      "loss": 0.0182,
+      "step": 2734
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.463171509515832e-07,
+      "loss": 0.0211,
+      "step": 2735
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.423710128426608e-07,
+      "loss": 0.0178,
+      "step": 2736
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.384349327650198e-07,
+      "loss": 0.0161,
+      "step": 2737
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.345089149950468e-07,
+      "loss": 0.0248,
+      "step": 2738
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.305929637981968e-07,
+      "loss": 0.0258,
+      "step": 2739
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.26687083428993e-07,
+      "loss": 0.0156,
+      "step": 2740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.227912781310099e-07,
+      "loss": 0.0156,
+      "step": 2741
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.189055521368793e-07,
+      "loss": 0.0221,
+      "step": 2742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.150299096682812e-07,
+      "loss": 0.0227,
+      "step": 2743
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.111643549359426e-07,
+      "loss": 0.0182,
+      "step": 2744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.073088921396287e-07,
+      "loss": 0.0209,
+      "step": 2745
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.034635254681388e-07,
+      "loss": 0.0244,
+      "step": 2746
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.996282590993064e-07,
+      "loss": 0.0192,
+      "step": 2747
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.958030971999885e-07,
+      "loss": 0.0183,
+      "step": 2748
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.919880439260685e-07,
+      "loss": 0.0227,
+      "step": 2749
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.881831034224417e-07,
+      "loss": 0.0228,
+      "step": 2750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.843882798230184e-07,
+      "loss": 0.014,
+      "step": 2751
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.806035772507169e-07,
+      "loss": 0.0179,
+      "step": 2752
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.768289998174649e-07,
+      "loss": 0.0237,
+      "step": 2753
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.730645516241785e-07,
+      "loss": 0.0165,
+      "step": 2754
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.693102367607795e-07,
+      "loss": 0.0251,
+      "step": 2755
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.655660593061719e-07,
+      "loss": 0.0256,
+      "step": 2756
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.618320233282549e-07,
+      "loss": 0.017,
+      "step": 2757
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.581081328839012e-07,
+      "loss": 0.0296,
+      "step": 2758
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.543943920189633e-07,
+      "loss": 0.0259,
+      "step": 2759
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.50690804768267e-07,
+      "loss": 0.0233,
+      "step": 2760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.46997375155608e-07,
+      "loss": 0.0206,
+      "step": 2761
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.433141071937476e-07,
+      "loss": 0.019,
+      "step": 2762
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.396410048843992e-07,
+      "loss": 0.0209,
+      "step": 2763
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.359780722182374e-07,
+      "loss": 0.0215,
+      "step": 2764
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.323253131748908e-07,
+      "loss": 0.0251,
+      "step": 2765
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.286827317229293e-07,
+      "loss": 0.0172,
+      "step": 2766
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.250503318198664e-07,
+      "loss": 0.0224,
+      "step": 2767
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.214281174121518e-07,
+      "loss": 0.019,
+      "step": 2768
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.178160924351772e-07,
+      "loss": 0.0124,
+      "step": 2769
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.142142608132551e-07,
+      "loss": 0.0232,
+      "step": 2770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.106226264596293e-07,
+      "loss": 0.0189,
+      "step": 2771
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.070411932764586e-07,
+      "loss": 0.0182,
+      "step": 2772
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.034699651548237e-07,
+      "loss": 0.0175,
+      "step": 2773
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.999089459747176e-07,
+      "loss": 0.0229,
+      "step": 2774
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.9635813960504e-07,
+      "loss": 0.0249,
+      "step": 2775
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.928175499035926e-07,
+      "loss": 0.015,
+      "step": 2776
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.89287180717083e-07,
+      "loss": 0.0143,
+      "step": 2777
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.857670358811096e-07,
+      "loss": 0.0172,
+      "step": 2778
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.822571192201654e-07,
+      "loss": 0.0179,
+      "step": 2779
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.787574345476299e-07,
+      "loss": 0.0196,
+      "step": 2780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.752679856657639e-07,
+      "loss": 0.0166,
+      "step": 2781
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.717887763657115e-07,
+      "loss": 0.0209,
+      "step": 2782
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.68319810427489e-07,
+      "loss": 0.0197,
+      "step": 2783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.64861091619987e-07,
+      "loss": 0.0222,
+      "step": 2784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.614126237009543e-07,
+      "loss": 0.0195,
+      "step": 2785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.579744104170137e-07,
+      "loss": 0.0177,
+      "step": 2786
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.545464555036418e-07,
+      "loss": 0.0241,
+      "step": 2787
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.511287626851681e-07,
+      "loss": 0.0237,
+      "step": 2788
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.477213356747746e-07,
+      "loss": 0.0184,
+      "step": 2789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.443241781744923e-07,
+      "loss": 0.0146,
+      "step": 2790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.4093729387519e-07,
+      "loss": 0.0227,
+      "step": 2791
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.375606864565786e-07,
+      "loss": 0.0217,
+      "step": 2792
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.341943595872012e-07,
+      "loss": 0.0168,
+      "step": 2793
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.308383169244346e-07,
+      "loss": 0.0234,
+      "step": 2794
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.274925621144799e-07,
+      "loss": 0.0248,
+      "step": 2795
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.241570987923616e-07,
+      "loss": 0.0199,
+      "step": 2796
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.208319305819209e-07,
+      "loss": 0.0164,
+      "step": 2797
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.17517061095818e-07,
+      "loss": 0.0311,
+      "step": 2798
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.142124939355209e-07,
+      "loss": 0.0206,
+      "step": 2799
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.109182326913053e-07,
+      "loss": 0.0272,
+      "step": 2800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.076342809422485e-07,
+      "loss": 0.0245,
+      "step": 2801
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.043606422562308e-07,
+      "loss": 0.0197,
+      "step": 2802
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.010973201899228e-07,
+      "loss": 0.0193,
+      "step": 2803
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.978443182887904e-07,
+      "loss": 0.0203,
+      "step": 2804
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.946016400870857e-07,
+      "loss": 0.0139,
+      "step": 2805
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.913692891078437e-07,
+      "loss": 0.0153,
+      "step": 2806
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.88147268862883e-07,
+      "loss": 0.0177,
+      "step": 2807
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.849355828527924e-07,
+      "loss": 0.0155,
+      "step": 2808
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.817342345669363e-07,
+      "loss": 0.021,
+      "step": 2809
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.785432274834468e-07,
+      "loss": 0.0163,
+      "step": 2810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7536256506922507e-07,
+      "loss": 0.016,
+      "step": 2811
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.721922507799248e-07,
+      "loss": 0.0256,
+      "step": 2812
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6903228805996406e-07,
+      "loss": 0.0148,
+      "step": 2813
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.658826803425087e-07,
+      "loss": 0.0181,
+      "step": 2814
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6274343104948163e-07,
+      "loss": 0.0188,
+      "step": 2815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.596145435915467e-07,
+      "loss": 0.0223,
+      "step": 2816
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.56496021368108e-07,
+      "loss": 0.0208,
+      "step": 2817
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.533878677673131e-07,
+      "loss": 0.016,
+      "step": 2818
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.502900861660398e-07,
+      "loss": 0.0212,
+      "step": 2819
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.472026799299045e-07,
+      "loss": 0.0221,
+      "step": 2820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.441256524132431e-07,
+      "loss": 0.0178,
+      "step": 2821
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.410590069591192e-07,
+      "loss": 0.029,
+      "step": 2822
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.380027468993153e-07,
+      "loss": 0.0174,
+      "step": 2823
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3495687555433544e-07,
+      "loss": 0.026,
+      "step": 2824
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3192139623339125e-07,
+      "loss": 0.0161,
+      "step": 2825
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2889631223440255e-07,
+      "loss": 0.0218,
+      "step": 2826
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2588162684400027e-07,
+      "loss": 0.0172,
+      "step": 2827
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.228773433375166e-07,
+      "loss": 0.0141,
+      "step": 2828
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.198834649789818e-07,
+      "loss": 0.0195,
+      "step": 2829
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1689999502111835e-07,
+      "loss": 0.0245,
+      "step": 2830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1392693670534223e-07,
+      "loss": 0.0243,
+      "step": 2831
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.10964293261763e-07,
+      "loss": 0.0198,
+      "step": 2832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0801206790916815e-07,
+      "loss": 0.0171,
+      "step": 2833
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.0248,
+      "step": 2834
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0213888429548877e-07,
+      "loss": 0.0176,
+      "step": 2835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9921793241537666e-07,
+      "loss": 0.0245,
+      "step": 2836
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9630741138818484e-07,
+      "loss": 0.0249,
+      "step": 2837
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9340732437607496e-07,
+      "loss": 0.022,
+      "step": 2838
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9051767452986777e-07,
+      "loss": 0.0128,
+      "step": 2839
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8763846498905187e-07,
+      "loss": 0.0225,
+      "step": 2840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8476969888176843e-07,
+      "loss": 0.0276,
+      "step": 2841
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8191137932481414e-07,
+      "loss": 0.0145,
+      "step": 2842
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.790635094236317e-07,
+      "loss": 0.02,
+      "step": 2843
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.762260922723182e-07,
+      "loss": 0.0202,
+      "step": 2844
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.733991309536078e-07,
+      "loss": 0.0177,
+      "step": 2845
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7058262853887805e-07,
+      "loss": 0.0266,
+      "step": 2846
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6777658808814234e-07,
+      "loss": 0.0186,
+      "step": 2847
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.649810126500475e-07,
+      "loss": 0.0253,
+      "step": 2848
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6219590526187063e-07,
+      "loss": 0.0204,
+      "step": 2849
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.594212689495147e-07,
+      "loss": 0.0187,
+      "step": 2850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.566571067275093e-07,
+      "loss": 0.0213,
+      "step": 2851
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5390342159900225e-07,
+      "loss": 0.0174,
+      "step": 2852
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.511602165557571e-07,
+      "loss": 0.0217,
+      "step": 2853
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4842749457815406e-07,
+      "loss": 0.028,
+      "step": 2854
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.457052586351817e-07,
+      "loss": 0.0159,
+      "step": 2855
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.429935116844374e-07,
+      "loss": 0.0192,
+      "step": 2856
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.402922566721223e-07,
+      "loss": 0.0215,
+      "step": 2857
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3760149653303877e-07,
+      "loss": 0.0209,
+      "step": 2858
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3492123419058497e-07,
+      "loss": 0.0234,
+      "step": 2859
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3225147255675716e-07,
+      "loss": 0.0178,
+      "step": 2860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.295922145321384e-07,
+      "loss": 0.0142,
+      "step": 2861
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.269434630059054e-07,
+      "loss": 0.0255,
+      "step": 2862
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.243052208558151e-07,
+      "loss": 0.019,
+      "step": 2863
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.216774909482101e-07,
+      "loss": 0.0188,
+      "step": 2864
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1906027613800903e-07,
+      "loss": 0.0219,
+      "step": 2865
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1645357926870957e-07,
+      "loss": 0.0158,
+      "step": 2866
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1385740317237733e-07,
+      "loss": 0.0152,
+      "step": 2867
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1127175066965276e-07,
+      "loss": 0.024,
+      "step": 2868
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.086966245697387e-07,
+      "loss": 0.0222,
+      "step": 2869
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.06132027670405e-07,
+      "loss": 0.0182,
+      "step": 2870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.035779627579782e-07,
+      "loss": 0.0164,
+      "step": 2871
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0103443260734557e-07,
+      "loss": 0.0218,
+      "step": 2872
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9850143998194635e-07,
+      "loss": 0.0126,
+      "step": 2873
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9597898763377595e-07,
+      "loss": 0.016,
+      "step": 2874
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9346707830336997e-07,
+      "loss": 0.0114,
+      "step": 2875
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9096571471981636e-07,
+      "loss": 0.0195,
+      "step": 2876
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8847489960074136e-07,
+      "loss": 0.0172,
+      "step": 2877
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8599463565231687e-07,
+      "loss": 0.013,
+      "step": 2878
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8352492556924295e-07,
+      "loss": 0.0216,
+      "step": 2879
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.810657720347587e-07,
+      "loss": 0.0175,
+      "step": 2880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.786171777206337e-07,
+      "loss": 0.0208,
+      "step": 2881
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7617914528716317e-07,
+      "loss": 0.0157,
+      "step": 2882
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.737516773831717e-07,
+      "loss": 0.0222,
+      "step": 2883
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.713347766459984e-07,
+      "loss": 0.0183,
+      "step": 2884
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.689284457015073e-07,
+      "loss": 0.0228,
+      "step": 2885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6653268716407944e-07,
+      "loss": 0.0171,
+      "step": 2886
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6414750363660814e-07,
+      "loss": 0.0178,
+      "step": 2887
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6177289771049274e-07,
+      "loss": 0.0155,
+      "step": 2888
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5940887196564625e-07,
+      "loss": 0.0167,
+      "step": 2889
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.570554289704863e-07,
+      "loss": 0.0175,
+      "step": 2890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5471257128193094e-07,
+      "loss": 0.0184,
+      "step": 2891
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5238030144539737e-07,
+      "loss": 0.0164,
+      "step": 2892
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.500586219947998e-07,
+      "loss": 0.0222,
+      "step": 2893
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4774753545254936e-07,
+      "loss": 0.0272,
+      "step": 2894
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4544704432954313e-07,
+      "loss": 0.0262,
+      "step": 2895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.431571511251729e-07,
+      "loss": 0.0206,
+      "step": 2896
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.408778583273108e-07,
+      "loss": 0.0209,
+      "step": 2897
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3860916841231265e-07,
+      "loss": 0.0222,
+      "step": 2898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3635108384502003e-07,
+      "loss": 0.0195,
+      "step": 2899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3410360707874724e-07,
+      "loss": 0.0178,
+      "step": 2900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3186674055528436e-07,
+      "loss": 0.0217,
+      "step": 2901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2964048670489293e-07,
+      "loss": 0.0173,
+      "step": 2902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2742484794630813e-07,
+      "loss": 0.0192,
+      "step": 2903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2521982668673003e-07,
+      "loss": 0.0155,
+      "step": 2904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2302542532182226e-07,
+      "loss": 0.0145,
+      "step": 2905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.208416462357088e-07,
+      "loss": 0.0142,
+      "step": 2906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.186684918009796e-07,
+      "loss": 0.0173,
+      "step": 2907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.165059643786749e-07,
+      "loss": 0.0183,
+      "step": 2908
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1435406631829192e-07,
+      "loss": 0.0199,
+      "step": 2909
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1221279995777833e-07,
+      "loss": 0.0285,
+      "step": 2910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1008216762353207e-07,
+      "loss": 0.0243,
+      "step": 2911
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0796217163039588e-07,
+      "loss": 0.0134,
+      "step": 2912
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0585281428166071e-07,
+      "loss": 0.025,
+      "step": 2913
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.037540978690533e-07,
+      "loss": 0.0184,
+      "step": 2914
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.016660246727431e-07,
+      "loss": 0.0229,
+      "step": 2915
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.995885969613365e-07,
+      "loss": 0.0256,
+      "step": 2916
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9752181699187133e-07,
+      "loss": 0.0173,
+      "step": 2917
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9546568700981927e-07,
+      "loss": 0.0199,
+      "step": 2918
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9342020924908e-07,
+      "loss": 0.0166,
+      "step": 2919
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9138538593198141e-07,
+      "loss": 0.0195,
+      "step": 2920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8936121926927508e-07,
+      "loss": 0.0214,
+      "step": 2921
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8734771146013297e-07,
+      "loss": 0.0182,
+      "step": 2922
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8534486469214962e-07,
+      "loss": 0.0214,
+      "step": 2923
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8335268114133443e-07,
+      "loss": 0.0233,
+      "step": 2924
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8137116297211266e-07,
+      "loss": 0.0245,
+      "step": 2925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7940031233732224e-07,
+      "loss": 0.0317,
+      "step": 2926
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7744013137821038e-07,
+      "loss": 0.0185,
+      "step": 2927
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7549062222443347e-07,
+      "loss": 0.0209,
+      "step": 2928
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7355178699405172e-07,
+      "loss": 0.0252,
+      "step": 2929
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7162362779353014e-07,
+      "loss": 0.0203,
+      "step": 2930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.69706146717733e-07,
+      "loss": 0.0145,
+      "step": 2931
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.677993458499272e-07,
+      "loss": 0.0145,
+      "step": 2932
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.659032272617689e-07,
+      "loss": 0.0167,
+      "step": 2933
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6401779301331467e-07,
+      "loss": 0.0226,
+      "step": 2934
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.621430451530115e-07,
+      "loss": 0.0176,
+      "step": 2935
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.602789857176945e-07,
+      "loss": 0.0146,
+      "step": 2936
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5842561673258705e-07,
+      "loss": 0.0326,
+      "step": 2937
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.565829402112995e-07,
+      "loss": 0.0211,
+      "step": 2938
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5475095815582152e-07,
+      "loss": 0.0159,
+      "step": 2939
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5292967255652657e-07,
+      "loss": 0.0251,
+      "step": 2940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5111908539216845e-07,
+      "loss": 0.0172,
+      "step": 2941
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4931919862987143e-07,
+      "loss": 0.0249,
+      "step": 2942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4753001422514125e-07,
+      "loss": 0.0188,
+      "step": 2943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4575153412185072e-07,
+      "loss": 0.0159,
+      "step": 2944
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4398376025224758e-07,
+      "loss": 0.0168,
+      "step": 2945
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4222669453694215e-07,
+      "loss": 0.017,
+      "step": 2946
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4048033888491518e-07,
+      "loss": 0.0222,
+      "step": 2947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3874469519350896e-07,
+      "loss": 0.021,
+      "step": 2948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3701976534842953e-07,
+      "loss": 0.0177,
+      "step": 2949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3530555122374335e-07,
+      "loss": 0.0183,
+      "step": 2950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.336020546818706e-07,
+      "loss": 0.0171,
+      "step": 2951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3190927757358974e-07,
+      "loss": 0.0211,
+      "step": 2952
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3022722173803516e-07,
+      "loss": 0.0121,
+      "step": 2953
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2855588900269057e-07,
+      "loss": 0.0232,
+      "step": 2954
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.26895281183389e-07,
+      "loss": 0.0294,
+      "step": 2955
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2524540008431173e-07,
+      "loss": 0.0263,
+      "step": 2956
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2360624749798822e-07,
+      "loss": 0.021,
+      "step": 2957
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2197782520528834e-07,
+      "loss": 0.0186,
+      "step": 2958
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2036013497542798e-07,
+      "loss": 0.0147,
+      "step": 2959
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.187531785659568e-07,
+      "loss": 0.0151,
+      "step": 2960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1715695772277047e-07,
+      "loss": 0.0197,
+      "step": 2961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1557147418009396e-07,
+      "loss": 0.0144,
+      "step": 2962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1399672966049269e-07,
+      "loss": 0.0244,
+      "step": 2963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1243272587485809e-07,
+      "loss": 0.0219,
+      "step": 2964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1087946452241871e-07,
+      "loss": 0.0149,
+      "step": 2965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0933694729072686e-07,
+      "loss": 0.0283,
+      "step": 2966
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0780517585566419e-07,
+      "loss": 0.0241,
+      "step": 2967
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0628415188143726e-07,
+      "loss": 0.0201,
+      "step": 2968
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0477387702057529e-07,
+      "loss": 0.0106,
+      "step": 2969
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0327435291393018e-07,
+      "loss": 0.0218,
+      "step": 2970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.0221,
+      "step": 2971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0030756346829151e-07,
+      "loss": 0.0225,
+      "step": 2972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.884030135259182e-08,
+      "loss": 0.0195,
+      "step": 2973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.738379643769446e-08,
+      "loss": 0.0167,
+      "step": 2974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.593805030602921e-08,
+      "loss": 0.0232,
+      "step": 2975
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.45030645283418e-08,
+      "loss": 0.0202,
+      "step": 2976
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.3078840663684e-08,
+      "loss": 0.0221,
+      "step": 2977
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.166538025941695e-08,
+      "loss": 0.0226,
+      "step": 2978
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.02626848512067e-08,
+      "loss": 0.027,
+      "step": 2979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.887075596302197e-08,
+      "loss": 0.0191,
+      "step": 2980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.748959510713861e-08,
+      "loss": 0.0201,
+      "step": 2981
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.611920378412963e-08,
+      "loss": 0.0248,
+      "step": 2982
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.475958348286962e-08,
+      "loss": 0.026,
+      "step": 2983
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.341073568052804e-08,
+      "loss": 0.0153,
+      "step": 2984
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.207266184257268e-08,
+      "loss": 0.022,
+      "step": 2985
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.074536342276618e-08,
+      "loss": 0.02,
+      "step": 2986
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.94288418631639e-08,
+      "loss": 0.0172,
+      "step": 2987
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.81230985941106e-08,
+      "loss": 0.0163,
+      "step": 2988
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.68281350342437e-08,
+      "loss": 0.0255,
+      "step": 2989
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.554395259048664e-08,
+      "loss": 0.0194,
+      "step": 2990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.427055265805006e-08,
+      "loss": 0.0179,
+      "step": 2991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.300793662043282e-08,
+      "loss": 0.0154,
+      "step": 2992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.175610584941317e-08,
+      "loss": 0.0142,
+      "step": 2993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.051506170505318e-08,
+      "loss": 0.0177,
+      "step": 2994
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.92848055356976e-08,
+      "loss": 0.0227,
+      "step": 2995
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.806533867796728e-08,
+      "loss": 0.0236,
+      "step": 2996
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.685666245676348e-08,
+      "loss": 0.0283,
+      "step": 2997
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.565877818526245e-08,
+      "loss": 0.0279,
+      "step": 2998
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.447168716491648e-08,
+      "loss": 0.0222,
+      "step": 2999
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.329539068545054e-08,
+      "loss": 0.0128,
+      "step": 3000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.212989002486236e-08,
+      "loss": 0.0153,
+      "step": 3001
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.097518644942013e-08,
+      "loss": 0.0214,
+      "step": 3002
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983128121366255e-08,
+      "loss": 0.0188,
+      "step": 3003
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.869817556039325e-08,
+      "loss": 0.022,
+      "step": 3004
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.757587072068638e-08,
+      "loss": 0.018,
+      "step": 3005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.646436791387766e-08,
+      "loss": 0.0201,
+      "step": 3006
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.536366834757223e-08,
+      "loss": 0.0184,
+      "step": 3007
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.427377321763239e-08,
+      "loss": 0.0141,
+      "step": 3008
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.319468370818537e-08,
+      "loss": 0.0193,
+      "step": 3009
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.212640099161559e-08,
+      "loss": 0.0213,
+      "step": 3010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.106892622857018e-08,
+      "loss": 0.021,
+      "step": 3011
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.002226056795123e-08,
+      "loss": 0.0161,
+      "step": 3012
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.898640514691577e-08,
+      "loss": 0.021,
+      "step": 3013
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.796136109088023e-08,
+      "loss": 0.0199,
+      "step": 3014
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.694712951351044e-08,
+      "loss": 0.0166,
+      "step": 3015
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.594371151673049e-08,
+      "loss": 0.0196,
+      "step": 3016
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4951108190709466e-08,
+      "loss": 0.0185,
+      "step": 3017
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.396932061387138e-08,
+      "loss": 0.0192,
+      "step": 3018
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.299834985288631e-08,
+      "loss": 0.0196,
+      "step": 3019
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.203819696267486e-08,
+      "loss": 0.0196,
+      "step": 3020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.10888629864048e-08,
+      "loss": 0.019,
+      "step": 3021
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.015034895548664e-08,
+      "loss": 0.0158,
+      "step": 3022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.922265588957586e-08,
+      "loss": 0.0219,
+      "step": 3023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8305784796576203e-08,
+      "loss": 0.0222,
+      "step": 3024
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.739973667262642e-08,
+      "loss": 0.0227,
+      "step": 3025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6504512502113506e-08,
+      "loss": 0.0178,
+      "step": 3026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.562011325765946e-08,
+      "loss": 0.0226,
+      "step": 3027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4746539900129e-08,
+      "loss": 0.0259,
+      "step": 3028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.38837933786218e-08,
+      "loss": 0.0149,
+      "step": 3029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3031874630478076e-08,
+      "loss": 0.0174,
+      "step": 3030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2190784581270786e-08,
+      "loss": 0.0194,
+      "step": 3031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1360524144810055e-08,
+      "loss": 0.0231,
+      "step": 3032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.054109422313989e-08,
+      "loss": 0.02,
+      "step": 3033
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9732495706535913e-08,
+      "loss": 0.0232,
+      "step": 3034
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8934729473508726e-08,
+      "loss": 0.0277,
+      "step": 3035
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.814779639079723e-08,
+      "loss": 0.0186,
+      "step": 3036
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.737169731337197e-08,
+      "loss": 0.0245,
+      "step": 3037
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6606433084432892e-08,
+      "loss": 0.0179,
+      "step": 3038
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.585200453540715e-08,
+      "loss": 0.0152,
+      "step": 3039
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5108412485951306e-08,
+      "loss": 0.0171,
+      "step": 3040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.437565774394579e-08,
+      "loss": 0.0284,
+      "step": 3041
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3653741105499338e-08,
+      "loss": 0.0205,
+      "step": 3042
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.294266335494566e-08,
+      "loss": 0.0229,
+      "step": 3043
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2242425264838997e-08,
+      "loss": 0.0222,
+      "step": 3044
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1553027595959676e-08,
+      "loss": 0.0222,
+      "step": 3045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0874471097311886e-08,
+      "loss": 0.0157,
+      "step": 3046
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.020675650611703e-08,
+      "loss": 0.0169,
+      "step": 3047
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9549884547819253e-08,
+      "loss": 0.0222,
+      "step": 3048
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.890385593608435e-08,
+      "loss": 0.0226,
+      "step": 3049
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.82686713727942e-08,
+      "loss": 0.0235,
+      "step": 3050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7644331548052342e-08,
+      "loss": 0.0179,
+      "step": 3051
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.703083714017617e-08,
+      "loss": 0.0208,
+      "step": 3052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6428188815703627e-08,
+      "loss": 0.0201,
+      "step": 3053
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.583638722938652e-08,
+      "loss": 0.0164,
+      "step": 3054
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5255433024192746e-08,
+      "loss": 0.0198,
+      "step": 3055
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4685326831305192e-08,
+      "loss": 0.0222,
+      "step": 3056
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4126069270120613e-08,
+      "loss": 0.0196,
+      "step": 3057
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3577660948249637e-08,
+      "loss": 0.0151,
+      "step": 3058
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3040102461514548e-08,
+      "loss": 0.0223,
+      "step": 3059
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2513394393950384e-08,
+      "loss": 0.0151,
+      "step": 3060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1997537317804953e-08,
+      "loss": 0.0208,
+      "step": 3061
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1492531793534378e-08,
+      "loss": 0.0248,
+      "step": 3062
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0998378369806439e-08,
+      "loss": 0.0175,
+      "step": 3063
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0515077583498346e-08,
+      "loss": 0.026,
+      "step": 3064
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0042629959696737e-08,
+      "loss": 0.0219,
+      "step": 3065
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.581036011696577e-09,
+      "loss": 0.0223,
+      "step": 3066
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.130296241001146e-09,
+      "loss": 0.0104,
+      "step": 3067
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.690411137318722e-09,
+      "loss": 0.0199,
+      "step": 3068
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.261381178568117e-09,
+      "loss": 0.0187,
+      "step": 3069
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.84320683087203e-09,
+      "loss": 0.0232,
+      "step": 3070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.435888548559256e-09,
+      "loss": 0.0189,
+      "step": 3071
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0394267741646925e-09,
+      "loss": 0.0143,
+      "step": 3072
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.653821938427119e-09,
+      "loss": 0.0179,
+      "step": 3073
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.279074460291412e-09,
+      "loss": 0.0216,
+      "step": 3074
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9151847469041125e-09,
+      "loss": 0.0202,
+      "step": 3075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.562153193616749e-09,
+      "loss": 0.0171,
+      "step": 3076
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2199801839836195e-09,
+      "loss": 0.0226,
+      "step": 3077
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.8886660897617955e-09,
+      "loss": 0.0189,
+      "step": 3078
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.568211270910005e-09,
+      "loss": 0.0196,
+      "step": 3079
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.25861607558864e-09,
+      "loss": 0.0257,
+      "step": 3080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.9598808401597466e-09,
+      "loss": 0.0201,
+      "step": 3081
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6720058891892564e-09,
+      "loss": 0.0257,
+      "step": 3082
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3949915354380968e-09,
+      "loss": 0.0128,
+      "step": 3083
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1288380798744078e-09,
+      "loss": 0.0136,
+      "step": 3084
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8735458116602164e-09,
+      "loss": 0.0188,
+      "step": 3085
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.629115008160321e-09,
+      "loss": 0.022,
+      "step": 3086
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.395545934940069e-09,
+      "loss": 0.0195,
+      "step": 3087
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1728388457620263e-09,
+      "loss": 0.0255,
+      "step": 3088
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9609939825881996e-09,
+      "loss": 0.0166,
+      "step": 3089
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7600115755789237e-09,
+      "loss": 0.0207,
+      "step": 3090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5698918430928634e-09,
+      "loss": 0.0173,
+      "step": 3091
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3906349916881223e-09,
+      "loss": 0.0148,
+      "step": 3092
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2222412161200237e-09,
+      "loss": 0.0234,
+      "step": 3093
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0647106993411094e-09,
+      "loss": 0.0211,
+      "step": 3094
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.180436125011405e-10,
+      "loss": 0.0205,
+      "step": 3095
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.822401149482073e-10,
+      "loss": 0.0211,
+      "step": 3096
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.573003542276191e-10,
+      "loss": 0.0168,
+      "step": 3097
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.432244660819041e-10,
+      "loss": 0.0218,
+      "step": 3098
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.4001257444747924e-10,
+      "loss": 0.0166,
+      "step": 3099
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4766479146242095e-10,
+      "loss": 0.0191,
+      "step": 3100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.66181217457584e-10,
+      "loss": 0.0238,
+      "step": 3101
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.955619409610421e-10,
+      "loss": 0.0166,
+      "step": 3102
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.358070386980881e-10,
+      "loss": 0.0218,
+      "step": 3103
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.691657559012356e-11,
+      "loss": 0.0148,
+      "step": 3104
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.88906047546589e-11,
+      "loss": 0.0232,
+      "step": 3105
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1729167505313288e-11,
+      "loss": 0.0176,
+      "step": 3106
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.432293351814721e-12,
+      "loss": 0.016,
+      "step": 3107
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0152,
+      "step": 3108
+    },
+    {
+      "epoch": 2.0,
+      "step": 3108,
+      "total_flos": 158152722481152.0,
+      "train_loss": 0.03794149063920729,
+      "train_runtime": 41293.2118,
+      "train_samples_per_second": 8.425,
+      "train_steps_per_second": 0.075
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3108,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 188888,
+  "total_flos": 158152722481152.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}