diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,483 +1,7197 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.970873786407767,
-  "global_step": 75,
+  "epoch": 2.9981167608286254,
+  "global_step": 1194,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.04,
-      "learning_rate": 9.999999999999999e-06,
-      "loss": 1.144,
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 12.2937,
       "step": 1
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9999999999999998e-05,
-      "loss": 1.3398,
+      "epoch": 0.01,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 12.1494,
       "step": 2
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3e-05,
-      "loss": 0.7985,
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 10.5337,
       "step": 3
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.998572332372787e-05,
-      "loss": 0.7269,
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 10.1005,
       "step": 4
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.9942920471376185e-05,
-      "loss": 0.6506,
+      "epoch": 0.01,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 9.3766,
       "step": 5
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.9871672920607158e-05,
-      "loss": 0.6182,
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 8.8333,
       "step": 6
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 2.977211629518312e-05,
-      "loss": 0.5439,
+      "epoch": 0.02,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 9.282,
       "step": 7
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 2.9644440106799003e-05,
-      "loss": 0.5743,
+      "epoch": 0.02,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 8.6448,
       "step": 8
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 2.9488887394336025e-05,
-      "loss": 0.5574,
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 8.5396,
       "step": 9
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 2.9305754261223403e-05,
-      "loss": 0.5959,
+      "epoch": 0.03,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 8.8695,
       "step": 10
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 2.9095389311788626e-05,
-      "loss": 0.5722,
+      "epoch": 0.03,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 8.7526,
       "step": 11
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.8858192987669303e-05,
-      "loss": 0.6372,
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 8.5884,
       "step": 12
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.8594616805549752e-05,
-      "loss": 0.5282,
+      "epoch": 0.03,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 8.6299,
       "step": 13
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 2.8305162497673325e-05,
-      "loss": 0.5699,
+      "epoch": 0.04,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 8.1304,
       "step": 14
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 2.7990381056766583e-05,
-      "loss": 0.5106,
+      "epoch": 0.04,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 8.361,
       "step": 15
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 2.7650871687193287e-05,
-      "loss": 0.4858,
+      "epoch": 0.04,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 8.3707,
       "step": 16
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 2.728728066433488e-05,
-      "loss": 0.5103,
+      "epoch": 0.04,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 8.4532,
       "step": 17
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 2.6900300104368527e-05,
-      "loss": 0.4594,
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 8.2189,
       "step": 18
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 2.649066664678467e-05,
-      "loss": 0.5136,
+      "epoch": 0.05,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 8.2484,
       "step": 19
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.605916005215186e-05,
-      "loss": 0.443,
+      "epoch": 0.05,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 8.5937,
       "step": 20
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.5606601717798212e-05,
-      "loss": 0.4905,
+      "epoch": 0.05,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 7.8933,
       "step": 21
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.5133853114234907e-05,
-      "loss": 0.5694,
+      "epoch": 0.06,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 7.9975,
       "step": 22
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.464181414529809e-05,
-      "loss": 0.4642,
+      "epoch": 0.06,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 7.631,
       "step": 23
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.413142143513081e-05,
-      "loss": 0.4738,
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 7.9762,
       "step": 24
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.360364654526569e-05,
-      "loss": 0.522,
+      "epoch": 0.06,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 7.7915,
       "step": 25
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 2.305949412520236e-05,
-      "loss": 0.7219,
+      "epoch": 0.07,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 7.7185,
       "step": 26
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 2.25e-05,
-      "loss": 0.4399,
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 6.9976,
       "step": 27
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 2.192622919852551e-05,
-      "loss": 0.4449,
+      "epoch": 0.07,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 7.2998,
       "step": 28
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 2.1339273926110493e-05,
-      "loss": 0.3547,
+      "epoch": 0.07,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 7.0075,
       "step": 29
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 2.074025148547635e-05,
-      "loss": 0.4001,
+      "epoch": 0.08,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 7.1437,
       "step": 30
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 2.0130302149885033e-05,
-      "loss": 0.386,
+      "epoch": 0.08,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 7.3111,
       "step": 31
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 1.9510586992564095e-05,
-      "loss": 0.4169,
+      "epoch": 0.08,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 7.0273,
       "step": 32
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 1.888228567653781e-05,
-      "loss": 0.3796,
+      "epoch": 0.08,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 6.7756,
       "step": 33
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 1.8246594209071543e-05,
-      "loss": 0.3544,
+      "epoch": 0.09,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 6.7993,
       "step": 34
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 1.760472266500396e-05,
-      "loss": 0.3437,
+      "epoch": 0.09,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 6.2466,
       "step": 35
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 1.6957892883300778e-05,
-      "loss": 0.3457,
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 6.3919,
       "step": 36
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 1.6307336141214877e-05,
-      "loss": 0.32,
+      "epoch": 0.09,
+      "learning_rate": 1.9999963199614804e-05,
+      "loss": 6.1181,
       "step": 37
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 1.5654290810480043e-05,
-      "loss": 0.342,
+      "epoch": 0.1,
+      "learning_rate": 1.999985279873006e-05,
+      "loss": 6.0641,
       "step": 38
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 1.5e-05,
-      "loss": 0.2976,
+      "epoch": 0.1,
+      "learning_rate": 1.999966879815833e-05,
+      "loss": 5.8343,
       "step": 39
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 1.4345709189519963e-05,
-      "loss": 0.2522,
+      "epoch": 0.1,
+      "learning_rate": 1.999941119925387e-05,
+      "loss": 5.7531,
       "step": 40
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.3692663858785127e-05,
-      "loss": 0.2336,
+      "epoch": 0.1,
+      "learning_rate": 1.9999080003912633e-05,
+      "loss": 5.7179,
       "step": 41
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.304210711669923e-05,
-      "loss": 0.2732,
+      "epoch": 0.11,
+      "learning_rate": 1.999867521457224e-05,
+      "loss": 5.4663,
       "step": 42
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.2395277334996045e-05,
-      "loss": 0.2377,
+      "epoch": 0.11,
+      "learning_rate": 1.9998196834211972e-05,
+      "loss": 5.3022,
       "step": 43
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 1.1753405790928458e-05,
-      "loss": 0.3129,
+      "epoch": 0.11,
+      "learning_rate": 1.9997644866352742e-05,
+      "loss": 5.1614,
       "step": 44
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 1.1117714323462188e-05,
-      "loss": 0.2584,
+      "epoch": 0.11,
+      "learning_rate": 1.9997019315057083e-05,
+      "loss": 4.9698,
       "step": 45
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 1.0489413007435905e-05,
-      "loss": 0.3123,
+      "epoch": 0.12,
+      "learning_rate": 1.9996320184929093e-05,
+      "loss": 4.9179,
       "step": 46
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 9.86969785011497e-06,
-      "loss": 0.2232,
+      "epoch": 0.12,
+      "learning_rate": 1.9995547481114427e-05,
+      "loss": 4.5208,
       "step": 47
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 9.259748514523654e-06,
-      "loss": 0.1568,
+      "epoch": 0.12,
+      "learning_rate": 1.9994701209300245e-05,
+      "loss": 4.2619,
       "step": 48
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 8.66072607388951e-06,
-      "loss": 0.2494,
+      "epoch": 0.12,
+      "learning_rate": 1.9993781375715172e-05,
+      "loss": 4.1518,
       "step": 49
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.073770801474494e-06,
-      "loss": 0.3024,
+      "epoch": 0.13,
+      "learning_rate": 1.9992787987129253e-05,
+      "loss": 3.9266,
       "step": 50
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 7.500000000000004e-06,
-      "loss": 0.551,
+      "epoch": 0.13,
+      "learning_rate": 1.9991721050853908e-05,
+      "loss": 3.7692,
       "step": 51
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 6.9405058747976406e-06,
-      "loss": 0.2052,
+      "epoch": 0.13,
+      "learning_rate": 1.9990580574741867e-05,
+      "loss": 3.5005,
       "step": 52
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 6.3963534547343126e-06,
-      "loss": 0.1905,
+      "epoch": 0.13,
+      "learning_rate": 1.998936656718712e-05,
+      "loss": 3.2996,
       "step": 53
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 5.86857856486919e-06,
-      "loss": 0.2084,
+      "epoch": 0.14,
+      "learning_rate": 1.9988079037124866e-05,
+      "loss": 3.1707,
       "step": 54
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 5.3581858547019095e-06,
-      "loss": 0.1803,
+      "epoch": 0.14,
+      "learning_rate": 1.9986717994031415e-05,
+      "loss": 2.989,
       "step": 55
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 4.8661468857650965e-06,
-      "loss": 0.2179,
+      "epoch": 0.14,
+      "learning_rate": 1.998528344792415e-05,
+      "loss": 2.7901,
       "step": 56
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 4.393398282201788e-06,
-      "loss": 0.2408,
+      "epoch": 0.14,
+      "learning_rate": 1.9983775409361447e-05,
+      "loss": 2.5576,
       "step": 57
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 3.940839947848141e-06,
-      "loss": 0.1756,
+      "epoch": 0.15,
+      "learning_rate": 1.9982193889442583e-05,
+      "loss": 2.3154,
       "step": 58
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 3.5093333532153316e-06,
-      "loss": 0.1596,
+      "epoch": 0.15,
+      "learning_rate": 1.9980538899807664e-05,
+      "loss": 2.0445,
       "step": 59
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 3.0996998956314744e-06,
-      "loss": 0.2365,
+      "epoch": 0.15,
+      "learning_rate": 1.9978810452637544e-05,
+      "loss": 1.9681,
       "step": 60
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 2.7127193356651214e-06,
-      "loss": 0.172,
+      "epoch": 0.15,
+      "learning_rate": 1.9977008560653724e-05,
+      "loss": 1.7775,
       "step": 61
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 2.3491283128067174e-06,
-      "loss": 0.1681,
+      "epoch": 0.16,
+      "learning_rate": 1.9975133237118276e-05,
+      "loss": 1.5088,
       "step": 62
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 2.0096189432334194e-06,
-      "loss": 0.21,
+      "epoch": 0.16,
+      "learning_rate": 1.9973184495833717e-05,
+      "loss": 1.383,
       "step": 63
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 1.6948375023266743e-06,
-      "loss": 0.2293,
+      "epoch": 0.16,
+      "learning_rate": 1.9971162351142935e-05,
+      "loss": 1.2045,
       "step": 64
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 1.4053831944502509e-06,
-      "loss": 0.1471,
+      "epoch": 0.16,
+      "learning_rate": 1.9969066817929073e-05,
+      "loss": 1.0182,
       "step": 65
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 1.1418070123306989e-06,
-      "loss": 0.2285,
+      "epoch": 0.17,
+      "learning_rate": 1.9966897911615417e-05,
+      "loss": 0.8853,
       "step": 66
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 9.046106882113753e-07,
-      "loss": 0.2122,
+      "epoch": 0.17,
+      "learning_rate": 1.996465564816528e-05,
+      "loss": 0.7538,
       "step": 67
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 6.942457387765977e-07,
-      "loss": 0.1508,
+      "epoch": 0.17,
+      "learning_rate": 1.9962340044081898e-05,
+      "loss": 0.6244,
       "step": 68
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 5.11112605663977e-07,
-      "loss": 0.2877,
+      "epoch": 0.17,
+      "learning_rate": 1.9959951116408295e-05,
+      "loss": 0.535,
       "step": 69
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 3.5555989320099957e-07,
-      "loss": 0.2275,
+      "epoch": 0.18,
+      "learning_rate": 1.9957488882727163e-05,
+      "loss": 0.4518,
       "step": 70
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 2.278837048168797e-07,
-      "loss": 0.2798,
+      "epoch": 0.18,
+      "learning_rate": 1.995495336116073e-05,
+      "loss": 0.3935,
       "step": 71
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 1.2832707939284428e-07,
-      "loss": 0.2241,
+      "epoch": 0.18,
+      "learning_rate": 1.995234457037063e-05,
+      "loss": 0.3193,
       "step": 72
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 5.7079528623816824e-08,
-      "loss": 0.1843,
+      "epoch": 0.18,
+      "learning_rate": 1.9949662529557764e-05,
+      "loss": 0.2764,
       "step": 73
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 1.4276676272133027e-08,
-      "loss": 0.1802,
+      "epoch": 0.19,
+      "learning_rate": 1.994690725846216e-05,
+      "loss": 0.2414,
       "step": 74
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 0.0,
-      "loss": 0.2303,
+      "epoch": 0.19,
+      "learning_rate": 1.9944078777362828e-05,
+      "loss": 0.2045,
       "step": 75
     },
     {
-      "epoch": 2.97,
-      "step": 75,
-      "total_flos": 2.4061184768475136e+16,
-      "train_loss": 0.3887578751643499,
-      "train_runtime": 5064.5824,
-      "train_samples_per_second": 0.97,
-      "train_steps_per_second": 0.015
+      "epoch": 0.19,
+      "learning_rate": 1.9941177107077604e-05,
+      "loss": 0.1661,
+      "step": 76
     },
     {
-      "epoch": 2.97,
-      "eval_loss": 0.23003457486629486,
-      "eval_runtime": 84.5662,
-      "eval_samples_per_second": 4.848,
-      "eval_steps_per_second": 0.307,
-      "step": 75
+      "epoch": 0.19,
+      "learning_rate": 1.9938202268963004e-05,
+      "loss": 0.1549,
+      "step": 77
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935154284914063e-05,
+      "loss": 0.1554,
+      "step": 78
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932033177364188e-05,
+      "loss": 0.1156,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928838969284964e-05,
+      "loss": 0.1274,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925571684186006e-05,
+      "loss": 0.1184,
+      "step": 81
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922231346114795e-05,
+      "loss": 0.1126,
+      "step": 82
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918817979656463e-05,
+      "loss": 0.1115,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991533160993366e-05,
+      "loss": 0.0911,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991177226260633e-05,
+      "loss": 0.095,
+      "step": 85
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908139963871547e-05,
+      "loss": 0.0917,
+      "step": 86
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990443474046331e-05,
+      "loss": 0.102,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900656619652343e-05,
+      "loss": 0.0959,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896805629245912e-05,
+      "loss": 0.1227,
+      "step": 89
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98928817975876e-05,
+      "loss": 0.0926,
+      "step": 90
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888885153557113e-05,
+      "loss": 0.1048,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988481572657006e-05,
+      "loss": 0.1281,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988067354657773e-05,
+      "loss": 0.1017,
+      "step": 93
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876458644066896e-05,
+      "loss": 0.0957,
+      "step": 94
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987217105005956e-05,
+      "loss": 0.1043,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867810796112742e-05,
+      "loss": 0.1007,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986337791431825e-05,
+      "loss": 0.0971,
+      "step": 97
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985887243730244e-05,
+      "loss": 0.098,
+      "step": 98
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985429439822596e-05,
+      "loss": 0.0889,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984964383078354e-05,
+      "loss": 0.1067,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984492076920371e-05,
+      "loss": 0.0899,
+      "step": 101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840125248248564e-05,
+      "loss": 0.099,
+      "step": 102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835257303213514e-05,
+      "loss": 0.0842,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830316969927002e-05,
+      "loss": 0.0998,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825304284750263e-05,
+      "loss": 0.0912,
+      "step": 105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820219284577052e-05,
+      "loss": 0.1089,
+      "step": 106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981506200683336e-05,
+      "loss": 0.1064,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809832489477144e-05,
+      "loss": 0.0899,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804530770998058e-05,
+      "loss": 0.1117,
+      "step": 109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799156890417156e-05,
+      "loss": 0.1238,
+      "step": 110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793710887286613e-05,
+      "loss": 0.1035,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978819280168944e-05,
+      "loss": 0.1012,
+      "step": 112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782602674239163e-05,
+      "loss": 0.1096,
+      "step": 113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776940546079552e-05,
+      "loss": 0.1037,
+      "step": 114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771206458884312e-05,
+      "loss": 0.1072,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976540045485676e-05,
+      "loss": 0.1055,
+      "step": 116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759522576729534e-05,
+      "loss": 0.0943,
+      "step": 117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975357286776427e-05,
+      "loss": 0.101,
+      "step": 118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747551371751285e-05,
+      "loss": 0.0973,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741458133009258e-05,
+      "loss": 0.0874,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973529319638489e-05,
+      "loss": 0.0772,
+      "step": 121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972905660725259e-05,
+      "loss": 0.0917,
+      "step": 122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722748411514137e-05,
+      "loss": 0.087,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716368655598333e-05,
+      "loss": 0.1037,
+      "step": 124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970991738646068e-05,
+      "loss": 0.1026,
+      "step": 125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970339465158301e-05,
+      "loss": 0.0848,
+      "step": 126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696800498973158e-05,
+      "loss": 0.1123,
+      "step": 127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690134977164594e-05,
+      "loss": 0.0898,
+      "step": 128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683398135216067e-05,
+      "loss": 0.1015,
+      "step": 129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967659002271126e-05,
+      "loss": 0.076,
+      "step": 130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669710689758404e-05,
+      "loss": 0.0883,
+      "step": 131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662760186989914e-05,
+      "loss": 0.0894,
+      "step": 132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655738565562035e-05,
+      "loss": 0.0891,
+      "step": 133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648645877154435e-05,
+      "loss": 0.0849,
+      "step": 134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964148217396985e-05,
+      "loss": 0.1031,
+      "step": 135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634247508733685e-05,
+      "loss": 0.0844,
+      "step": 136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626941934693636e-05,
+      "loss": 0.1028,
+      "step": 137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619565505619288e-05,
+      "loss": 0.085,
+      "step": 138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961211827580173e-05,
+      "loss": 0.0751,
+      "step": 139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604600300053146e-05,
+      "loss": 0.0809,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597011633706414e-05,
+      "loss": 0.0932,
+      "step": 141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589352332614708e-05,
+      "loss": 0.0808,
+      "step": 142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581622453151072e-05,
+      "loss": 0.0849,
+      "step": 143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573822052208013e-05,
+      "loss": 0.0892,
+      "step": 144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565951187197085e-05,
+      "loss": 0.084,
+      "step": 145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955800991604846e-05,
+      "loss": 0.0765,
+      "step": 146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95499982972105e-05,
+      "loss": 0.0883,
+      "step": 147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541916389649346e-05,
+      "loss": 0.0796,
+      "step": 148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953376425284846e-05,
+      "loss": 0.0805,
+      "step": 149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525541946808187e-05,
+      "loss": 0.0781,
+      "step": 150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517249532045346e-05,
+      "loss": 0.0907,
+      "step": 151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508887069592735e-05,
+      "loss": 0.1014,
+      "step": 152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950045462099873e-05,
+      "loss": 0.1071,
+      "step": 153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491952248326805e-05,
+      "loss": 0.0979,
+      "step": 154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948338001415507e-05,
+      "loss": 0.0985,
+      "step": 155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474737981575833e-05,
+      "loss": 0.085,
+      "step": 156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946602621419512e-05,
+      "loss": 0.1161,
+      "step": 157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457244776132208e-05,
+      "loss": 0.0746,
+      "step": 158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944839373201916e-05,
+      "loss": 0.0847,
+      "step": 159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439473147000344e-05,
+      "loss": 0.0821,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430483086731945e-05,
+      "loss": 0.0919,
+      "step": 161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942142361738151e-05,
+      "loss": 0.0894,
+      "step": 162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412294805627423e-05,
+      "loss": 0.0911,
+      "step": 163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403096718658446e-05,
+      "loss": 0.087,
+      "step": 164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393829424173205e-05,
+      "loss": 0.0864,
+      "step": 165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384492990379703e-05,
+      "loss": 0.0756,
+      "step": 166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937508748599481e-05,
+      "loss": 0.0869,
+      "step": 167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936561298024377e-05,
+      "loss": 0.0739,
+      "step": 168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356069542859666e-05,
+      "loss": 0.0811,
+      "step": 169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934645724408294e-05,
+      "loss": 0.088,
+      "step": 170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336776154660842e-05,
+      "loss": 0.1019,
+      "step": 171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327026345846945e-05,
+      "loss": 0.0789,
+      "step": 172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317207889400595e-05,
+      "loss": 0.0873,
+      "step": 173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307320857586377e-05,
+      "loss": 0.0745,
+      "step": 174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929736532317361e-05,
+      "loss": 0.0825,
+      "step": 175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92873413594358e-05,
+      "loss": 0.0727,
+      "step": 176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277249040150093e-05,
+      "loss": 0.0645,
+      "step": 177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267088439596728e-05,
+      "loss": 0.0845,
+      "step": 178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256859632558513e-05,
+      "loss": 0.0784,
+      "step": 179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246562694320258e-05,
+      "loss": 0.0869,
+      "step": 180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236197700668214e-05,
+      "loss": 0.0835,
+      "step": 181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225764727889543e-05,
+      "loss": 0.0779,
+      "step": 182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921526385277172e-05,
+      "loss": 0.1,
+      "step": 183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204695152601997e-05,
+      "loss": 0.0743,
+      "step": 184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194058705166826e-05,
+      "loss": 0.0599,
+      "step": 185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183354588751274e-05,
+      "loss": 0.0752,
+      "step": 186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172582882138466e-05,
+      "loss": 0.0857,
+      "step": 187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916174366460899e-05,
+      "loss": 0.0792,
+      "step": 188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915083701594032e-05,
+      "loss": 0.091,
+      "step": 189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139863016406237e-05,
+      "loss": 0.0745,
+      "step": 190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912882174677622e-05,
+      "loss": 0.088,
+      "step": 191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117713288314864e-05,
+      "loss": 0.0584,
+      "step": 192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106537722781276e-05,
+      "loss": 0.091,
+      "step": 193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095295132428485e-05,
+      "loss": 0.0684,
+      "step": 194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908398560000282e-05,
+      "loss": 0.0805,
+      "step": 195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072609208743307e-05,
+      "loss": 0.08,
+      "step": 196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061166042381063e-05,
+      "loss": 0.0875,
+      "step": 197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904965618513868e-05,
+      "loss": 0.0875,
+      "step": 198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903807972172959e-05,
+      "loss": 0.0635,
+      "step": 199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026436737357454e-05,
+      "loss": 0.0794,
+      "step": 200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014727317715536e-05,
+      "loss": 0.0854,
+      "step": 201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900295154898607e-05,
+      "loss": 0.0805,
+      "step": 202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991109517839613e-05,
+      "loss": 0.0772,
+      "step": 203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979201311434434e-05,
+      "loss": 0.0716,
+      "step": 204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967227017415845e-05,
+      "loss": 0.0735,
+      "step": 205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955186723915573e-05,
+      "loss": 0.0657,
+      "step": 206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943080519551108e-05,
+      "loss": 0.0795,
+      "step": 207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930908493425048e-05,
+      "loss": 0.072,
+      "step": 208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891867073512444e-05,
+      "loss": 0.0856,
+      "step": 209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906367334720125e-05,
+      "loss": 0.0758,
+      "step": 210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889399838276608e-05,
+      "loss": 0.0869,
+      "step": 211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888156397029875e-05,
+      "loss": 0.0875,
+      "step": 212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886906418883636e-05,
+      "loss": 0.0795,
+      "step": 213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885649913037827e-05,
+      "loss": 0.09,
+      "step": 214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843868887404282e-05,
+      "loss": 0.0963,
+      "step": 215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831173552873946e-05,
+      "loss": 0.0783,
+      "step": 216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881841322022591e-05,
+      "loss": 0.0787,
+      "step": 217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805587983377208e-05,
+      "loss": 0.081,
+      "step": 218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792697936722565e-05,
+      "loss": 0.0709,
+      "step": 219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779743175133718e-05,
+      "loss": 0.065,
+      "step": 220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876672379395871e-05,
+      "loss": 0.0723,
+      "step": 221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753639889021197e-05,
+      "loss": 0.0728,
+      "step": 222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874049155661972e-05,
+      "loss": 0.0708,
+      "step": 223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727278893527015e-05,
+      "loss": 0.0799,
+      "step": 224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714001996989312e-05,
+      "loss": 0.0738,
+      "step": 225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700660964725583e-05,
+      "loss": 0.0659,
+      "step": 226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687255894926853e-05,
+      "loss": 0.0796,
+      "step": 227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673786886255478e-05,
+      "loss": 0.0894,
+      "step": 228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.0681,
+      "step": 229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646657449296394e-05,
+      "loss": 0.0829,
+      "step": 230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863299722068344e-05,
+      "loss": 0.0925,
+      "step": 231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619273452545857e-05,
+      "loss": 0.0714,
+      "step": 232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605486245891633e-05,
+      "loss": 0.0697,
+      "step": 233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591635702195672e-05,
+      "loss": 0.0682,
+      "step": 234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577721923399047e-05,
+      "loss": 0.0703,
+      "step": 235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563745011908243e-05,
+      "loss": 0.0583,
+      "step": 236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549705070594396e-05,
+      "loss": 0.0639,
+      "step": 237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8535602202792567e-05,
+      "loss": 0.0618,
+      "step": 238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852143651230094e-05,
+      "loss": 0.1002,
+      "step": 239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507208103380093e-05,
+      "loss": 0.0817,
+      "step": 240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492917080752208e-05,
+      "loss": 0.0661,
+      "step": 241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478563549600318e-05,
+      "loss": 0.0733,
+      "step": 242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464147615567517e-05,
+      "loss": 0.0739,
+      "step": 243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844966938475619e-05,
+      "loss": 0.0897,
+      "step": 244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435128963727225e-05,
+      "loss": 0.0838,
+      "step": 245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420526459499252e-05,
+      "loss": 0.0703,
+      "step": 246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405861979547816e-05,
+      "loss": 0.0867,
+      "step": 247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391135631804626e-05,
+      "loss": 0.0591,
+      "step": 248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376347524656735e-05,
+      "loss": 0.0641,
+      "step": 249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361497766945747e-05,
+      "loss": 0.0672,
+      "step": 250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346586467967028e-05,
+      "loss": 0.0775,
+      "step": 251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331613737468888e-05,
+      "loss": 0.0884,
+      "step": 252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831657968565177e-05,
+      "loss": 0.055,
+      "step": 253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301484423167456e-05,
+      "loss": 0.0733,
+      "step": 254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8286328061118243e-05,
+      "loss": 0.0747,
+      "step": 255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271110711056125e-05,
+      "loss": 0.0559,
+      "step": 256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255832484981968e-05,
+      "loss": 0.0819,
+      "step": 257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240493495344695e-05,
+      "loss": 0.0657,
+      "step": 258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225093855040448e-05,
+      "loss": 0.0658,
+      "step": 259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209633677411767e-05,
+      "loss": 0.0716,
+      "step": 260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8194113076246753e-05,
+      "loss": 0.073,
+      "step": 261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178532165778225e-05,
+      "loss": 0.073,
+      "step": 262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816289106068288e-05,
+      "loss": 0.0589,
+      "step": 263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147189876080463e-05,
+      "loss": 0.0569,
+      "step": 264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131428727532903e-05,
+      "loss": 0.0811,
+      "step": 265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811560773104346e-05,
+      "loss": 0.073,
+      "step": 266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099727003055894e-05,
+      "loss": 0.0743,
+      "step": 267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808378666045358e-05,
+      "loss": 0.0796,
+      "step": 268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067786820558673e-05,
+      "loss": 0.06,
+      "step": 269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8051727601131228e-05,
+      "loss": 0.0628,
+      "step": 270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035609120368334e-05,
+      "loss": 0.0624,
+      "step": 271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801943149690325e-05,
+      "loss": 0.0558,
+      "step": 272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800319484980453e-05,
+      "loss": 0.0883,
+      "step": 273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798689929857516e-05,
+      "loss": 0.0675,
+      "step": 274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970544963151637e-05,
+      "loss": 0.0772,
+      "step": 275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954131963903134e-05,
+      "loss": 0.0806,
+      "step": 276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937660421630595e-05,
+      "loss": 0.0622,
+      "step": 277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921130457565835e-05,
+      "loss": 0.0745,
+      "step": 278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904542193370665e-05,
+      "loss": 0.0481,
+      "step": 279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887895751135985e-05,
+      "loss": 0.0633,
+      "step": 280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78711912533809e-05,
+      "loss": 0.0711,
+      "step": 281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785442882305179e-05,
+      "loss": 0.076,
+      "step": 282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783760858352144e-05,
+      "loss": 0.0812,
+      "step": 283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7820730658588106e-05,
+      "loss": 0.0556,
+      "step": 284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7803795172474618e-05,
+      "loss": 0.0624,
+      "step": 285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786802249827454e-05,
+      "loss": 0.0587,
+      "step": 286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769752015715842e-05,
+      "loss": 0.0626,
+      "step": 287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775264459563081e-05,
+      "loss": 0.0692,
+      "step": 288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735480115484294e-05,
+      "loss": 0.0564,
+      "step": 289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771825870160819e-05,
+      "loss": 0.0635,
+      "step": 290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770098048075342e-05,
+      "loss": 0.0688,
+      "step": 291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683645580089038e-05,
+      "loss": 0.0535,
+      "step": 292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666254127201236e-05,
+      "loss": 0.062,
+      "step": 293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764880625009245e-05,
+      "loss": 0.0663,
+      "step": 294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631302077180403e-05,
+      "loss": 0.0692,
+      "step": 295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761374173729715e-05,
+      "loss": 0.0734,
+      "step": 296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7596125359688154e-05,
+      "loss": 0.0693,
+      "step": 297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7578453074011302e-05,
+      "loss": 0.0622,
+      "step": 298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560725010335985e-05,
+      "loss": 0.0777,
+      "step": 299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542941299142113e-05,
+      "loss": 0.0617,
+      "step": 300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7525102071319176e-05,
+      "loss": 0.0745,
+      "step": 301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507207458165257e-05,
+      "loss": 0.0842,
+      "step": 302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489257591386092e-05,
+      "loss": 0.0709,
+      "step": 303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471252603094084e-05,
+      "loss": 0.0629,
+      "step": 304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453192625807334e-05,
+      "loss": 0.0662,
+      "step": 305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435077792448666e-05,
+      "loss": 0.0572,
+      "step": 306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7416908236344647e-05,
+      "loss": 0.079,
+      "step": 307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398684091224608e-05,
+      "loss": 0.0656,
+      "step": 308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738040549121967e-05,
+      "loss": 0.071,
+      "step": 309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736207257086173e-05,
+      "loss": 0.0574,
+      "step": 310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343685465082488e-05,
+      "loss": 0.0543,
+      "step": 311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325244309212476e-05,
+      "loss": 0.0689,
+      "step": 312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306749238980008e-05,
+      "loss": 0.0675,
+      "step": 313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288200390510227e-05,
+      "loss": 0.0737,
+      "step": 314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7269597900324096e-05,
+      "loss": 0.0677,
+      "step": 315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725094190533737e-05,
+      "loss": 0.0707,
+      "step": 316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232232542859606e-05,
+      "loss": 0.0564,
+      "step": 317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213469950593156e-05,
+      "loss": 0.0645,
+      "step": 318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194654266632146e-05,
+      "loss": 0.0646,
+      "step": 319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717578562946146e-05,
+      "loss": 0.0639,
+      "step": 320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715686417795572e-05,
+      "loss": 0.088,
+      "step": 321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137890051378264e-05,
+      "loss": 0.0558,
+      "step": 322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118863389380126e-05,
+      "loss": 0.0664,
+      "step": 323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709978433199901e-05,
+      "loss": 0.0526,
+      "step": 324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080653019658244e-05,
+      "loss": 0.0674,
+      "step": 325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706146959316576e-05,
+      "loss": 0.0604,
+      "step": 326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042234193713058e-05,
+      "loss": 0.0633,
+      "step": 327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022946962874157e-05,
+      "loss": 0.0785,
+      "step": 328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700360804260456e-05,
+      "loss": 0.0636,
+      "step": 329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6984217575240212e-05,
+      "loss": 0.0642,
+      "step": 330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964775703496447e-05,
+      "loss": 0.0466,
+      "step": 331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694528257046694e-05,
+      "loss": 0.0634,
+      "step": 332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692573831962265e-05,
+      "loss": 0.0691,
+      "step": 333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6906143094810774e-05,
+      "loss": 0.0535,
+      "step": 334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688649704025367e-05,
+      "loss": 0.0718,
+      "step": 335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6866800300547814e-05,
+      "loss": 0.0616,
+      "step": 336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847053020662726e-05,
+      "loss": 0.0739,
+      "step": 337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827255345939915e-05,
+      "loss": 0.0687,
+      "step": 338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807407422091785e-05,
+      "loss": 0.0722,
+      "step": 339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787509395200582e-05,
+      "loss": 0.0572,
+      "step": 340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6767561411717327e-05,
+      "loss": 0.0735,
+      "step": 341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674756361846071e-05,
+      "loss": 0.0648,
+      "step": 342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672751616261603e-05,
+      "loss": 0.0729,
+      "step": 343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6707419191734104e-05,
+      "loss": 0.0817,
+      "step": 344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668727285373019e-05,
+      "loss": 0.0675,
+      "step": 345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666707729688289e-05,
+      "loss": 0.0625,
+      "step": 346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646832669833047e-05,
+      "loss": 0.0762,
+      "step": 347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626539121582687e-05,
+      "loss": 0.0752,
+      "step": 348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660619680149388e-05,
+      "loss": 0.0607,
+      "step": 349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658580585928768e-05,
+      "loss": 0.088,
+      "step": 350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565366445042982e-05,
+      "loss": 0.0581,
+      "step": 351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544878709195457e-05,
+      "loss": 0.0619,
+      "step": 352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652434280253641e-05,
+      "loss": 0.0797,
+      "step": 353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650375887621171e-05,
+      "loss": 0.0634,
+      "step": 354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483127081720626e-05,
+      "loss": 0.0555,
+      "step": 355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646244757091476e-05,
+      "loss": 0.0462,
+      "step": 356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644172049599691e-05,
+      "loss": 0.0626,
+      "step": 357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642094600951994e-05,
+      "loss": 0.0635,
+      "step": 358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6400124264385673e-05,
+      "loss": 0.0779,
+      "step": 359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637925541384375e-05,
+      "loss": 0.0641,
+      "step": 360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635833961149053e-05,
+      "loss": 0.0527,
+      "step": 361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337377011267924e-05,
+      "loss": 0.0674,
+      "step": 362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631636776746228e-05,
+      "loss": 0.0604,
+      "step": 363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295312034703257e-05,
+      "loss": 0.0702,
+      "step": 364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274209967962668e-05,
+      "loss": 0.0637,
+      "step": 365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253061722553353e-05,
+      "loss": 0.0566,
+      "step": 366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6231867454128023e-05,
+      "loss": 0.0641,
+      "step": 367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621062731867813e-05,
+      "loss": 0.0637,
+      "step": 368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189341472532705e-05,
+      "loss": 0.0676,
+      "step": 369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168010072357216e-05,
+      "loss": 0.0467,
+      "step": 370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614663327515241e-05,
+      "loss": 0.0621,
+      "step": 371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612521123825317e-05,
+      "loss": 0.0533,
+      "step": 372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6103744119327325e-05,
+      "loss": 0.0582,
+      "step": 373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082232076374532e-05,
+      "loss": 0.0728,
+      "step": 374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6060675267725083e-05,
+      "loss": 0.0709,
+      "step": 375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039073852038746e-05,
+      "loss": 0.0562,
+      "step": 376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017427988303613e-05,
+      "loss": 0.0695,
+      "step": 377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5995737835834905e-05,
+      "loss": 0.0551,
+      "step": 378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974003554273815e-05,
+      "loss": 0.0496,
+      "step": 379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5952225303586323e-05,
+      "loss": 0.0724,
+      "step": 380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593040324406204e-05,
+      "loss": 0.0695,
+      "step": 381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590853753631301e-05,
+      "loss": 0.0459,
+      "step": 382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886628341272513e-05,
+      "loss": 0.0679,
+      "step": 383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586467582019392e-05,
+      "loss": 0.0798,
+      "step": 384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584268013464948e-05,
+      "loss": 0.0656,
+      "step": 385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5820641446529127e-05,
+      "loss": 0.0589,
+      "step": 386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579855991803931e-05,
+      "loss": 0.058,
+      "step": 387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577643571170177e-05,
+      "loss": 0.0672,
+      "step": 388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575426899035238e-05,
+      "loss": 0.0642,
+      "step": 389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732059917139912e-05,
+      "loss": 0.0786,
+      "step": 390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5709808655524858e-05,
+      "loss": 0.0598,
+      "step": 391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5687515369278216e-05,
+      "loss": 0.0465,
+      "step": 392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566518022248029e-05,
+      "loss": 0.0578,
+      "step": 393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564280337951948e-05,
+      "loss": 0.0628,
+      "step": 394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5620385005091077e-05,
+      "loss": 0.0584,
+      "step": 395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597925264196048e-05,
+      "loss": 0.0652,
+      "step": 396
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557542432213981e-05,
+      "loss": 0.068,
+      "step": 397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552882344531023e-05,
+      "loss": 0.0662,
+      "step": 398
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530299497280395e-05,
+      "loss": 0.0453,
+      "step": 399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507675946599413e-05,
+      "loss": 0.0509,
+      "step": 400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548501185899915e-05,
+      "loss": 0.0636,
+      "step": 401
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546230740128904e-05,
+      "loss": 0.0708,
+      "step": 402
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5439562740575644e-05,
+      "loss": 0.0496,
+      "step": 403
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5416778044261413e-05,
+      "loss": 0.0518,
+      "step": 404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393953480043468e-05,
+      "loss": 0.0663,
+      "step": 405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371089215912363e-05,
+      "loss": 0.062,
+      "step": 406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534818542015084e-05,
+      "loss": 0.062,
+      "step": 407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53252422613326e-05,
+      "loss": 0.0573,
+      "step": 408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530225990832106e-05,
+      "loss": 0.0696,
+      "step": 409
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279238530268112e-05,
+      "loss": 0.0677,
+      "step": 410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256178296612869e-05,
+      "loss": 0.0541,
+      "step": 411
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5233079377080424e-05,
+      "loss": 0.0531,
+      "step": 412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209941941680614e-05,
+      "loss": 0.0378,
+      "step": 413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186766160706738e-05,
+      "loss": 0.0646,
+      "step": 414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163552204734328e-05,
+      "loss": 0.041,
+      "step": 415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5140300244619894e-05,
+      "loss": 0.0528,
+      "step": 416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117010451499654e-05,
+      "loss": 0.0532,
+      "step": 417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093682996788274e-05,
+      "loss": 0.0679,
+      "step": 418
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070318052177624e-05,
+      "loss": 0.0491,
+      "step": 419
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.504691578963549e-05,
+      "loss": 0.051,
+      "step": 420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023476381404334e-05,
+      "loss": 0.0559,
+      "step": 421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0521,
+      "step": 422
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976486818210467e-05,
+      "loss": 0.0568,
+      "step": 423
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4952937009094567e-05,
+      "loss": 0.0484,
+      "step": 424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929350745980707e-05,
+      "loss": 0.0586,
+      "step": 425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905728202465596e-05,
+      "loss": 0.0566,
+      "step": 426
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882069552412982e-05,
+      "loss": 0.0569,
+      "step": 427
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4858374969952345e-05,
+      "loss": 0.0573,
+      "step": 428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4834644629477643e-05,
+      "loss": 0.0588,
+      "step": 429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810878705646005e-05,
+      "loss": 0.0647,
+      "step": 430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478707737337647e-05,
+      "loss": 0.0557,
+      "step": 431
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4763240807848667e-05,
+      "loss": 0.0524,
+      "step": 432
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739369184501557e-05,
+      "loss": 0.0421,
+      "step": 433
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715462679032134e-05,
+      "loss": 0.0718,
+      "step": 434
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691521467394109e-05,
+      "loss": 0.049,
+      "step": 435
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4667545725796655e-05,
+      "loss": 0.0509,
+      "step": 436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4643535630703067e-05,
+      "loss": 0.0512,
+      "step": 437
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619491358829502e-05,
+      "loss": 0.0552,
+      "step": 438
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4595413087143656e-05,
+      "loss": 0.0535,
+      "step": 439
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4571300992863453e-05,
+      "loss": 0.0601,
+      "step": 440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547155253455769e-05,
+      "loss": 0.058,
+      "step": 441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452297604663511e-05,
+      "loss": 0.0515,
+      "step": 442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.44987635503623e-05,
+      "loss": 0.0488,
+      "step": 443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4474517942843173e-05,
+      "loss": 0.0541,
+      "step": 444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450239402527271e-05,
+      "loss": 0.0584,
+      "step": 445
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425928108106519e-05,
+      "loss": 0.0532,
+      "step": 446
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401584238513921e-05,
+      "loss": 0.0541,
+      "step": 447
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4377207972922229e-05,
+      "loss": 0.0454,
+      "step": 448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435279949074264e-05,
+      "loss": 0.0469,
+      "step": 449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328358971623455e-05,
+      "loss": 0.0469,
+      "step": 450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4303886595448785e-05,
+      "loss": 0.0647,
+      "step": 451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4279382542337202e-05,
+      "loss": 0.0476,
+      "step": 452
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254846992640423e-05,
+      "loss": 0.0514,
+      "step": 453
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4230280126941987e-05,
+      "loss": 0.0467,
+      "step": 454
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205682126055915e-05,
+      "loss": 0.0602,
+      "step": 455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181053171025392e-05,
+      "loss": 0.0479,
+      "step": 456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156393443121424e-05,
+      "loss": 0.0404,
+      "step": 457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131703123841503e-05,
+      "loss": 0.0562,
+      "step": 458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4106982394908285e-05,
+      "loss": 0.0527,
+      "step": 459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408223143826824e-05,
+      "loss": 0.0722,
+      "step": 460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4057450436090316e-05,
+      "loss": 0.0509,
+      "step": 461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032639570764595e-05,
+      "loss": 0.0478,
+      "step": 462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007799024900962e-05,
+      "loss": 0.0549,
+      "step": 463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3982928981327742e-05,
+      "loss": 0.053,
+      "step": 464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958029623090379e-05,
+      "loss": 0.0522,
+      "step": 465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393310113345006e-05,
+      "loss": 0.0532,
+      "step": 466
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908143695882397e-05,
+      "loss": 0.0712,
+      "step": 467
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883157494076048e-05,
+      "loss": 0.0414,
+      "step": 468
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858142711931384e-05,
+      "loss": 0.0554,
+      "step": 469
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833099533559129e-05,
+      "loss": 0.0392,
+      "step": 470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3808028143279007e-05,
+      "loss": 0.057,
+      "step": 471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378292872561838e-05,
+      "loss": 0.0469,
+      "step": 472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757801465310895e-05,
+      "loss": 0.0511,
+      "step": 473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732646547295128e-05,
+      "loss": 0.0522,
+      "step": 474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707464156713208e-05,
+      "loss": 0.0554,
+      "step": 475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682254478909474e-05,
+      "loss": 0.0566,
+      "step": 476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657017699429092e-05,
+      "loss": 0.0619,
+      "step": 477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631754004016708e-05,
+      "loss": 0.0446,
+      "step": 478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3606463578615064e-05,
+      "loss": 0.0374,
+      "step": 479
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358114660936364e-05,
+      "loss": 0.037,
+      "step": 480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555803282597279e-05,
+      "loss": 0.0591,
+      "step": 481
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353043378484482e-05,
+      "loss": 0.0463,
+      "step": 482
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3505038302827724e-05,
+      "loss": 0.0521,
+      "step": 483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479617023458686e-05,
+      "loss": 0.0488,
+      "step": 484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454170133840291e-05,
+      "loss": 0.0386,
+      "step": 485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34286978212636e-05,
+      "loss": 0.0613,
+      "step": 486
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3403200273206798e-05,
+      "loss": 0.0464,
+      "step": 487
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377677677333804e-05,
+      "loss": 0.0524,
+      "step": 488
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335213022149289e-05,
+      "loss": 0.0632,
+      "step": 489
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326558093715294e-05,
+      "loss": 0.0598,
+      "step": 490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300961482213855e-05,
+      "loss": 0.0486,
+      "step": 491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32753405753816e-05,
+      "loss": 0.0467,
+      "step": 492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3249695561790378e-05,
+      "loss": 0.0455,
+      "step": 493
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3224026630189465e-05,
+      "loss": 0.0414,
+      "step": 494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198333969504176e-05,
+      "loss": 0.0458,
+      "step": 495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3172617768834472e-05,
+      "loss": 0.049,
+      "step": 496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146878217453572e-05,
+      "loss": 0.0529,
+      "step": 497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121115504806554e-05,
+      "loss": 0.0538,
+      "step": 498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095329820508974e-05,
+      "loss": 0.0581,
+      "step": 499
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069521354345449e-05,
+      "loss": 0.0461,
+      "step": 500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304369029626828e-05,
+      "loss": 0.0538,
+      "step": 501
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017836836396046e-05,
+      "loss": 0.0369,
+      "step": 502
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991961165012203e-05,
+      "loss": 0.0572,
+      "step": 503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966063472563686e-05,
+      "loss": 0.0425,
+      "step": 504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940143949659504e-05,
+      "loss": 0.051,
+      "step": 505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2914202787069345e-05,
+      "loss": 0.0509,
+      "step": 506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888240175722163e-05,
+      "loss": 0.0397,
+      "step": 507
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862256306704777e-05,
+      "loss": 0.0357,
+      "step": 508
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836251371260469e-05,
+      "loss": 0.049,
+      "step": 509
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810225560787561e-05,
+      "loss": 0.052,
+      "step": 510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278417906683803e-05,
+      "loss": 0.046,
+      "step": 511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758112081116071e-05,
+      "loss": 0.0564,
+      "step": 512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273202479547671e-05,
+      "loss": 0.034,
+      "step": 513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2705917401924382e-05,
+      "loss": 0.0534,
+      "step": 514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267979009261151e-05,
+      "loss": 0.038,
+      "step": 515
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653643059837109e-05,
+      "loss": 0.0383,
+      "step": 516
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2627476496045349e-05,
+      "loss": 0.0562,
+      "step": 517
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601290593824155e-05,
+      "loss": 0.0561,
+      "step": 518
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575085545903793e-05,
+      "loss": 0.0499,
+      "step": 519
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548861545155424e-05,
+      "loss": 0.0444,
+      "step": 520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522618784589724e-05,
+      "loss": 0.0574,
+      "step": 521
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496357457355423e-05,
+      "loss": 0.0395,
+      "step": 522
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470077756737919e-05,
+      "loss": 0.042,
+      "step": 523
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2443779876157831e-05,
+      "loss": 0.0678,
+      "step": 524
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417464009169585e-05,
+      "loss": 0.046,
+      "step": 525
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239113034945999e-05,
+      "loss": 0.0537,
+      "step": 526
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2364779090846811e-05,
+      "loss": 0.0632,
+      "step": 527
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338410427277342e-05,
+      "loss": 0.0511,
+      "step": 528
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2312024552826977e-05,
+      "loss": 0.0349,
+      "step": 529
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2285621661697787e-05,
+      "loss": 0.0552,
+      "step": 530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2259201948217076e-05,
+      "loss": 0.0366,
+      "step": 531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232765606835986e-05,
+      "loss": 0.0625,
+      "step": 532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206312832128013e-05,
+      "loss": 0.0459,
+      "step": 533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2179843818787625e-05,
+      "loss": 0.0396,
+      "step": 534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2153358761628793e-05,
+      "loss": 0.0611,
+      "step": 535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212685785558358e-05,
+      "loss": 0.0449,
+      "step": 536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100341295700702e-05,
+      "loss": 0.0561,
+      "step": 537
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207380927714407e-05,
+      "loss": 0.0533,
+      "step": 538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2047261995191397e-05,
+      "loss": 0.046,
+      "step": 539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202069964523272e-05,
+      "loss": 0.0479,
+      "step": 540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994122422768978e-05,
+      "loss": 0.0471,
+      "step": 541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967530523410578e-05,
+      "loss": 0.0555,
+      "step": 542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1940924142875947e-05,
+      "loss": 0.0443,
+      "step": 543
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1914303476990096e-05,
+      "loss": 0.0755,
+      "step": 544
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1887668721683176e-05,
+      "loss": 0.0427,
+      "step": 545
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186102007298904e-05,
+      "loss": 0.0598,
+      "step": 546
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834357727043795e-05,
+      "loss": 0.0351,
+      "step": 547
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807681880084358e-05,
+      "loss": 0.0516,
+      "step": 548
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780992728447018e-05,
+      "loss": 0.0404,
+      "step": 549
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754290468565995e-05,
+      "loss": 0.047,
+      "step": 550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172757529697197e-05,
+      "loss": 0.0502,
+      "step": 551
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700847410290667e-05,
+      "loss": 0.0562,
+      "step": 552
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167410700524139e-05,
+      "loss": 0.0505,
+      "step": 553
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1647354278635583e-05,
+      "loss": 0.0535,
+      "step": 554
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1620589427375375e-05,
+      "loss": 0.0482,
+      "step": 555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593812648452128e-05,
+      "loss": 0.0427,
+      "step": 556
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567024138945003e-05,
+      "loss": 0.0459,
+      "step": 557
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1540224096019495e-05,
+      "loss": 0.0407,
+      "step": 558
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513412716925978e-05,
+      "loss": 0.0422,
+      "step": 559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148659019899827e-05,
+      "loss": 0.0456,
+      "step": 560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459756739652175e-05,
+      "loss": 0.0527,
+      "step": 561
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432912536384013e-05,
+      "loss": 0.0375,
+      "step": 562
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406057786769194e-05,
+      "loss": 0.0371,
+      "step": 563
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137919268846074e-05,
+      "loss": 0.0437,
+      "step": 564
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135231743918785e-05,
+      "loss": 0.05,
+      "step": 565
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325432236754424e-05,
+      "loss": 0.0412,
+      "step": 566
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.129853727903762e-05,
+      "loss": 0.045,
+      "step": 567
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271632763986408e-05,
+      "loss": 0.0396,
+      "step": 568
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244718889620085e-05,
+      "loss": 0.0448,
+      "step": 569
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121779585402684e-05,
+      "loss": 0.0535,
+      "step": 570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1190863855362294e-05,
+      "loss": 0.0466,
+      "step": 571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1163923091848026e-05,
+      "loss": 0.0389,
+      "step": 572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136973761770136e-05,
+      "loss": 0.0323,
+      "step": 573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1110016063477763e-05,
+      "loss": 0.046,
+      "step": 574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1083050195381648e-05,
+      "loss": 0.0432,
+      "step": 575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105607635595266e-05,
+      "loss": 0.0545,
+      "step": 576
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102909474372033e-05,
+      "loss": 0.0463,
+      "step": 577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002105557271405e-05,
+      "loss": 0.0347,
+      "step": 578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975108995248378e-05,
+      "loss": 0.0395,
+      "step": 579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948105256348021e-05,
+      "loss": 0.036,
+      "step": 580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0921094539319936e-05,
+      "loss": 0.0346,
+      "step": 581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0894077042965084e-05,
+      "loss": 0.0425,
+      "step": 582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0867052966134314e-05,
+      "loss": 0.0414,
+      "step": 583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0840022507726915e-05,
+      "loss": 0.0471,
+      "step": 584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0812985866689143e-05,
+      "loss": 0.0411,
+      "step": 585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0785943242012763e-05,
+      "loss": 0.058,
+      "step": 586
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075889483273357e-05,
+      "loss": 0.0411,
+      "step": 587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0731840837929946e-05,
+      "loss": 0.059,
+      "step": 588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704781456721372e-05,
+      "loss": 0.0504,
+      "step": 589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677716888266979e-05,
+      "loss": 0.0462,
+      "step": 590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650647331764079e-05,
+      "loss": 0.0461,
+      "step": 591
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623572986446689e-05,
+      "loss": 0.0438,
+      "step": 592
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059649405158408e-05,
+      "loss": 0.0445,
+      "step": 593
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569410726479301e-05,
+      "loss": 0.0556,
+      "step": 594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0542323210467704e-05,
+      "loss": 0.0472,
+      "step": 595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0515231702915498e-05,
+      "loss": 0.0506,
+      "step": 596
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488136403218265e-05,
+      "loss": 0.0374,
+      "step": 597
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461037510799499e-05,
+      "loss": 0.0574,
+      "step": 598
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0433935225109135e-05,
+      "loss": 0.0505,
+      "step": 599
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406829745622085e-05,
+      "loss": 0.0499,
+      "step": 600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0379721271836765e-05,
+      "loss": 0.049,
+      "step": 601
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035261000327363e-05,
+      "loss": 0.0367,
+      "step": 602
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0325496139473702e-05,
+      "loss": 0.0467,
+      "step": 603
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0298379879997119e-05,
+      "loss": 0.0362,
+      "step": 604
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0271261424421628e-05,
+      "loss": 0.0376,
+      "step": 605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244140972341155e-05,
+      "loss": 0.0389,
+      "step": 606
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0217018723364316e-05,
+      "loss": 0.0409,
+      "step": 607
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189894877112956e-05,
+      "loss": 0.0433,
+      "step": 608
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162769633220673e-05,
+      "loss": 0.04,
+      "step": 609
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135643191331344e-05,
+      "loss": 0.0483,
+      "step": 610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0108515751097678e-05,
+      "loss": 0.057,
+      "step": 611
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008138751217973e-05,
+      "loss": 0.0504,
+      "step": 612
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054258674243418e-05,
+      "loss": 0.0402,
+      "step": 613
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027129436959082e-05,
+      "loss": 0.0486,
+      "step": 614
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.0389,
+      "step": 615
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97287056304092e-06,
+      "loss": 0.0448,
+      "step": 616
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945741325756589e-06,
+      "loss": 0.0392,
+      "step": 617
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918612487820274e-06,
+      "loss": 0.04,
+      "step": 618
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.891484248902322e-06,
+      "loss": 0.04,
+      "step": 619
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86435680866866e-06,
+      "loss": 0.0381,
+      "step": 620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837230366779332e-06,
+      "loss": 0.0492,
+      "step": 621
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.810105122887049e-06,
+      "loss": 0.0519,
+      "step": 622
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.782981276635686e-06,
+      "loss": 0.04,
+      "step": 623
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755859027658848e-06,
+      "loss": 0.0576,
+      "step": 624
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728738575578377e-06,
+      "loss": 0.0368,
+      "step": 625
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701620120002885e-06,
+      "loss": 0.0492,
+      "step": 626
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.674503860526297e-06,
+      "loss": 0.039,
+      "step": 627
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647389996726375e-06,
+      "loss": 0.0493,
+      "step": 628
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.620278728163237e-06,
+      "loss": 0.0376,
+      "step": 629
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.593170254377915e-06,
+      "loss": 0.041,
+      "step": 630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.566064774890868e-06,
+      "loss": 0.0415,
+      "step": 631
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538962489200503e-06,
+      "loss": 0.0389,
+      "step": 632
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511863596781733e-06,
+      "loss": 0.0491,
+      "step": 633
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.484768297084504e-06,
+      "loss": 0.0452,
+      "step": 634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.457676789532299e-06,
+      "loss": 0.0535,
+      "step": 635
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430589273520704e-06,
+      "loss": 0.0549,
+      "step": 636
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.403505948415923e-06,
+      "loss": 0.0378,
+      "step": 637
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376427013553311e-06,
+      "loss": 0.0467,
+      "step": 638
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.349352668235925e-06,
+      "loss": 0.049,
+      "step": 639
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322283111733023e-06,
+      "loss": 0.0428,
+      "step": 640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29521854327863e-06,
+      "loss": 0.0374,
+      "step": 641
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268159162070058e-06,
+      "loss": 0.0428,
+      "step": 642
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.241105167266433e-06,
+      "loss": 0.0419,
+      "step": 643
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214056757987238e-06,
+      "loss": 0.0465,
+      "step": 644
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18701413331086e-06,
+      "loss": 0.037,
+      "step": 645
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.159977492273086e-06,
+      "loss": 0.0444,
+      "step": 646
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13294703386569e-06,
+      "loss": 0.0355,
+      "step": 647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105922957034921e-06,
+      "loss": 0.0452,
+      "step": 648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.078905460680064e-06,
+      "loss": 0.0398,
+      "step": 649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05189474365198e-06,
+      "loss": 0.046,
+      "step": 650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.024891004751625e-06,
+      "loss": 0.0364,
+      "step": 651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997894442728598e-06,
+      "loss": 0.0439,
+      "step": 652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970905256279675e-06,
+      "loss": 0.0434,
+      "step": 653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943923644047343e-06,
+      "loss": 0.0378,
+      "step": 654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.916949804618353e-06,
+      "loss": 0.0492,
+      "step": 655
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.889983936522242e-06,
+      "loss": 0.0358,
+      "step": 656
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863026238229869e-06,
+      "loss": 0.0392,
+      "step": 657
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836076908151981e-06,
+      "loss": 0.035,
+      "step": 658
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.80913614463771e-06,
+      "loss": 0.0462,
+      "step": 659
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782204145973162e-06,
+      "loss": 0.0439,
+      "step": 660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755281110379922e-06,
+      "loss": 0.0478,
+      "step": 661
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728367236013595e-06,
+      "loss": 0.0402,
+      "step": 662
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.701462720962382e-06,
+      "loss": 0.0332,
+      "step": 663
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674567763245581e-06,
+      "loss": 0.0479,
+      "step": 664
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.647682560812152e-06,
+      "loss": 0.041,
+      "step": 665
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620807311539258e-06,
+      "loss": 0.0436,
+      "step": 666
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.593942213230808e-06,
+      "loss": 0.0468,
+      "step": 667
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567087463615988e-06,
+      "loss": 0.0442,
+      "step": 668
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540243260347825e-06,
+      "loss": 0.0331,
+      "step": 669
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.513409801001731e-06,
+      "loss": 0.0434,
+      "step": 670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.486587283074026e-06,
+      "loss": 0.0402,
+      "step": 671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45977590398051e-06,
+      "loss": 0.0353,
+      "step": 672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432975861054999e-06,
+      "loss": 0.0405,
+      "step": 673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406187351547872e-06,
+      "loss": 0.0359,
+      "step": 674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.379410572624629e-06,
+      "loss": 0.0432,
+      "step": 675
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.352645721364419e-06,
+      "loss": 0.049,
+      "step": 676
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32589299475861e-06,
+      "loss": 0.0364,
+      "step": 677
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299152589709336e-06,
+      "loss": 0.0528,
+      "step": 678
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.272424703028032e-06,
+      "loss": 0.0372,
+      "step": 679
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.245709531434005e-06,
+      "loss": 0.0479,
+      "step": 680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219007271552984e-06,
+      "loss": 0.0304,
+      "step": 681
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.192318119915644e-06,
+      "loss": 0.0396,
+      "step": 682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16564227295621e-06,
+      "loss": 0.0492,
+      "step": 683
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.138979927010964e-06,
+      "loss": 0.0418,
+      "step": 684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112331278316824e-06,
+      "loss": 0.0431,
+      "step": 685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.085696523009907e-06,
+      "loss": 0.0429,
+      "step": 686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.059075857124056e-06,
+      "loss": 0.0391,
+      "step": 687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032469476589424e-06,
+      "loss": 0.043,
+      "step": 688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005877577231027e-06,
+      "loss": 0.0404,
+      "step": 689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979300354767282e-06,
+      "loss": 0.0511,
+      "step": 690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952738004808603e-06,
+      "loss": 0.0464,
+      "step": 691
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926190722855931e-06,
+      "loss": 0.0304,
+      "step": 692
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899658704299302e-06,
+      "loss": 0.0302,
+      "step": 693
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873142144416423e-06,
+      "loss": 0.0309,
+      "step": 694
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846641238371212e-06,
+      "loss": 0.0301,
+      "step": 695
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82015618121238e-06,
+      "loss": 0.0458,
+      "step": 696
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79368716787199e-06,
+      "loss": 0.0354,
+      "step": 697
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767234393164017e-06,
+      "loss": 0.0318,
+      "step": 698
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740798051782922e-06,
+      "loss": 0.0343,
+      "step": 699
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71437833830222e-06,
+      "loss": 0.0428,
+      "step": 700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.687975447173025e-06,
+      "loss": 0.0307,
+      "step": 701
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66158957272266e-06,
+      "loss": 0.0427,
+      "step": 702
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635220909153192e-06,
+      "loss": 0.0461,
+      "step": 703
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608869650540014e-06,
+      "loss": 0.0419,
+      "step": 704
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582535990830416e-06,
+      "loss": 0.0325,
+      "step": 705
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.556220123842173e-06,
+      "loss": 0.0363,
+      "step": 706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.529922243262085e-06,
+      "loss": 0.0383,
+      "step": 707
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503642542644581e-06,
+      "loss": 0.0411,
+      "step": 708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.47738121541028e-06,
+      "loss": 0.0383,
+      "step": 709
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451138454844575e-06,
+      "loss": 0.0323,
+      "step": 710
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424914454096211e-06,
+      "loss": 0.0347,
+      "step": 711
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.398709406175846e-06,
+      "loss": 0.04,
+      "step": 712
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.372523503954654e-06,
+      "loss": 0.03,
+      "step": 713
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.346356940162895e-06,
+      "loss": 0.0387,
+      "step": 714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320209907388491e-06,
+      "loss": 0.0337,
+      "step": 715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.294082598075619e-06,
+      "loss": 0.0268,
+      "step": 716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267975204523294e-06,
+      "loss": 0.0346,
+      "step": 717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241887918883932e-06,
+      "loss": 0.0319,
+      "step": 718
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215820933161976e-06,
+      "loss": 0.0382,
+      "step": 719
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.189774439212442e-06,
+      "loss": 0.0406,
+      "step": 720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163748628739533e-06,
+      "loss": 0.0221,
+      "step": 721
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137743693295225e-06,
+      "loss": 0.0331,
+      "step": 722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.11175982427784e-06,
+      "loss": 0.0344,
+      "step": 723
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0857972129306584e-06,
+      "loss": 0.0371,
+      "step": 724
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.059856050340501e-06,
+      "loss": 0.0411,
+      "step": 725
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.033936527436318e-06,
+      "loss": 0.0441,
+      "step": 726
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0080388349877985e-06,
+      "loss": 0.0374,
+      "step": 727
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.982163163603957e-06,
+      "loss": 0.0359,
+      "step": 728
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.95630970373172e-06,
+      "loss": 0.0435,
+      "step": 729
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930478645654554e-06,
+      "loss": 0.041,
+      "step": 730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.90467017949103e-06,
+      "loss": 0.0338,
+      "step": 731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878884495193448e-06,
+      "loss": 0.0558,
+      "step": 732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853121782546434e-06,
+      "loss": 0.04,
+      "step": 733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.827382231165531e-06,
+      "loss": 0.0393,
+      "step": 734
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801666030495826e-06,
+      "loss": 0.037,
+      "step": 735
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.775973369810539e-06,
+      "loss": 0.0449,
+      "step": 736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7503044382096235e-06,
+      "loss": 0.0325,
+      "step": 737
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724659424618401e-06,
+      "loss": 0.037,
+      "step": 738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699038517786149e-06,
+      "loss": 0.036,
+      "step": 739
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673441906284708e-06,
+      "loss": 0.0372,
+      "step": 740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6478697785071125e-06,
+      "loss": 0.04,
+      "step": 741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6223223226661994e-06,
+      "loss": 0.0257,
+      "step": 742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.596799726793204e-06,
+      "loss": 0.026,
+      "step": 743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.571302178736404e-06,
+      "loss": 0.0498,
+      "step": 744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545829866159711e-06,
+      "loss": 0.0384,
+      "step": 745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520382976541313e-06,
+      "loss": 0.0299,
+      "step": 746
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49496169717228e-06,
+      "loss": 0.0411,
+      "step": 747
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4695662151551805e-06,
+      "loss": 0.039,
+      "step": 748
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444196717402721e-06,
+      "loss": 0.032,
+      "step": 749
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418853390636363e-06,
+      "loss": 0.0301,
+      "step": 750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.393536421384938e-06,
+      "loss": 0.0416,
+      "step": 751
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.368245995983293e-06,
+      "loss": 0.0369,
+      "step": 752
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342982300570913e-06,
+      "loss": 0.0376,
+      "step": 753
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31774552109053e-06,
+      "loss": 0.032,
+      "step": 754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.292535843286795e-06,
+      "loss": 0.0429,
+      "step": 755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267353452704876e-06,
+      "loss": 0.0333,
+      "step": 756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2421985346891055e-06,
+      "loss": 0.0349,
+      "step": 757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.217071274381623e-06,
+      "loss": 0.0292,
+      "step": 758
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.191971856720997e-06,
+      "loss": 0.0372,
+      "step": 759
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166900466440871e-06,
+      "loss": 0.0239,
+      "step": 760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.141857288068621e-06,
+      "loss": 0.0294,
+      "step": 761
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116842505923955e-06,
+      "loss": 0.0331,
+      "step": 762
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091856304117606e-06,
+      "loss": 0.0274,
+      "step": 763
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.066898866549943e-06,
+      "loss": 0.0275,
+      "step": 764
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041970376909624e-06,
+      "loss": 0.0341,
+      "step": 765
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0170710186722605e-06,
+      "loss": 0.0322,
+      "step": 766
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9922009750990425e-06,
+      "loss": 0.0269,
+      "step": 767
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.967360429235407e-06,
+      "loss": 0.0392,
+      "step": 768
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9425495639096894e-06,
+      "loss": 0.0396,
+      "step": 769
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917768561731763e-06,
+      "loss": 0.0369,
+      "step": 770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893017605091718e-06,
+      "loss": 0.0302,
+      "step": 771
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868296876158501e-06,
+      "loss": 0.0331,
+      "step": 772
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843606556878581e-06,
+      "loss": 0.039,
+      "step": 773
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8189468289746075e-06,
+      "loss": 0.0292,
+      "step": 774
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7943178739440865e-06,
+      "loss": 0.0415,
+      "step": 775
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.769719873058015e-06,
+      "loss": 0.0346,
+      "step": 776
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745153007359578e-06,
+      "loss": 0.0407,
+      "step": 777
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.720617457662801e-06,
+      "loss": 0.0303,
+      "step": 778
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696113404551218e-06,
+      "loss": 0.032,
+      "step": 779
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.671641028376547e-06,
+      "loss": 0.0442,
+      "step": 780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647200509257364e-06,
+      "loss": 0.032,
+      "step": 781
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622792027077773e-06,
+      "loss": 0.0325,
+      "step": 782
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598415761486085e-06,
+      "loss": 0.0317,
+      "step": 783
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.57407189189348e-06,
+      "loss": 0.0545,
+      "step": 784
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549760597472731e-06,
+      "loss": 0.0396,
+      "step": 785
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525482057156833e-06,
+      "loss": 0.0319,
+      "step": 786
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501236449637701e-06,
+      "loss": 0.0326,
+      "step": 787
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4770239533648885e-06,
+      "loss": 0.0261,
+      "step": 788
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452844746544234e-06,
+      "loss": 0.0377,
+      "step": 789
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4286990071365516e-06,
+      "loss": 0.0273,
+      "step": 790
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40458691285635e-06,
+      "loss": 0.0351,
+      "step": 791
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380508641170499e-06,
+      "loss": 0.0394,
+      "step": 792
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.356464369296934e-06,
+      "loss": 0.0331,
+      "step": 793
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.332454274203349e-06,
+      "loss": 0.034,
+      "step": 794
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3084785326058925e-06,
+      "loss": 0.0304,
+      "step": 795
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2845373209678705e-06,
+      "loss": 0.0364,
+      "step": 796
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.260630815498444e-06,
+      "loss": 0.0311,
+      "step": 797
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.236759192151336e-06,
+      "loss": 0.0365,
+      "step": 798
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212922626623533e-06,
+      "loss": 0.0229,
+      "step": 799
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.189121294353994e-06,
+      "loss": 0.0396,
+      "step": 800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.165355370522358e-06,
+      "loss": 0.0243,
+      "step": 801
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.141625030047659e-06,
+      "loss": 0.0306,
+      "step": 802
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.11793044758702e-06,
+      "loss": 0.0151,
+      "step": 803
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094271797534404e-06,
+      "loss": 0.0412,
+      "step": 804
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070649254019298e-06,
+      "loss": 0.0304,
+      "step": 805
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.047062990905436e-06,
+      "loss": 0.0307,
+      "step": 806
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02351318178953e-06,
+      "loss": 0.0222,
+      "step": 807
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.0394,
+      "step": 808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97652361859567e-06,
+      "loss": 0.0248,
+      "step": 809
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.953084210364508e-06,
+      "loss": 0.0306,
+      "step": 810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92968194782238e-06,
+      "loss": 0.02,
+      "step": 811
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906317003211728e-06,
+      "loss": 0.0306,
+      "step": 812
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.882989548500349e-06,
+      "loss": 0.0267,
+      "step": 813
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.859699755380106e-06,
+      "loss": 0.0198,
+      "step": 814
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836447795265673e-06,
+      "loss": 0.0261,
+      "step": 815
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813233839293265e-06,
+      "loss": 0.0249,
+      "step": 816
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7900580583193875e-06,
+      "loss": 0.0224,
+      "step": 817
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.766920622919575e-06,
+      "loss": 0.0234,
+      "step": 818
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743821703387137e-06,
+      "loss": 0.0352,
+      "step": 819
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7207614697318895e-06,
+      "loss": 0.0298,
+      "step": 820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.69774009167894e-06,
+      "loss": 0.0305,
+      "step": 821
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.674757738667405e-06,
+      "loss": 0.0257,
+      "step": 822
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651814579849161e-06,
+      "loss": 0.0277,
+      "step": 823
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6289107840876366e-06,
+      "loss": 0.0325,
+      "step": 824
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606046519956534e-06,
+      "loss": 0.0405,
+      "step": 825
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5832219557385896e-06,
+      "loss": 0.0362,
+      "step": 826
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.560437259424359e-06,
+      "loss": 0.0284,
+      "step": 827
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.537692598710962e-06,
+      "loss": 0.02,
+      "step": 828
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.514988141000853e-06,
+      "loss": 0.0286,
+      "step": 829
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.492324053400592e-06,
+      "loss": 0.0347,
+      "step": 830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469700502719607e-06,
+      "loss": 0.0278,
+      "step": 831
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.447117655468978e-06,
+      "loss": 0.0273,
+      "step": 832
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4245756778601955e-06,
+      "loss": 0.0252,
+      "step": 833
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.402074735803955e-06,
+      "loss": 0.029,
+      "step": 834
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.379614994908922e-06,
+      "loss": 0.0328,
+      "step": 835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35719662048052e-06,
+      "loss": 0.0302,
+      "step": 836
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.334819777519712e-06,
+      "loss": 0.0367,
+      "step": 837
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.312484630721786e-06,
+      "loss": 0.0294,
+      "step": 838
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290191344475143e-06,
+      "loss": 0.0349,
+      "step": 839
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267940082860088e-06,
+      "loss": 0.0404,
+      "step": 840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.245731009647624e-06,
+      "loss": 0.0255,
+      "step": 841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.223564288298233e-06,
+      "loss": 0.0292,
+      "step": 842
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.201440081960693e-06,
+      "loss": 0.0266,
+      "step": 843
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.179358553470876e-06,
+      "loss": 0.0332,
+      "step": 844
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.157319865350523e-06,
+      "loss": 0.0255,
+      "step": 845
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.135324179806079e-06,
+      "loss": 0.0284,
+      "step": 846
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113371658727489e-06,
+      "loss": 0.0271,
+      "step": 847
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091462463686995e-06,
+      "loss": 0.0328,
+      "step": 848
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.06959675593796e-06,
+      "loss": 0.0203,
+      "step": 849
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.047774696413679e-06,
+      "loss": 0.02,
+      "step": 850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.02599644572619e-06,
+      "loss": 0.0228,
+      "step": 851
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004262164165098e-06,
+      "loss": 0.0316,
+      "step": 852
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982572011696388e-06,
+      "loss": 0.0266,
+      "step": 853
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.960926147961253e-06,
+      "loss": 0.0268,
+      "step": 854
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.939324732274924e-06,
+      "loss": 0.0317,
+      "step": 855
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917767923625471e-06,
+      "loss": 0.0348,
+      "step": 856
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.896255880672677e-06,
+      "loss": 0.0255,
+      "step": 857
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.874788761746836e-06,
+      "loss": 0.0351,
+      "step": 858
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853366724847588e-06,
+      "loss": 0.0218,
+      "step": 859
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831989927642784e-06,
+      "loss": 0.0328,
+      "step": 860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.810658527467298e-06,
+      "loss": 0.0317,
+      "step": 861
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7893726813218734e-06,
+      "loss": 0.0283,
+      "step": 862
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.76813254587198e-06,
+      "loss": 0.0223,
+      "step": 863
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.74693827744665e-06,
+      "loss": 0.0211,
+      "step": 864
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.725790032037334e-06,
+      "loss": 0.0368,
+      "step": 865
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.704687965296746e-06,
+      "loss": 0.0293,
+      "step": 866
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6836322325377226e-06,
+      "loss": 0.0209,
+      "step": 867
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6626229887320807e-06,
+      "loss": 0.0274,
+      "step": 868
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6416603885094726e-06,
+      "loss": 0.031,
+      "step": 869
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6207445861562497e-06,
+      "loss": 0.0316,
+      "step": 870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59987573561433e-06,
+      "loss": 0.028,
+      "step": 871
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5790539904800605e-06,
+      "loss": 0.0251,
+      "step": 872
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5582795040030905e-06,
+      "loss": 0.0358,
+      "step": 873
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5375524290852394e-06,
+      "loss": 0.0272,
+      "step": 874
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.516872918279377e-06,
+      "loss": 0.0219,
+      "step": 875
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4962411237882945e-06,
+      "loss": 0.0287,
+      "step": 876
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4756571974635935e-06,
+      "loss": 0.0286,
+      "step": 877
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4551212908045497e-06,
+      "loss": 0.0358,
+      "step": 878
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4346335549570186e-06,
+      "loss": 0.0302,
+      "step": 879
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414194140712325e-06,
+      "loss": 0.022,
+      "step": 880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3938031985061216e-06,
+      "loss": 0.0347,
+      "step": 881
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.373460878417315e-06,
+      "loss": 0.021,
+      "step": 882
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3531673301669563e-06,
+      "loss": 0.0233,
+      "step": 883
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3329227031171165e-06,
+      "loss": 0.0268,
+      "step": 884
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312727146269812e-06,
+      "loss": 0.0316,
+      "step": 885
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.292580808265897e-06,
+      "loss": 0.031,
+      "step": 886
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2724838373839716e-06,
+      "loss": 0.0331,
+      "step": 887
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.252436381539291e-06,
+      "loss": 0.018,
+      "step": 888
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2324385882826726e-06,
+      "loss": 0.0363,
+      "step": 889
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2124906047994165e-06,
+      "loss": 0.0305,
+      "step": 890
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.192592577908221e-06,
+      "loss": 0.0266,
+      "step": 891
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1727446540600882e-06,
+      "loss": 0.0235,
+      "step": 892
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1529469793372735e-06,
+      "loss": 0.0196,
+      "step": 893
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1331996994521917e-06,
+      "loss": 0.024,
+      "step": 894
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1135029597463328e-06,
+      "loss": 0.0286,
+      "step": 895
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0938569051892276e-06,
+      "loss": 0.0202,
+      "step": 896
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0742616803773518e-06,
+      "loss": 0.0249,
+      "step": 897
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.054717429533063e-06,
+      "loss": 0.0325,
+      "step": 898
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0352242965035562e-06,
+      "loss": 0.0242,
+      "step": 899
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.015782424759792e-06,
+      "loss": 0.0267,
+      "step": 900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9963919573954447e-06,
+      "loss": 0.0235,
+      "step": 901
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.977053037125849e-06,
+      "loss": 0.0277,
+      "step": 902
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9577658062869465e-06,
+      "loss": 0.0344,
+      "step": 903
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9385304068342414e-06,
+      "loss": 0.0335,
+      "step": 904
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9193469803417574e-06,
+      "loss": 0.0214,
+      "step": 905
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.900215668000991e-06,
+      "loss": 0.0236,
+      "step": 906
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8811366106198734e-06,
+      "loss": 0.0225,
+      "step": 907
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8621099486217376e-06,
+      "loss": 0.0333,
+      "step": 908
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.843135822044283e-06,
+      "loss": 0.0306,
+      "step": 909
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8242143705385417e-06,
+      "loss": 0.0315,
+      "step": 910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8053457333678535e-06,
+      "loss": 0.0363,
+      "step": 911
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.786530049406844e-06,
+      "loss": 0.0266,
+      "step": 912
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.767767457140399e-06,
+      "loss": 0.0289,
+      "step": 913
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7490580946626355e-06,
+      "loss": 0.0217,
+      "step": 914
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.730402099675904e-06,
+      "loss": 0.0198,
+      "step": 915
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7117996094897737e-06,
+      "loss": 0.0367,
+      "step": 916
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6932507610199964e-06,
+      "loss": 0.0244,
+      "step": 917
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.674755690787526e-06,
+      "loss": 0.0233,
+      "step": 918
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6563145349175136e-06,
+      "loss": 0.0325,
+      "step": 919
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.637927429138276e-06,
+      "loss": 0.0239,
+      "step": 920
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6195945087803345e-06,
+      "loss": 0.0237,
+      "step": 921
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6013159087753927e-06,
+      "loss": 0.0308,
+      "step": 922
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5830917636553563e-06,
+      "loss": 0.0211,
+      "step": 923
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.564922207551337e-06,
+      "loss": 0.0321,
+      "step": 924
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.546807374192668e-06,
+      "loss": 0.0258,
+      "step": 925
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5287473969059174e-06,
+      "loss": 0.0282,
+      "step": 926
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.510742408613912e-06,
+      "loss": 0.0221,
+      "step": 927
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4927925418347455e-06,
+      "loss": 0.0306,
+      "step": 928
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.474897928680827e-06,
+      "loss": 0.0315,
+      "step": 929
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4570587008578896e-06,
+      "loss": 0.024,
+      "step": 930
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4392749896640157e-06,
+      "loss": 0.0162,
+      "step": 931
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.421546925988697e-06,
+      "loss": 0.0253,
+      "step": 932
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4038746403118495e-06,
+      "loss": 0.0285,
+      "step": 933
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.386258262702851e-06,
+      "loss": 0.0301,
+      "step": 934
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3686979228195994e-06,
+      "loss": 0.03,
+      "step": 935
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.351193749907551e-06,
+      "loss": 0.025,
+      "step": 936
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333745872798767e-06,
+      "loss": 0.032,
+      "step": 937
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3163544199109656e-06,
+      "loss": 0.0276,
+      "step": 938
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2990195192465803e-06,
+      "loss": 0.0267,
+      "step": 939
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281741298391815e-06,
+      "loss": 0.0294,
+      "step": 940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2645198845157078e-06,
+      "loss": 0.0302,
+      "step": 941
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2473554043691915e-06,
+      "loss": 0.0286,
+      "step": 942
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2302479842841608e-06,
+      "loss": 0.0254,
+      "step": 943
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2131977501725465e-06,
+      "loss": 0.0201,
+      "step": 944
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1962048275253845e-06,
+      "loss": 0.0294,
+      "step": 945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.179269341411896e-06,
+      "loss": 0.0186,
+      "step": 946
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1623914164785618e-06,
+      "loss": 0.0269,
+      "step": 947
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1455711769482113e-06,
+      "loss": 0.0282,
+      "step": 948
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1288087466191053e-06,
+      "loss": 0.0365,
+      "step": 949
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1121042488640166e-06,
+      "loss": 0.0281,
+      "step": 950
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.095457806629335e-06,
+      "loss": 0.0275,
+      "step": 951
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0788695424341674e-06,
+      "loss": 0.0355,
+      "step": 952
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.062339578369408e-06,
+      "loss": 0.034,
+      "step": 953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045868036096864e-06,
+      "loss": 0.0262,
+      "step": 954
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0294550368483668e-06,
+      "loss": 0.0337,
+      "step": 955
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.013100701424844e-06,
+      "loss": 0.0327,
+      "step": 956
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9968051501954676e-06,
+      "loss": 0.0294,
+      "step": 957
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9805685030967527e-06,
+      "loss": 0.0282,
+      "step": 958
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.964390879631669e-06,
+      "loss": 0.0273,
+      "step": 959
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9482723988687734e-06,
+      "loss": 0.0409,
+      "step": 960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.932213179441327e-06,
+      "loss": 0.026,
+      "step": 961
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916213339546421e-06,
+      "loss": 0.0307,
+      "step": 962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9002729969441113e-06,
+      "loss": 0.0285,
+      "step": 963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8843922689565418e-06,
+      "loss": 0.0227,
+      "step": 964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8685712724671e-06,
+      "loss": 0.0242,
+      "step": 965
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8528101239195394e-06,
+      "loss": 0.0279,
+      "step": 966
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8371089393171193e-06,
+      "loss": 0.0241,
+      "step": 967
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8214678342217774e-06,
+      "loss": 0.0252,
+      "step": 968
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8058869237532506e-06,
+      "loss": 0.0297,
+      "step": 969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.790366322588236e-06,
+      "loss": 0.022,
+      "step": 970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7749061449595562e-06,
+      "loss": 0.0274,
+      "step": 971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7595065046553085e-06,
+      "loss": 0.023,
+      "step": 972
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7441675150180337e-06,
+      "loss": 0.0312,
+      "step": 973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.728889288943877e-06,
+      "loss": 0.0228,
+      "step": 974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.713671938881758e-06,
+      "loss": 0.0281,
+      "step": 975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6985155768325456e-06,
+      "loss": 0.0297,
+      "step": 976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6834203143482341e-06,
+      "loss": 0.0205,
+      "step": 977
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6683862625311165e-06,
+      "loss": 0.0217,
+      "step": 978
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6534135320329736e-06,
+      "loss": 0.0237,
+      "step": 979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6385022330542533e-06,
+      "loss": 0.025,
+      "step": 980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.623652475343268e-06,
+      "loss": 0.0363,
+      "step": 981
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6088643681953752e-06,
+      "loss": 0.0223,
+      "step": 982
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5941380204521851e-06,
+      "loss": 0.0262,
+      "step": 983
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.579473540500751e-06,
+      "loss": 0.0309,
+      "step": 984
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.564871036272777e-06,
+      "loss": 0.0199,
+      "step": 985
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5503306152438146e-06,
+      "loss": 0.0226,
+      "step": 986
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5358523844324834e-06,
+      "loss": 0.0264,
+      "step": 987
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5214364503996838e-06,
+      "loss": 0.0304,
+      "step": 988
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5070829192477943e-06,
+      "loss": 0.0302,
+      "step": 989
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4927918966199095e-06,
+      "loss": 0.0271,
+      "step": 990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.478563487699065e-06,
+      "loss": 0.0287,
+      "step": 991
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4643977972074386e-06,
+      "loss": 0.0268,
+      "step": 992
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4502949294056056e-06,
+      "loss": 0.0259,
+      "step": 993
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4362549880917609e-06,
+      "loss": 0.0306,
+      "step": 994
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4222780766009536e-06,
+      "loss": 0.0234,
+      "step": 995
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4083642978043278e-06,
+      "loss": 0.0243,
+      "step": 996
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3945137541083697e-06,
+      "loss": 0.0236,
+      "step": 997
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3807265474541465e-06,
+      "loss": 0.0227,
+      "step": 998
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3670027793165642e-06,
+      "loss": 0.032,
+      "step": 999
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3533425507036057e-06,
+      "loss": 0.0309,
+      "step": 1000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.0262,
+      "step": 1001
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3262131137445266e-06,
+      "loss": 0.031,
+      "step": 1002
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.312744105073146e-06,
+      "loss": 0.0202,
+      "step": 1003
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2993390352744184e-06,
+      "loss": 0.0224,
+      "step": 1004
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2859980030106922e-06,
+      "loss": 0.0278,
+      "step": 1005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2727211064729862e-06,
+      "loss": 0.0269,
+      "step": 1006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2595084433802829e-06,
+      "loss": 0.0254,
+      "step": 1007
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2463601109788058e-06,
+      "loss": 0.0245,
+      "step": 1008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2332762060412905e-06,
+      "loss": 0.0136,
+      "step": 1009
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.220256824866285e-06,
+      "loss": 0.028,
+      "step": 1010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2073020632774391e-06,
+      "loss": 0.0197,
+      "step": 1011
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1944120166227947e-06,
+      "loss": 0.022,
+      "step": 1012
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.18158677977409e-06,
+      "loss": 0.02,
+      "step": 1013
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1688264471260546e-06,
+      "loss": 0.0315,
+      "step": 1014
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1561311125957208e-06,
+      "loss": 0.0201,
+      "step": 1015
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1435008696217298e-06,
+      "loss": 0.0285,
+      "step": 1016
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1309358111636405e-06,
+      "loss": 0.029,
+      "step": 1017
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1184360297012532e-06,
+      "loss": 0.0244,
+      "step": 1018
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1060016172339206e-06,
+      "loss": 0.0276,
+      "step": 1019
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.093632665279878e-06,
+      "loss": 0.0212,
+      "step": 1020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.081329264875567e-06,
+      "loss": 0.0193,
+      "step": 1021
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0690915065749564e-06,
+      "loss": 0.0241,
+      "step": 1022
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0569194804488914e-06,
+      "loss": 0.0253,
+      "step": 1023
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0448132760844287e-06,
+      "loss": 0.0293,
+      "step": 1024
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.032772982584157e-06,
+      "loss": 0.0341,
+      "step": 1025
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0207986885655664e-06,
+      "loss": 0.0265,
+      "step": 1026
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0088904821603884e-06,
+      "loss": 0.0269,
+      "step": 1027
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.970484510139323e-07,
+      "loss": 0.0309,
+      "step": 1028
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.852726822844639e-07,
+      "loss": 0.0229,
+      "step": 1029
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.735632626425463e-07,
+      "loss": 0.0298,
+      "step": 1030
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.619202782704118e-07,
+      "loss": 0.0258,
+      "step": 1031
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.503438148613208e-07,
+      "loss": 0.032,
+      "step": 1032
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.388339576189376e-07,
+      "loss": 0.0224,
+      "step": 1033
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.273907912566959e-07,
+      "loss": 0.0189,
+      "step": 1034
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.16014399997186e-07,
+      "loss": 0.0288,
+      "step": 1035
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.047048675715164e-07,
+      "loss": 0.0248,
+      "step": 1036
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.93462277218724e-07,
+      "loss": 0.0183,
+      "step": 1037
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.822867116851397e-07,
+      "loss": 0.0297,
+      "step": 1038
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.7117825322378e-07,
+      "loss": 0.0262,
+      "step": 1039
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.601369835937623e-07,
+      "loss": 0.0395,
+      "step": 1040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.491629840596805e-07,
+      "loss": 0.0302,
+      "step": 1041
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.382563353910122e-07,
+      "loss": 0.0287,
+      "step": 1042
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.274171178615354e-07,
+      "loss": 0.0281,
+      "step": 1043
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.166454112487287e-07,
+      "loss": 0.0234,
+      "step": 1044
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.059412948331768e-07,
+      "loss": 0.0288,
+      "step": 1045
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.953048473980041e-07,
+      "loss": 0.0152,
+      "step": 1046
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.847361472282822e-07,
+      "loss": 0.0245,
+      "step": 1047
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.742352721104607e-07,
+      "loss": 0.0263,
+      "step": 1048
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.638022993317873e-07,
+      "loss": 0.0227,
+      "step": 1049
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.534373056797451e-07,
+      "loss": 0.0317,
+      "step": 1050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.431403674414883e-07,
+      "loss": 0.0301,
+      "step": 1051
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.329115604032732e-07,
+      "loss": 0.0287,
+      "step": 1052
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.227509598499094e-07,
+      "loss": 0.0214,
+      "step": 1053
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.126586405641989e-07,
+      "loss": 0.0268,
+      "step": 1054
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.026346768263892e-07,
+      "loss": 0.0222,
+      "step": 1055
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.926791424136259e-07,
+      "loss": 0.0236,
+      "step": 1056
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.827921105994096e-07,
+      "loss": 0.0281,
+      "step": 1057
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.729736541530551e-07,
+      "loss": 0.0255,
+      "step": 1058
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.63223845339157e-07,
+      "loss": 0.0273,
+      "step": 1059
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.535427559170638e-07,
+      "loss": 0.0365,
+      "step": 1060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.439304571403349e-07,
+      "loss": 0.0233,
+      "step": 1061
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.343870197562307e-07,
+      "loss": 0.0274,
+      "step": 1062
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.249125140051893e-07,
+      "loss": 0.0322,
+      "step": 1063
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.155070096202986e-07,
+      "loss": 0.0202,
+      "step": 1064
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.061705758267978e-07,
+      "loss": 0.0304,
+      "step": 1065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.969032813415577e-07,
+      "loss": 0.038,
+      "step": 1066
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.877051943725798e-07,
+      "loss": 0.03,
+      "step": 1067
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.785763826184931e-07,
+      "loss": 0.0223,
+      "step": 1068
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.695169132680556e-07,
+      "loss": 0.0226,
+      "step": 1069
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.605268529996588e-07,
+      "loss": 0.0295,
+      "step": 1070
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.51606267980842e-07,
+      "loss": 0.0276,
+      "step": 1071
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.427552238677925e-07,
+      "loss": 0.0323,
+      "step": 1072
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.339737858048811e-07,
+      "loss": 0.0249,
+      "step": 1073
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.252620184241697e-07,
+      "loss": 0.0215,
+      "step": 1074
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.166199858449317e-07,
+      "loss": 0.0269,
+      "step": 1075
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.080477516731974e-07,
+      "loss": 0.0255,
+      "step": 1076
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.995453790012706e-07,
+      "loss": 0.0253,
+      "step": 1077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.911129304072648e-07,
+      "loss": 0.0233,
+      "step": 1078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.827504679546569e-07,
+      "loss": 0.0291,
+      "step": 1079
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7445805319181305e-07,
+      "loss": 0.0179,
+      "step": 1080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.662357471515433e-07,
+      "loss": 0.0181,
+      "step": 1081
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5808361035065364e-07,
+      "loss": 0.0231,
+      "step": 1082
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5000170278949984e-07,
+      "loss": 0.0217,
+      "step": 1083
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.419900839515434e-07,
+      "loss": 0.0276,
+      "step": 1084
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3404881280291744e-07,
+      "loss": 0.0349,
+      "step": 1085
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.261779477919892e-07,
+      "loss": 0.0185,
+      "step": 1086
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.183775468489304e-07,
+      "loss": 0.025,
+      "step": 1087
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1064766738529326e-07,
+      "loss": 0.0253,
+      "step": 1088
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0298836629358626e-07,
+      "loss": 0.0229,
+      "step": 1089
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9539969994685676e-07,
+      "loss": 0.0375,
+      "step": 1090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.878817241982713e-07,
+      "loss": 0.0278,
+      "step": 1091
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.804344943807126e-07,
+      "loss": 0.0276,
+      "step": 1092
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7305806530636647e-07,
+      "loss": 0.0235,
+      "step": 1093
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6575249126631683e-07,
+      "loss": 0.0243,
+      "step": 1094
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.585178260301514e-07,
+      "loss": 0.0325,
+      "step": 1095
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5135412284556637e-07,
+      "loss": 0.019,
+      "step": 1096
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.442614344379669e-07,
+      "loss": 0.0286,
+      "step": 1097
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.372398130100851e-07,
+      "loss": 0.023,
+      "step": 1098
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.302893102415994e-07,
+      "loss": 0.0283,
+      "step": 1099
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2340997728874223e-07,
+      "loss": 0.0251,
+      "step": 1100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.166018647839353e-07,
+      "loss": 0.0151,
+      "step": 1101
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0986502283541055e-07,
+      "loss": 0.0221,
+      "step": 1102
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.031995010268451e-07,
+      "loss": 0.0266,
+      "step": 1103
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9660534841699175e-07,
+      "loss": 0.0324,
+      "step": 1104
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.900826135393231e-07,
+      "loss": 0.0134,
+      "step": 1105
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8363134440166806e-07,
+      "loss": 0.0171,
+      "step": 1106
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.772515884858673e-07,
+      "loss": 0.0313,
+      "step": 1107
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7094339274741254e-07,
+      "loss": 0.0251,
+      "step": 1108
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6470680361511346e-07,
+      "loss": 0.028,
+      "step": 1109
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.585418669907458e-07,
+      "loss": 0.0213,
+      "step": 1110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.52448628248716e-07,
+      "loss": 0.027,
+      "step": 1111
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4642713223573254e-07,
+      "loss": 0.0218,
+      "step": 1112
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.40477423270471e-07,
+      "loss": 0.0224,
+      "step": 1113
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.345995451432448e-07,
+      "loss": 0.0214,
+      "step": 1114
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2879354111569163e-07,
+      "loss": 0.0275,
+      "step": 1115
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2305945392044893e-07,
+      "loss": 0.0245,
+      "step": 1116
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.173973257608397e-07,
+      "loss": 0.02,
+      "step": 1117
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1180719831056184e-07,
+      "loss": 0.0276,
+      "step": 1118
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0628911271338593e-07,
+      "loss": 0.0331,
+      "step": 1119
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.008431095828467e-07,
+      "loss": 0.0353,
+      "step": 1120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9546922900194553e-07,
+      "loss": 0.0231,
+      "step": 1121
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9016751052285952e-07,
+      "loss": 0.0274,
+      "step": 1122
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8493799316664395e-07,
+      "loss": 0.021,
+      "step": 1123
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7978071542294916e-07,
+      "loss": 0.0236,
+      "step": 1124
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.746957152497375e-07,
+      "loss": 0.0209,
+      "step": 1125
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6968303007300124e-07,
+      "loss": 0.0335,
+      "step": 1126
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6474269678648956e-07,
+      "loss": 0.019,
+      "step": 1127
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.598747517514365e-07,
+      "loss": 0.0279,
+      "step": 1128
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.550792307962934e-07,
+      "loss": 0.023,
+      "step": 1129
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5035616921646234e-07,
+      "loss": 0.0339,
+      "step": 1130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.457056017740399e-07,
+      "loss": 0.029,
+      "step": 1131
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4112756269756278e-07,
+      "loss": 0.0159,
+      "step": 1132
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3662208568175017e-07,
+      "loss": 0.0192,
+      "step": 1133
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3218920388725853e-07,
+      "loss": 0.025,
+      "step": 1134
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2782894994044393e-07,
+      "loss": 0.0329,
+      "step": 1135
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2354135593310669e-07,
+      "loss": 0.0226,
+      "step": 1136
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1932645342227046e-07,
+      "loss": 0.0219,
+      "step": 1137
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1518427342994243e-07,
+      "loss": 0.0227,
+      "step": 1138
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1111484644288684e-07,
+      "loss": 0.0311,
+      "step": 1139
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0711820241240067e-07,
+      "loss": 0.0233,
+      "step": 1140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0319437075409056e-07,
+      "loss": 0.0325,
+      "step": 1141
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.934338034765956e-08,
+      "loss": 0.0243,
+      "step": 1142
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.556525953669516e-08,
+      "loss": 0.0236,
+      "step": 1143
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.186003612845496e-08,
+      "loss": 0.0305,
+      "step": 1144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.82277373936713e-08,
+      "loss": 0.0234,
+      "step": 1145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.466839006634364e-08,
+      "loss": 0.0329,
+      "step": 1146
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.118202034353872e-08,
+      "loss": 0.0274,
+      "step": 1147
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.77686538852085e-08,
+      "loss": 0.0199,
+      "step": 1148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.442831581399357e-08,
+      "loss": 0.0235,
+      "step": 1149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.116103071503788e-08,
+      "loss": 0.0338,
+      "step": 1150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.796682263581322e-08,
+      "loss": 0.0201,
+      "step": 1151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.484571508593718e-08,
+      "loss": 0.0212,
+      "step": 1152
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.179773103699993e-08,
+      "loss": 0.0236,
+      "step": 1153
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8822892922399956e-08,
+      "loss": 0.0226,
+      "step": 1154
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.592122263717414e-08,
+      "loss": 0.0279,
+      "step": 1155
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3092741537841276e-08,
+      "loss": 0.0347,
+      "step": 1156
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.033747044223769e-08,
+      "loss": 0.0209,
+      "step": 1157
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7655429629372975e-08,
+      "loss": 0.0177,
+      "step": 1158
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.504663883927341e-08,
+      "loss": 0.0291,
+      "step": 1159
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.251111727283985e-08,
+      "loss": 0.0187,
+      "step": 1160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.004888359170678e-08,
+      "loss": 0.0258,
+      "step": 1161
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7659955918103455e-08,
+      "loss": 0.0167,
+      "step": 1162
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5344351834721844e-08,
+      "loss": 0.0252,
+      "step": 1163
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.310208838458562e-08,
+      "loss": 0.0209,
+      "step": 1164
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.093318207092799e-08,
+      "loss": 0.0257,
+      "step": 1165
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8837648857066304e-08,
+      "loss": 0.0274,
+      "step": 1166
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.681550416628431e-08,
+      "loss": 0.0274,
+      "step": 1167
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4866762881725627e-08,
+      "loss": 0.0278,
+      "step": 1168
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2991439346274902e-08,
+      "loss": 0.0154,
+      "step": 1169
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.118954736245682e-08,
+      "loss": 0.0232,
+      "step": 1170
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9461100192337267e-08,
+      "loss": 0.0224,
+      "step": 1171
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7806110557418988e-08,
+      "loss": 0.0324,
+      "step": 1172
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6224590638553863e-08,
+      "loss": 0.0312,
+      "step": 1173
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4716552075849655e-08,
+      "loss": 0.0222,
+      "step": 1174
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3282005968587864e-08,
+      "loss": 0.0244,
+      "step": 1175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1920962875137109e-08,
+      "loss": 0.0141,
+      "step": 1176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0633432812878763e-08,
+      "loss": 0.0284,
+      "step": 1177
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.419425258135884e-09,
+      "loss": 0.0314,
+      "step": 1178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.278949146094394e-09,
+      "loss": 0.0238,
+      "step": 1179
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.212012870748686e-09,
+      "loss": 0.0202,
+      "step": 1180
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.218624284831664e-09,
+      "loss": 0.017,
+      "step": 1181
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.2987906997581385e-09,
+      "loss": 0.0229,
+      "step": 1182
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.452518885575963e-09,
+      "loss": 0.0181,
+      "step": 1183
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6798150709105306e-09,
+      "loss": 0.0262,
+      "step": 1184
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9806849429203642e-09,
+      "loss": 0.0336,
+      "step": 1185
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3551336472582563e-09,
+      "loss": 0.0204,
+      "step": 1186
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8031657880301924e-09,
+      "loss": 0.0258,
+      "step": 1187
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.324785427760933e-09,
+      "loss": 0.0189,
+      "step": 1188
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.199960873673697e-10,
+      "loss": 0.0209,
+      "step": 1189
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.888007461307688e-10,
+      "loss": 0.023,
+      "step": 1190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3120184167345636e-10,
+      "loss": 0.0245,
+      "step": 1191
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4720126994327567e-10,
+      "loss": 0.0209,
+      "step": 1192
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6800385199153853e-11,
+      "loss": 0.0312,
+      "step": 1193
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.017,
+      "step": 1194
+    },
+    {
+      "epoch": 3.0,
+      "step": 1194,
+      "total_flos": 7.638268890547487e+17,
+      "train_loss": 0.38923207115088676,
+      "train_runtime": 44465.4246,
+      "train_samples_per_second": 3.439,
+      "train_steps_per_second": 0.027
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.028035340830683708,
+      "eval_runtime": 304.3464,
+      "eval_samples_per_second": 41.867,
+      "eval_steps_per_second": 1.311,
+      "step": 1194
     }
   ],
-  "max_steps": 75,
+  "max_steps": 1194,
   "num_train_epochs": 3,
-  "total_flos": 2.4061184768475136e+16,
+  "total_flos": 7.638268890547487e+17,
   "trial_name": null,
   "trial_params": null
 }