diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,38682 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999223903764066,
+  "eval_steps": 500,
+  "global_step": 6442,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.6997,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263218254025221e-07,
+      "loss": 1.4604,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.342003565741054e-07,
+      "loss": 1.511,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526436508050443e-06,
+      "loss": 1.44,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2220814333545174e-06,
+      "loss": 1.444,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3605221819766275e-06,
+      "loss": 1.4222,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4775721671301136e-06,
+      "loss": 1.4141,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789654762075664e-06,
+      "loss": 1.4233,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6684007131482108e-06,
+      "loss": 1.3823,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7484032587570397e-06,
+      "loss": 1.3929,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8207743643763844e-06,
+      "loss": 1.3562,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8868440073791497e-06,
+      "loss": 0.9379,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9476221872440144e-06,
+      "loss": 1.336,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0038939925326357e-06,
+      "loss": 0.86,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0562817899286226e-06,
+      "loss": 1.3466,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1052873016100886e-06,
+      "loss": 1.3665,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1513209038718584e-06,
+      "loss": 1.3431,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.194722538550733e-06,
+      "loss": 1.3343,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2357769630532252e-06,
+      "loss": 0.8422,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2747250841595615e-06,
+      "loss": 0.792,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.311772523704219e-06,
+      "loss": 1.3071,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.347096189778907e-06,
+      "loss": 1.3147,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.380849386033331e-06,
+      "loss": 1.2676,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.413165832781672e-06,
+      "loss": 1.3264,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4441628667090348e-06,
+      "loss": 1.2895,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4739440126465366e-06,
+      "loss": 1.293,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.502601069722316e-06,
+      "loss": 1.341,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.530215817935158e-06,
+      "loss": 1.3293,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.556861425126305e-06,
+      "loss": 1.3002,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5826036153311447e-06,
+      "loss": 1.2522,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.607501645485261e-06,
+      "loss": 1.3078,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.631609127012611e-06,
+      "loss": 1.2617,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6549747209504897e-06,
+      "loss": 1.2612,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.677642729274381e-06,
+      "loss": 1.295,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6996536004846307e-06,
+      "loss": 1.2329,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.721044363953255e-06,
+      "loss": 1.2692,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7418490047471415e-06,
+      "loss": 1.2806,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.762098788455747e-06,
+      "loss": 1.3087,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7818225438181194e-06,
+      "loss": 1.2707,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8010469095620836e-06,
+      "loss": 1.233,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.819796550759532e-06,
+      "loss": 1.2178,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.838094349106741e-06,
+      "loss": 1.2374,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8559615708121775e-06,
+      "loss": 1.2632,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8734180151814286e-06,
+      "loss": 1.2264,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8904821465027275e-06,
+      "loss": 1.2551,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9071712114358532e-06,
+      "loss": 0.7733,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9235013437754734e-06,
+      "loss": 1.242,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9394876581841944e-06,
+      "loss": 1.2606,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.955144334260227e-06,
+      "loss": 1.2807,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.970484692111557e-06,
+      "loss": 1.2876,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9855212604459638e-06,
+      "loss": 0.7206,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0002658380490587e-06,
+      "loss": 1.2439,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0147295494061477e-06,
+      "loss": 1.2275,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0289228951248383e-06,
+      "loss": 1.2293,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.042855797730902e-06,
+      "loss": 1.2154,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0565376433376804e-06,
+      "loss": 1.2221,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06997731962733e-06,
+      "loss": 1.259,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.083183250528827e-06,
+      "loss": 1.2122,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.096163427931584e-06,
+      "loss": 1.186,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.108925440733667e-06,
+      "loss": 1.2271,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.121476501488993e-06,
+      "loss": 1.2328,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.133823470887783e-06,
+      "loss": 1.2398,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1459728802783244e-06,
+      "loss": 0.7516,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.157930952415133e-06,
+      "loss": 1.2116,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.169703620598531e-06,
+      "loss": 1.2226,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1812965463530115e-06,
+      "loss": 1.2577,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.192715135776243e-06,
+      "loss": 1.245,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.203964554676903e-06,
+      "loss": 1.258,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2150497426074365e-06,
+      "loss": 1.2162,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2259754258871533e-06,
+      "loss": 1.2358,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2367461297016074e-06,
+      "loss": 1.2223,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2473661893557772e-06,
+      "loss": 1.2344,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2578397607510923e-06,
+      "loss": 1.1773,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.268170830149664e-06,
+      "loss": 1.219,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2783632232831393e-06,
+      "loss": 1.23,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.288420613858269e-06,
+      "loss": 1.2056,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2983465315064983e-06,
+      "loss": 1.2514,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.308144369220642e-06,
+      "loss": 1.2155,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3178173903178495e-06,
+      "loss": 1.2144,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3273687349646058e-06,
+      "loss": 1.194,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3368014262964216e-06,
+      "loss": 1.2513,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.346118376162054e-06,
+      "loss": 1.1853,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.355322390519571e-06,
+      "loss": 1.2551,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3644161745092633e-06,
+      "loss": 1.2394,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.373402337226376e-06,
+      "loss": 1.2108,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3822833962146993e-06,
+      "loss": 1.2249,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.39106178170041e-06,
+      "loss": 0.7066,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3997398405839512e-06,
+      "loss": 1.251,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.408319840206372e-06,
+      "loss": 1.1949,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.41680397190525e-06,
+      "loss": 1.1782,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4251943543741276e-06,
+      "loss": 1.2124,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4334930368383754e-06,
+      "loss": 1.2132,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4417020020593663e-06,
+      "loss": 1.2415,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4498231691779956e-06,
+      "loss": 1.2377,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4578583964077424e-06,
+      "loss": 1.2183,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.465809483586716e-06,
+      "loss": 1.2141,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4736781745974772e-06,
+      "loss": 1.1881,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4814661596627497e-06,
+      "loss": 1.2148,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.489175077524595e-06,
+      "loss": 1.2455,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4968065175140795e-06,
+      "loss": 1.1521,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.504362021517959e-06,
+      "loss": 1.1442,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.511843085848486e-06,
+      "loss": 0.7065,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5192511630219986e-06,
+      "loss": 1.228,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.526587663451581e-06,
+      "loss": 1.255,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5338539570587357e-06,
+      "loss": 1.2191,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54105137480867e-06,
+      "loss": 1.2206,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5481812101734914e-06,
+      "loss": 1.186,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5552447205273605e-06,
+      "loss": 0.7098,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5622431284773286e-06,
+      "loss": 1.2118,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.569177623133424e-06,
+      "loss": 1.2175,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.576049361321247e-06,
+      "loss": 0.6933,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.582859468740202e-06,
+      "loss": 1.2065,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5896090410702415e-06,
+      "loss": 1.2043,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5962991450298524e-06,
+      "loss": 1.2265,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6029308193878483e-06,
+      "loss": 0.6824,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.609505075931349e-06,
+      "loss": 1.2274,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.616022900392225e-06,
+      "loss": 1.2177,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.622485253334106e-06,
+      "loss": 1.1999,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6288930710019724e-06,
+      "loss": 1.1599,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.635247266136189e-06,
+      "loss": 1.1792,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6415487287527687e-06,
+      "loss": 1.2027,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.647798326891515e-06,
+      "loss": 1.1514,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.653996907333637e-06,
+      "loss": 1.2333,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.660145296290305e-06,
+      "loss": 1.2027,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6662443000635524e-06,
+      "loss": 1.213,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6722947056808465e-06,
+      "loss": 1.2478,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6782972815045803e-06,
+      "loss": 1.2408,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.684252777817655e-06,
+      "loss": 0.6991,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6901619273862825e-06,
+      "loss": 1.151,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6960254460010533e-06,
+      "loss": 1.2664,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.701844032997277e-06,
+      "loss": 1.1937,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.707618371755534e-06,
+      "loss": 1.211,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.713349130183339e-06,
+      "loss": 0.6936,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.719036961178765e-06,
+      "loss": 1.195,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7246825030768334e-06,
+      "loss": 1.1553,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7302863800794253e-06,
+      "loss": 1.1698,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.735849202669451e-06,
+      "loss": 1.1859,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7413715680099586e-06,
+      "loss": 1.1854,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7468540603288257e-06,
+      "loss": 1.2103,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7522972512896746e-06,
+      "loss": 1.2231,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.757701700349579e-06,
+      "loss": 1.2263,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7630679551041295e-06,
+      "loss": 1.1974,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7683965516203984e-06,
+      "loss": 1.1948,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7736880147582994e-06,
+      "loss": 1.215,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7789428584808222e-06,
+      "loss": 1.1633,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7841615861536144e-06,
+      "loss": 1.179,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789344690834332e-06,
+      "loss": 0.6824,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.794492655552186e-06,
+      "loss": 1.2186,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7996059535780702e-06,
+      "loss": 1.1834,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8046850486856615e-06,
+      "loss": 1.1867,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8097303954038432e-06,
+      "loss": 1.2078,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.814742439260792e-06,
+      "loss": 1.2158,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.819721617020069e-06,
+      "loss": 1.2094,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8246683569090205e-06,
+      "loss": 1.1613,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.829583078839778e-06,
+      "loss": 1.2034,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.834466194623164e-06,
+      "loss": 1.1546,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.839318108175754e-06,
+      "loss": 1.2136,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8441392157203725e-06,
+      "loss": 1.1999,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.848929905980253e-06,
+      "loss": 1.1871,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.853690560367128e-06,
+      "loss": 1.1699,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.858421553163444e-06,
+      "loss": 1.1669,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.863123251698943e-06,
+      "loss": 1.1838,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.867796016521808e-06,
+      "loss": 1.2083,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.872440201564576e-06,
+      "loss": 0.692,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.877056154305006e-06,
+      "loss": 1.1757,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.881644215922093e-06,
+      "loss": 1.193,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.886204721447382e-06,
+      "loss": 1.1592,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.890737999911785e-06,
+      "loss": 1.2161,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.895244374488029e-06,
+      "loss": 1.2183,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8997241626288986e-06,
+      "loss": 1.1699,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.904177676201436e-06,
+      "loss": 1.2091,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.908605221617221e-06,
+      "loss": 0.664,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9130070999588795e-06,
+      "loss": 1.1522,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.917383607102933e-06,
+      "loss": 1.2207,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.921735033839148e-06,
+      "loss": 1.2431,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.926061665986473e-06,
+      "loss": 1.2209,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.930363784505689e-06,
+      "loss": 1.2028,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.934641665608895e-06,
+      "loss": 1.2214,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9388955808659186e-06,
+      "loss": 0.6866,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.943125797307773e-06,
+      "loss": 1.1956,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.947332577527231e-06,
+      "loss": 0.6935,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.95151617977665e-06,
+      "loss": 1.186,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.955676858063098e-06,
+      "loss": 1.2178,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9598148622408975e-06,
+      "loss": 1.2106,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.963930438101659e-06,
+      "loss": 0.6608,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.968023827461888e-06,
+      "loss": 1.183,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.972095268248243e-06,
+      "loss": 1.1725,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.976144994580518e-06,
+      "loss": 1.2199,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.98017323685243e-06,
+      "loss": 1.2174,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.984180221810264e-06,
+      "loss": 1.1707,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.98816617262947e-06,
+      "loss": 0.6939,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.992131308989239e-06,
+      "loss": 1.1607,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.996075847145162e-06,
+      "loss": 1.1851,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999999999999e-06,
+      "loss": 1.1766,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999747176351e-06,
+      "loss": 1.1851,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999998988705469e-06,
+      "loss": 1.1988,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999997724587546e-06,
+      "loss": 1.1891,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999995954822901e-06,
+      "loss": 1.2192,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999993679411981e-06,
+      "loss": 0.6839,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999990898355362e-06,
+      "loss": 1.168,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999987611653747e-06,
+      "loss": 1.178,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999983819307967e-06,
+      "loss": 1.1959,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99997952131898e-06,
+      "loss": 1.1502,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999974717687875e-06,
+      "loss": 1.1923,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999969408415863e-06,
+      "loss": 1.2183,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999963593504289e-06,
+      "loss": 1.1656,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999957272954622e-06,
+      "loss": 1.2099,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99995044676846e-06,
+      "loss": 1.2014,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99994311494753e-06,
+      "loss": 1.1927,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999935277493683e-06,
+      "loss": 1.1744,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999269344089036e-06,
+      "loss": 1.1313,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999180856952996e-06,
+      "loss": 0.661,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999908731355108e-06,
+      "loss": 1.2144,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999898871390694e-06,
+      "loss": 0.7184,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999888505804551e-06,
+      "loss": 1.1833,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999877634599299e-06,
+      "loss": 1.1623,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999866257777686e-06,
+      "loss": 1.1826,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99985437534259e-06,
+      "loss": 1.1646,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999841987297013e-06,
+      "loss": 1.182,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999829093644089e-06,
+      "loss": 1.1734,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999815694387077e-06,
+      "loss": 1.1914,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999801789529365e-06,
+      "loss": 0.7231,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999787379074467e-06,
+      "loss": 1.2141,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997724630260284e-06,
+      "loss": 0.693,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999757041387819e-06,
+      "loss": 1.1907,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999741114163737e-06,
+      "loss": 1.2042,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999724681357812e-06,
+      "loss": 1.1771,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999707742974197e-06,
+      "loss": 1.1708,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999690299017172e-06,
+      "loss": 1.2061,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999672349491152e-06,
+      "loss": 1.1841,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999653894400672e-06,
+      "loss": 1.2113,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999634933750398e-06,
+      "loss": 1.1834,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999615467545124e-06,
+      "loss": 1.2052,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999595495789773e-06,
+      "loss": 1.2044,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995750184893915e-06,
+      "loss": 1.1781,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995540356491585e-06,
+      "loss": 1.1514,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999532547274379e-06,
+      "loss": 1.1897,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999510553370486e-06,
+      "loss": 1.1882,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999488053943038e-06,
+      "loss": 1.2039,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999465048997727e-06,
+      "loss": 1.2106,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999441538540365e-06,
+      "loss": 1.1931,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994175225769e-06,
+      "loss": 1.1705,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993930011134005e-06,
+      "loss": 1.1701,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999367974156068e-06,
+      "loss": 1.1637,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999342441711229e-06,
+      "loss": 1.2067,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99931640378534e-06,
+      "loss": 1.1977,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999289860384982e-06,
+      "loss": 1.1876,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999262811516868e-06,
+      "loss": 1.1704,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999235257187834e-06,
+      "loss": 1.1407,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999207197404849e-06,
+      "loss": 1.1595,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999178632175006e-06,
+      "loss": 1.177,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999149561505526e-06,
+      "loss": 1.1771,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9991199854037615e-06,
+      "loss": 1.1834,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999089903877187e-06,
+      "loss": 0.7063,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99905931693341e-06,
+      "loss": 1.1526,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9990282245801626e-06,
+      "loss": 0.7188,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998996626825305e-06,
+      "loss": 1.196,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998964523676827e-06,
+      "loss": 1.1544,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998931915142846e-06,
+      "loss": 1.1862,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9988988012316035e-06,
+      "loss": 1.1918,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9988651819514735e-06,
+      "loss": 0.6783,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998831057310955e-06,
+      "loss": 1.2079,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998796427318675e-06,
+      "loss": 1.1729,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998761291983391e-06,
+      "loss": 1.1828,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998725651313984e-06,
+      "loss": 1.1499,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998689505319464e-06,
+      "loss": 1.1528,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986528540089716e-06,
+      "loss": 1.2101,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998615697391773e-06,
+      "loss": 1.1905,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998578035477261e-06,
+      "loss": 1.2011,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998539868274959e-06,
+      "loss": 1.2195,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9985011957945145e-06,
+      "loss": 1.127,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998462018045706e-06,
+      "loss": 1.2007,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998422335038439e-06,
+      "loss": 1.1881,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983821467827465e-06,
+      "loss": 1.1904,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998341453288787e-06,
+      "loss": 1.1833,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998300254566851e-06,
+      "loss": 1.1803,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998258550627353e-06,
+      "loss": 1.1752,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9982163414808386e-06,
+      "loss": 1.1615,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998173627137977e-06,
+      "loss": 0.7142,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998130407609569e-06,
+      "loss": 1.2126,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998086682906541e-06,
+      "loss": 0.672,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998042453039947e-06,
+      "loss": 1.14,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99799771802097e-06,
+      "loss": 1.1991,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997952477860921e-06,
+      "loss": 1.1539,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997906732571236e-06,
+      "loss": 0.6987,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978604821634815e-06,
+      "loss": 1.1494,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997813726649351e-06,
+      "loss": 0.673,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9977664660406646e-06,
+      "loss": 1.1753,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997718700349371e-06,
+      "loss": 1.154,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997670429587547e-06,
+      "loss": 1.2017,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997621653767396e-06,
+      "loss": 1.1597,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9975723729012505e-06,
+      "loss": 0.6717,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997522587001569e-06,
+      "loss": 1.1659,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99747229608094e-06,
+      "loss": 1.1937,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997421500152075e-06,
+      "loss": 1.1901,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99737019922782e-06,
+      "loss": 1.1652,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997318393321143e-06,
+      "loss": 1.1619,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997266082445142e-06,
+      "loss": 1.1647,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997213266613044e-06,
+      "loss": 1.1726,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997159945838201e-06,
+      "loss": 1.1542,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997106120134093e-06,
+      "loss": 0.6907,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997051789514329e-06,
+      "loss": 0.6828,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996996953992646e-06,
+      "loss": 1.1612,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996941613582906e-06,
+      "loss": 1.1834,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9968857682991016e-06,
+      "loss": 1.223,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996829418155351e-06,
+      "loss": 0.7015,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9967725631659025e-06,
+      "loss": 1.1759,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996715203345127e-06,
+      "loss": 1.2109,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99665733870753e-06,
+      "loss": 1.1656,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99659896926774e-06,
+      "loss": 1.2029,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996540095040514e-06,
+      "loss": 1.1353,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996480716040736e-06,
+      "loss": 1.1827,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996420832283419e-06,
+      "loss": 1.1777,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996360443783703e-06,
+      "loss": 1.2046,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996299550556855e-06,
+      "loss": 1.203,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996238152618272e-06,
+      "loss": 1.1746,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996176249983476e-06,
+      "loss": 1.1517,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9961138426681165e-06,
+      "loss": 0.6969,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996050930687972e-06,
+      "loss": 1.1666,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995987514058949e-06,
+      "loss": 1.1686,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99592359279708e-06,
+      "loss": 1.1478,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995859166918526e-06,
+      "loss": 1.1831,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995794236439576e-06,
+      "loss": 1.1806,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9957288013766445e-06,
+      "loss": 1.1581,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995662861746276e-06,
+      "loss": 1.1661,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9955964175651424e-06,
+      "loss": 1.1507,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99552946885004e-06,
+      "loss": 1.1547,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995462015617897e-06,
+      "loss": 1.1827,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995394057885767e-06,
+      "loss": 1.1298,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99532559567083e-06,
+      "loss": 1.1539,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995256628990396e-06,
+      "loss": 1.1566,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995187157861902e-06,
+      "loss": 0.6841,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99511718230291e-06,
+      "loss": 1.1828,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995046702331114e-06,
+      "loss": 1.1785,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99497571796433e-06,
+      "loss": 0.6634,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994904229220507e-06,
+      "loss": 1.1876,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994832236117718e-06,
+      "loss": 1.1257,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994759738674164e-06,
+      "loss": 1.2092,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994686736908176e-06,
+      "loss": 1.1981,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994613230838209e-06,
+      "loss": 1.1615,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994539220482848e-06,
+      "loss": 1.1726,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994464705860803e-06,
+      "loss": 1.1792,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9943896869909145e-06,
+      "loss": 1.179,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994314163892148e-06,
+      "loss": 1.1442,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9942381365836e-06,
+      "loss": 0.6716,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994161605084488e-06,
+      "loss": 0.6757,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994084569414164e-06,
+      "loss": 1.222,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994007029592104e-06,
+      "loss": 1.1731,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993928985637911e-06,
+      "loss": 1.162,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9938504375713175e-06,
+      "loss": 1.1769,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993771385412181e-06,
+      "loss": 1.2044,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993691829180489e-06,
+      "loss": 1.1921,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993611768896353e-06,
+      "loss": 1.1632,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993531204580017e-06,
+      "loss": 1.1775,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993450136251848e-06,
+      "loss": 1.1711,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993368563932343e-06,
+      "loss": 0.7148,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993286487642123e-06,
+      "loss": 1.1919,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993203907401942e-06,
+      "loss": 1.1398,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9931208232326755e-06,
+      "loss": 1.1792,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993037235155332e-06,
+      "loss": 1.1706,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992953143191042e-06,
+      "loss": 1.1789,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992868547361066e-06,
+      "loss": 1.1755,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9927834476867945e-06,
+      "loss": 1.219,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99269784418974e-06,
+      "loss": 1.1638,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992611736891546e-06,
+      "loss": 1.1579,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9925251258139824e-06,
+      "loss": 1.175,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992438010978946e-06,
+      "loss": 1.117,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992350392408463e-06,
+      "loss": 1.1532,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9922622701246845e-06,
+      "loss": 0.6806,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9921736441498894e-06,
+      "loss": 1.1561,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992084514506486e-06,
+      "loss": 1.137,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991994881217006e-06,
+      "loss": 0.7027,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991904744304112e-06,
+      "loss": 1.1256,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991814103790594e-06,
+      "loss": 1.1762,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9917229596993665e-06,
+      "loss": 1.2014,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991631312053473e-06,
+      "loss": 1.1833,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991539160876086e-06,
+      "loss": 1.1301,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991446506190501e-06,
+      "loss": 1.2185,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991353348020144e-06,
+      "loss": 1.1949,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991259686388569e-06,
+      "loss": 1.204,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991165521319453e-06,
+      "loss": 1.1709,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991070852836606e-06,
+      "loss": 1.1861,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990975680963962e-06,
+      "loss": 1.1303,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990880005725581e-06,
+      "loss": 1.1174,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990783827145654e-06,
+      "loss": 1.1546,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990687145248495e-06,
+      "loss": 1.1978,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99058996005855e-06,
+      "loss": 1.1911,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990492271600388e-06,
+      "loss": 0.7157,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990394079898707e-06,
+      "loss": 1.131,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990295384978333e-06,
+      "loss": 1.1686,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990196186864218e-06,
+      "loss": 1.1405,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9900964855814415e-06,
+      "loss": 0.6984,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989996281155211e-06,
+      "loss": 1.1685,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9898955736108595e-06,
+      "loss": 1.1438,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989794362973848e-06,
+      "loss": 0.7084,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989692649269767e-06,
+      "loss": 1.1404,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98959043252433e-06,
+      "loss": 1.1581,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989487712763382e-06,
+      "loss": 1.1616,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98938449001289e-06,
+      "loss": 0.6944,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989280764298954e-06,
+      "loss": 1.1759,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9891765356477955e-06,
+      "loss": 1.1577,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989071804085769e-06,
+      "loss": 1.1878,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988966569639351e-06,
+      "loss": 0.677,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988860832335149e-06,
+      "loss": 1.1765,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988754592199893e-06,
+      "loss": 1.1616,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988647849260446e-06,
+      "loss": 1.1531,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988540603543794e-06,
+      "loss": 1.1437,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988432855077051e-06,
+      "loss": 1.1456,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988324603887459e-06,
+      "loss": 0.7026,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988215850002386e-06,
+      "loss": 1.1413,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988106593449328e-06,
+      "loss": 1.1214,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987996834255906e-06,
+      "loss": 1.1672,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9878865724498714e-06,
+      "loss": 1.1593,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987775808059101e-06,
+      "loss": 1.1502,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987664541111598e-06,
+      "loss": 1.2029,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987552771635494e-06,
+      "loss": 1.1662,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987440499659046e-06,
+      "loss": 1.1995,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98732772521064e-06,
+      "loss": 1.1706,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987214448318787e-06,
+      "loss": 1.18,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987100669012126e-06,
+      "loss": 1.1161,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986986387319425e-06,
+      "loss": 1.1222,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986871603269575e-06,
+      "loss": 0.6672,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986756316891598e-06,
+      "loss": 1.1839,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9866405282146396e-06,
+      "loss": 0.6785,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9865242372679746e-06,
+      "loss": 1.1862,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986407444081003e-06,
+      "loss": 1.1817,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986290148683255e-06,
+      "loss": 1.1683,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9861723511043835e-06,
+      "loss": 1.1229,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986054051374172e-06,
+      "loss": 1.1126,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985935249522528e-06,
+      "loss": 0.7094,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98581594557949e-06,
+      "loss": 1.171,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985696139575218e-06,
+      "loss": 1.1443,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9855758315400035e-06,
+      "loss": 1.1237,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985455021504263e-06,
+      "loss": 1.1748,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985333709498539e-06,
+      "loss": 1.1336,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985211895553504e-06,
+      "loss": 1.125,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985089579699953e-06,
+      "loss": 1.1225,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9849667619688126e-06,
+      "loss": 0.656,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984843442391133e-06,
+      "loss": 1.1469,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984719620998092e-06,
+      "loss": 1.1656,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984595297820995e-06,
+      "loss": 1.1346,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9844704728912745e-06,
+      "loss": 1.1815,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984345146240488e-06,
+      "loss": 1.1617,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984219317900321e-06,
+      "loss": 1.1299,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984092987902588e-06,
+      "loss": 1.1668,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983966156279225e-06,
+      "loss": 1.1251,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983838823062301e-06,
+      "loss": 1.1571,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983710988284007e-06,
+      "loss": 1.1875,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983582651976663e-06,
+      "loss": 1.1496,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9834538141727155e-06,
+      "loss": 1.1377,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983324474904739e-06,
+      "loss": 1.1277,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983194634205431e-06,
+      "loss": 1.1595,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98306429210762e-06,
+      "loss": 1.1626,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982933448644261e-06,
+      "loss": 1.2028,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9828021038484305e-06,
+      "loss": 1.172,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982670257753338e-06,
+      "loss": 1.1943,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982537910392318e-06,
+      "loss": 0.694,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982405061798829e-06,
+      "loss": 1.1172,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982271712006459e-06,
+      "loss": 1.1949,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982137861048924e-06,
+      "loss": 1.1587,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982003508960062e-06,
+      "loss": 0.6577,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981868655773841e-06,
+      "loss": 1.2017,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981733301524356e-06,
+      "loss": 1.116,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981597446245826e-06,
+      "loss": 1.1755,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981461089972601e-06,
+      "loss": 1.1487,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981324232739153e-06,
+      "loss": 1.1207,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9811868745800834e-06,
+      "loss": 1.1084,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98104901553012e-06,
+      "loss": 1.1313,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980910655624116e-06,
+      "loss": 1.1952,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980771794897054e-06,
+      "loss": 1.1686,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980632433384038e-06,
+      "loss": 0.6608,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980492571120304e-06,
+      "loss": 1.1612,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9803522081412126e-06,
+      "loss": 1.1398,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980211344482251e-06,
+      "loss": 1.14,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980069980179031e-06,
+      "loss": 0.6789,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979928115267294e-06,
+      "loss": 1.1509,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979785749782908e-06,
+      "loss": 1.1322,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979642883761865e-06,
+      "loss": 1.2132,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979499517240286e-06,
+      "loss": 0.6401,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979355650254415e-06,
+      "loss": 1.1686,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979211282840628e-06,
+      "loss": 1.1372,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9790664150354225e-06,
+      "loss": 1.1795,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.978921046875425e-06,
+      "loss": 1.1609,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978775178397389e-06,
+      "loss": 0.6418,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978628809638192e-06,
+      "loss": 1.1782,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9784819406348404e-06,
+      "loss": 1.1689,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978334571424466e-06,
+      "loss": 1.1558,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978186702044327e-06,
+      "loss": 1.1687,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9780383325318075e-06,
+      "loss": 1.0936,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977889462924421e-06,
+      "loss": 1.1487,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977740093259803e-06,
+      "loss": 1.1652,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9775902235757194e-06,
+      "loss": 1.1215,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97743985391006e-06,
+      "loss": 1.1865,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977288984300841e-06,
+      "loss": 1.1584,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977137614786208e-06,
+      "loss": 1.1292,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976985745404429e-06,
+      "loss": 1.1454,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9768333761939e-06,
+      "loss": 1.1454,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976680507193145e-06,
+      "loss": 1.1816,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976527138440811e-06,
+      "loss": 1.1447,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976373269975675e-06,
+      "loss": 0.6654,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976218901836638e-06,
+      "loss": 1.1874,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9760640340627276e-06,
+      "loss": 1.1144,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9759086666930985e-06,
+      "loss": 1.1849,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975752799767031e-06,
+      "loss": 1.1468,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975596433323932e-06,
+      "loss": 1.1568,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975439567403334e-06,
+      "loss": 1.1359,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975282202044898e-06,
+      "loss": 1.1755,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975124337288408e-06,
+      "loss": 1.1522,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974965973173777e-06,
+      "loss": 1.1658,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974807109741042e-06,
+      "loss": 1.1215,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974647747030369e-06,
+      "loss": 1.1808,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974487885082048e-06,
+      "loss": 1.1242,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974327523936495e-06,
+      "loss": 0.6794,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974166663634255e-06,
+      "loss": 1.1574,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974005304215995e-06,
+      "loss": 1.1512,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973843445722512e-06,
+      "loss": 1.1047,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9736810881947275e-06,
+      "loss": 0.6552,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97351823167369e-06,
+      "loss": 1.1965,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973354876200572e-06,
+      "loss": 0.6915,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9731910218166734e-06,
+      "loss": 1.1551,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973026668563422e-06,
+      "loss": 1.1609,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972861816482369e-06,
+      "loss": 1.1678,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972696465615194e-06,
+      "loss": 1.1374,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972530616003701e-06,
+      "loss": 1.1545,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97236426768982e-06,
+      "loss": 1.136,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972197420715609e-06,
+      "loss": 1.1607,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972030075123251e-06,
+      "loss": 0.6861,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971862230955053e-06,
+      "loss": 1.1661,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971693888253452e-06,
+      "loss": 1.1655,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971525047061008e-06,
+      "loss": 0.6904,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971355707420409e-06,
+      "loss": 1.1979,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971185869374467e-06,
+      "loss": 1.1749,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971015532966121e-06,
+      "loss": 1.1325,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970844698238437e-06,
+      "loss": 1.1319,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970673365234605e-06,
+      "loss": 1.1594,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970501533997943e-06,
+      "loss": 1.1157,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970329204571894e-06,
+      "loss": 0.6731,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970156377000026e-06,
+      "loss": 0.6491,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9699830513260346e-06,
+      "loss": 1.1334,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969809227593741e-06,
+      "loss": 0.6433,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9696349058470905e-06,
+      "loss": 1.1912,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969460086130158e-06,
+      "loss": 1.2072,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96928476848714e-06,
+      "loss": 1.1743,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969108952962361e-06,
+      "loss": 1.174,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968932639600273e-06,
+      "loss": 1.1532,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968755828445451e-06,
+      "loss": 1.1524,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9685785195425976e-06,
+      "loss": 1.1506,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96840071293654e-06,
+      "loss": 1.1429,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968222408672232e-06,
+      "loss": 1.1428,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.968043606794753e-06,
+      "loss": 1.1297,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9678643073493096e-06,
+      "loss": 1.1526,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967684510381233e-06,
+      "loss": 1.1505,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967504215935978e-06,
+      "loss": 1.1481,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967323424059128e-06,
+      "loss": 1.1518,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967142134796393e-06,
+      "loss": 1.1657,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9669603481936055e-06,
+      "loss": 1.1511,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9667780642967266e-06,
+      "loss": 1.175,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966595283151841e-06,
+      "loss": 1.2148,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966412004805161e-06,
+      "loss": 1.1656,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9662282293030235e-06,
+      "loss": 1.1454,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966043956691891e-06,
+      "loss": 1.1555,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965859187018352e-06,
+      "loss": 1.1215,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965673920329121e-06,
+      "loss": 1.172,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965488156671037e-06,
+      "loss": 1.1436,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965301896091066e-06,
+      "loss": 1.1574,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9651151386363e-06,
+      "loss": 1.1786,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964927884353953e-06,
+      "loss": 1.1513,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964740133291371e-06,
+      "loss": 1.1716,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964551885496018e-06,
+      "loss": 1.1328,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964363141015491e-06,
+      "loss": 1.1482,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964173899897506e-06,
+      "loss": 1.1305,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963984162189911e-06,
+      "loss": 1.1414,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963793927940673e-06,
+      "loss": 1.1299,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96360319719789e-06,
+      "loss": 1.1764,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963411970009782e-06,
+      "loss": 1.1453,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963220246424696e-06,
+      "loss": 1.1187,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963028026491104e-06,
+      "loss": 1.1428,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962835310257605e-06,
+      "loss": 1.136,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96264209777292e-06,
+      "loss": 1.146,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9624483890859e-06,
+      "loss": 1.1216,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962254184245517e-06,
+      "loss": 0.6762,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962059483300872e-06,
+      "loss": 1.1547,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96186428630119e-06,
+      "loss": 1.1415,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961668593295822e-06,
+      "loss": 0.6647,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961472404334241e-06,
+      "loss": 1.1186,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961275719466051e-06,
+      "loss": 1.1331,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9610785387409784e-06,
+      "loss": 1.1671,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960880862208874e-06,
+      "loss": 1.1633,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960682689919715e-06,
+      "loss": 1.175,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9604840219236056e-06,
+      "loss": 1.1472,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960284858270772e-06,
+      "loss": 1.1209,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9600851990115686e-06,
+      "loss": 1.1951,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959885044196474e-06,
+      "loss": 1.1903,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959684393876091e-06,
+      "loss": 1.1357,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9594832481011495e-06,
+      "loss": 1.1462,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9592816069225035e-06,
+      "loss": 1.206,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959079470391134e-06,
+      "loss": 1.1247,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958876838558144e-06,
+      "loss": 1.1508,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958673711474765e-06,
+      "loss": 1.0963,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958470089192352e-06,
+      "loss": 1.1874,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958265971762386e-06,
+      "loss": 1.1644,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958061359236472e-06,
+      "loss": 1.1414,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957856251666339e-06,
+      "loss": 1.136,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957650649103847e-06,
+      "loss": 1.1572,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957444551600975e-06,
+      "loss": 1.1533,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957237959209829e-06,
+      "loss": 0.6989,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9570308719826415e-06,
+      "loss": 1.1626,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956823289971768e-06,
+      "loss": 1.1411,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956615213229691e-06,
+      "loss": 0.6895,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956406641809017e-06,
+      "loss": 1.1791,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956197575762478e-06,
+      "loss": 1.1471,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.955988015142929e-06,
+      "loss": 0.6644,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955777960003355e-06,
+      "loss": 1.1481,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95556741039686e-06,
+      "loss": 1.1861,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955356366376679e-06,
+      "loss": 1.1825,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9551448279961655e-06,
+      "loss": 1.1479,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954932795308804e-06,
+      "loss": 1.1268,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9547202683682e-06,
+      "loss": 1.1474,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954507247228086e-06,
+      "loss": 1.1246,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954293731942319e-06,
+      "loss": 1.1363,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95407972256488e-06,
+      "loss": 1.1726,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953865219149875e-06,
+      "loss": 1.1579,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953650221751537e-06,
+      "loss": 1.1619,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953434730424223e-06,
+      "loss": 1.1419,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953218745222412e-06,
+      "loss": 1.1468,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953002266200711e-06,
+      "loss": 1.1463,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952785293413852e-06,
+      "loss": 1.1676,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9525678269166904e-06,
+      "loss": 1.1339,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9523498667642055e-06,
+      "loss": 1.177,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952131413011505e-06,
+      "loss": 1.1483,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951912465713818e-06,
+      "loss": 1.1444,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951693024926501e-06,
+      "loss": 1.1484,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95147309070503e-06,
+      "loss": 0.7034,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951252663105014e-06,
+      "loss": 1.1708,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951031742182179e-06,
+      "loss": 1.1744,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9508103279923815e-06,
+      "loss": 1.2062,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950588420591598e-06,
+      "loss": 1.1537,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950366020035934e-06,
+      "loss": 1.1851,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950143126381617e-06,
+      "loss": 1.2109,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949919739684999e-06,
+      "loss": 1.1433,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949695860002558e-06,
+      "loss": 1.161,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949471487390896e-06,
+      "loss": 0.6946,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94924662190674e-06,
+      "loss": 1.1347,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949021263606941e-06,
+      "loss": 1.1439,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948795412548474e-06,
+      "loss": 0.6662,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948569068788442e-06,
+      "loss": 1.1464,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9483422323840674e-06,
+      "loss": 1.1475,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948114903392701e-06,
+      "loss": 1.1228,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947887081871817e-06,
+      "loss": 1.1682,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947658767879014e-06,
+      "loss": 1.1826,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947429961472015e-06,
+      "loss": 1.2041,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947200662708668e-06,
+      "loss": 1.2008,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946970871646945e-06,
+      "loss": 1.1663,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946740588344942e-06,
+      "loss": 1.1369,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946509812860882e-06,
+      "loss": 1.1153,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946278545253107e-06,
+      "loss": 1.1529,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94604678558009e-06,
+      "loss": 1.1402,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9458145339004245e-06,
+      "loss": 1.1543,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945581790272829e-06,
+      "loss": 1.134,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945348554756147e-06,
+      "loss": 1.1085,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945114827409345e-06,
+      "loss": 1.1408,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944880608291515e-06,
+      "loss": 1.173,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944645897461875e-06,
+      "loss": 1.187,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944410694979762e-06,
+      "loss": 1.1243,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944175000904644e-06,
+      "loss": 1.1341,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943938815296109e-06,
+      "loss": 1.1558,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94370213821387e-06,
+      "loss": 1.1383,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943464969717765e-06,
+      "loss": 0.6856,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9432273098677536e-06,
+      "loss": 1.1692,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942989158723925e-06,
+      "loss": 1.1726,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942750516346488e-06,
+      "loss": 0.6806,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942511382795776e-06,
+      "loss": 1.1673,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942271758132251e-06,
+      "loss": 1.1725,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942031642416492e-06,
+      "loss": 1.1539,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941791035709208e-06,
+      "loss": 1.1343,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9415499380712295e-06,
+      "loss": 1.1485,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941308349563513e-06,
+      "loss": 0.7076,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941066270247135e-06,
+      "loss": 1.1543,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9408237001833e-06,
+      "loss": 1.1718,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940580639433337e-06,
+      "loss": 1.1155,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940337088058696e-06,
+      "loss": 1.0928,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940093046120952e-06,
+      "loss": 1.1246,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9398485136818065e-06,
+      "loss": 1.1731,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939603490803081e-06,
+      "loss": 1.1432,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939357977546726e-06,
+      "loss": 1.16,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93911197397481e-06,
+      "loss": 1.135,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93886548014953e-06,
+      "loss": 1.17,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9386184961332044e-06,
+      "loss": 0.6923,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938371021988278e-06,
+      "loss": 1.1397,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938123057777318e-06,
+      "loss": 1.1553,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937874603563015e-06,
+      "loss": 1.1326,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937625659408184e-06,
+      "loss": 1.1506,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9373762253757635e-06,
+      "loss": 1.1523,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9371263015288175e-06,
+      "loss": 1.1872,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9368758879305324e-06,
+      "loss": 1.1456,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936624984644217e-06,
+      "loss": 1.1356,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936373591733309e-06,
+      "loss": 1.1746,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936121709261364e-06,
+      "loss": 1.1157,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9358693372920645e-06,
+      "loss": 1.1561,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935616475889216e-06,
+      "loss": 1.1207,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935363125116747e-06,
+      "loss": 1.11,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9351092850387115e-06,
+      "loss": 0.6946,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934854955719286e-06,
+      "loss": 1.1413,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934600137222772e-06,
+      "loss": 1.1325,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934344829613592e-06,
+      "loss": 1.152,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9340890329562945e-06,
+      "loss": 1.1203,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933832747315551e-06,
+      "loss": 1.1467,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933575972756158e-06,
+      "loss": 1.1085,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933318709343031e-06,
+      "loss": 1.1624,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933060957141214e-06,
+      "loss": 1.1042,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932802716215874e-06,
+      "loss": 1.1524,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932543986632298e-06,
+      "loss": 1.2026,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932284768455901e-06,
+      "loss": 1.1591,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9320250617522194e-06,
+      "loss": 1.1542,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931764866586911e-06,
+      "loss": 1.1297,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931504183025762e-06,
+      "loss": 1.1819,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931243011134678e-06,
+      "loss": 0.6859,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930981350979689e-06,
+      "loss": 1.1155,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93071920262695e-06,
+      "loss": 0.7116,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930456566142738e-06,
+      "loss": 1.1697,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930193441593454e-06,
+      "loss": 1.1216,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92992982904562e-06,
+      "loss": 1.1428,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929665728565886e-06,
+      "loss": 1.1299,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929401140221022e-06,
+      "loss": 1.1316,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929136064077921e-06,
+      "loss": 1.1383,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928870500203603e-06,
+      "loss": 1.111,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928604448665207e-06,
+      "loss": 1.1061,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928337909529998e-06,
+      "loss": 1.1404,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9280708828653625e-06,
+      "loss": 1.1748,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927803368738812e-06,
+      "loss": 1.1333,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92753536721798e-06,
+      "loss": 1.1248,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9272668783706235e-06,
+      "loss": 1.1217,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926997902264623e-06,
+      "loss": 1.1213,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926728438967982e-06,
+      "loss": 0.7024,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926458488548827e-06,
+      "loss": 1.1753,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9261880510754085e-06,
+      "loss": 0.6726,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925917126616099e-06,
+      "loss": 0.6978,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9256457152393945e-06,
+      "loss": 1.1474,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925373817013915e-06,
+      "loss": 1.1629,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925101432008401e-06,
+      "loss": 1.1793,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924828560291719e-06,
+      "loss": 1.1239,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924555201932859e-06,
+      "loss": 1.1301,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92428135700093e-06,
+      "loss": 1.1314,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924007025565168e-06,
+      "loss": 1.1232,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9237322076949285e-06,
+      "loss": 1.1565,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923456903459695e-06,
+      "loss": 1.1224,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923181112929068e-06,
+      "loss": 1.1376,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9229048361727764e-06,
+      "loss": 0.7062,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922628073260667e-06,
+      "loss": 1.1197,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9223508242627135e-06,
+      "loss": 1.1696,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922073089249011e-06,
+      "loss": 1.196,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921794868289777e-06,
+      "loss": 1.1656,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921516161455353e-06,
+      "loss": 1.1472,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921236968816202e-06,
+      "loss": 1.1306,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92095729044291e-06,
+      "loss": 1.1446,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920677126406188e-06,
+      "loss": 1.1615,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9203964767768665e-06,
+      "loss": 1.1086,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920115341625902e-06,
+      "loss": 1.1695,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919833721024371e-06,
+      "loss": 1.1452,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919551615043473e-06,
+      "loss": 1.1742,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919269023754533e-06,
+      "loss": 1.1445,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9189859472289945e-06,
+      "loss": 1.1578,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918702385538428e-06,
+      "loss": 1.1076,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918418338754524e-06,
+      "loss": 1.1748,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918133806949096e-06,
+      "loss": 1.1025,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91784879019408e-06,
+      "loss": 0.6799,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917563288561536e-06,
+      "loss": 1.1425,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9172773021236446e-06,
+      "loss": 1.1419,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91699083095271e-06,
+      "loss": 0.6789,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91670387512116e-06,
+      "loss": 1.1645,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916416434701543e-06,
+      "loss": 1.1929,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9161285097665306e-06,
+      "loss": 0.6689,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915840100388917e-06,
+      "loss": 1.1456,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915551206641619e-06,
+      "loss": 1.1437,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915261828597675e-06,
+      "loss": 1.1609,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914971966330249e-06,
+      "loss": 1.1272,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914681619912623e-06,
+      "loss": 1.1569,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914390789418204e-06,
+      "loss": 0.6852,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914099474920521e-06,
+      "loss": 1.1344,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913807676493225e-06,
+      "loss": 1.1327,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913515394210088e-06,
+      "loss": 1.128,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91322262814501e-06,
+      "loss": 1.1252,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912929378372005e-06,
+      "loss": 1.123,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912635644965215e-06,
+      "loss": 1.1748,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912341427998904e-06,
+      "loss": 1.1683,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912046727547454e-06,
+      "loss": 1.1826,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9117515436853755e-06,
+      "loss": 1.1437,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911455876487296e-06,
+      "loss": 1.1202,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911159726027968e-06,
+      "loss": 1.1567,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910863092382264e-06,
+      "loss": 1.1448,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9105659756251825e-06,
+      "loss": 1.1136,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9102683758318395e-06,
+      "loss": 1.1541,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909970293077475e-06,
+      "loss": 1.1338,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909671727437453e-06,
+      "loss": 1.1194,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909372678987259e-06,
+      "loss": 1.1482,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909073147802496e-06,
+      "loss": 1.1417,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908773133958894e-06,
+      "loss": 0.685,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908472637532305e-06,
+      "loss": 1.1434,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908171658598699e-06,
+      "loss": 1.1699,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907870197234174e-06,
+      "loss": 1.1081,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907568253514944e-06,
+      "loss": 1.1479,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907265827517348e-06,
+      "loss": 1.1688,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9069629193178465e-06,
+      "loss": 0.662,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906659528993023e-06,
+      "loss": 1.1638,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.90635565661958e-06,
+      "loss": 1.1045,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.906051302274344e-06,
+      "loss": 1.1658,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905746466034264e-06,
+      "loss": 1.1302,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905441147976409e-06,
+      "loss": 1.1542,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90513534817797e-06,
+      "loss": 1.1077,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9048290667162626e-06,
+      "loss": 0.675,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90452230366872e-06,
+      "loss": 1.1195,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9042150591129e-06,
+      "loss": 1.1439,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90390733312648e-06,
+      "loss": 1.1723,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903599125787263e-06,
+      "loss": 0.7103,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903290437173169e-06,
+      "loss": 1.1332,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902981267362242e-06,
+      "loss": 1.1299,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902671616432648e-06,
+      "loss": 1.1493,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902361484462674e-06,
+      "loss": 1.1078,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902050871530728e-06,
+      "loss": 1.1131,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9017397777153405e-06,
+      "loss": 1.1118,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901428203095165e-06,
+      "loss": 1.1419,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901116147748973e-06,
+      "loss": 1.1246,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900803611755661e-06,
+      "loss": 1.1552,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900490595194243e-06,
+      "loss": 1.1364,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.90017709814386e-06,
+      "loss": 1.1096,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89986312068377e-06,
+      "loss": 1.1389,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899548662893352e-06,
+      "loss": 1.1581,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899233724852113e-06,
+      "loss": 1.1578,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898918306639673e-06,
+      "loss": 1.1019,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898602408335778e-06,
+      "loss": 1.1443,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898286030020296e-06,
+      "loss": 1.1801,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897969171773213e-06,
+      "loss": 1.1549,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897651833674639e-06,
+      "loss": 1.1051,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897334015804805e-06,
+      "loss": 1.1563,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897015718244062e-06,
+      "loss": 1.1096,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896696941072885e-06,
+      "loss": 1.105,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896377684371866e-06,
+      "loss": 1.1233,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896057948221722e-06,
+      "loss": 1.0996,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89573773270329e-06,
+      "loss": 1.1152,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895417037897526e-06,
+      "loss": 1.1336,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895095863885512e-06,
+      "loss": 1.1257,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8947742107484474e-06,
+      "loss": 1.0857,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894452078567654e-06,
+      "loss": 1.127,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894129467424573e-06,
+      "loss": 1.1443,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89380637740077e-06,
+      "loss": 1.1339,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8934828085779284e-06,
+      "loss": 1.1493,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893158761037855e-06,
+      "loss": 1.1586,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892834234862476e-06,
+      "loss": 1.1197,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892509230133839e-06,
+      "loss": 1.1392,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892183746934115e-06,
+      "loss": 0.6446,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891857785345591e-06,
+      "loss": 1.1733,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8915313454506796e-06,
+      "loss": 1.171,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891204427331912e-06,
+      "loss": 1.1267,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890877031071941e-06,
+      "loss": 0.6962,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890549156753539e-06,
+      "loss": 1.123,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890220804459602e-06,
+      "loss": 1.1199,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889891974273145e-06,
+      "loss": 1.1182,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889562666277303e-06,
+      "loss": 1.1233,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889232880555333e-06,
+      "loss": 1.1065,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888902617190614e-06,
+      "loss": 1.1306,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888571876266644e-06,
+      "loss": 1.0792,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888240657867041e-06,
+      "loss": 1.1487,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887908962075545e-06,
+      "loss": 1.1236,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887576788976018e-06,
+      "loss": 1.1593,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8872441386524395e-06,
+      "loss": 1.1469,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8869110111889115e-06,
+      "loss": 1.1167,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886577406669658e-06,
+      "loss": 1.1049,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886243325179021e-06,
+      "loss": 1.1306,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8859087668014644e-06,
+      "loss": 1.1388,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8855737316215725e-06,
+      "loss": 1.1229,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88523821972405e-06,
+      "loss": 1.136,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884902231193722e-06,
+      "loss": 0.7325,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884565766115536e-06,
+      "loss": 0.7181,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8842288245745555e-06,
+      "loss": 1.1282,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883891406655969e-06,
+      "loss": 1.1549,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883553512445084e-06,
+      "loss": 0.6824,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883215142027327e-06,
+      "loss": 1.0995,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882876295488247e-06,
+      "loss": 1.2102,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882536972913513e-06,
+      "loss": 1.1456,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882197174388912e-06,
+      "loss": 1.1626,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881856900000355e-06,
+      "loss": 1.1448,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881516149833869e-06,
+      "loss": 1.1092,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881174923975606e-06,
+      "loss": 1.1257,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880833222511835e-06,
+      "loss": 1.1289,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880491045528946e-06,
+      "loss": 1.1946,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88014839311345e-06,
+      "loss": 1.1075,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8798052653519765e-06,
+      "loss": 1.1146,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879461662331279e-06,
+      "loss": 1.1441,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879117584138225e-06,
+      "loss": 0.693,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878773030859808e-06,
+      "loss": 0.6806,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878428002583139e-06,
+      "loss": 1.1138,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878082499395447e-06,
+      "loss": 1.1222,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877736521384086e-06,
+      "loss": 1.0877,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877390068636527e-06,
+      "loss": 1.1109,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877043141240361e-06,
+      "loss": 1.1418,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876695739283299e-06,
+      "loss": 1.143,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876347862853174e-06,
+      "loss": 1.1072,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875999512037935e-06,
+      "loss": 1.1454,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8756506869256555e-06,
+      "loss": 1.1639,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875301387604525e-06,
+      "loss": 0.6604,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874951614162856e-06,
+      "loss": 0.7042,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8746013666890785e-06,
+      "loss": 1.1166,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874250645271744e-06,
+      "loss": 1.1418,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873899449999524e-06,
+      "loss": 1.1461,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873547780961207e-06,
+      "loss": 1.1344,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873195638245704e-06,
+      "loss": 1.1449,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872843021942045e-06,
+      "loss": 1.1255,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872489932139381e-06,
+      "loss": 1.1536,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87213636892698e-06,
+      "loss": 1.1325,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871782332394231e-06,
+      "loss": 1.1374,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871427822630644e-06,
+      "loss": 0.6711,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871072839725847e-06,
+      "loss": 1.1729,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870717383769587e-06,
+      "loss": 1.1422,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870361454851733e-06,
+      "loss": 1.1619,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870005053062272e-06,
+      "loss": 1.1545,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869648178491311e-06,
+      "loss": 1.1341,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869290831229075e-06,
+      "loss": 1.0997,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8689330113659105e-06,
+      "loss": 1.134,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868574718992284e-06,
+      "loss": 1.1349,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.86821595419878e-06,
+      "loss": 0.6697,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867856717076101e-06,
+      "loss": 1.1258,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8674970077150725e-06,
+      "loss": 1.1463,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8671368262066356e-06,
+      "loss": 0.7071,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866776172641854e-06,
+      "loss": 0.6918,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866415047111911e-06,
+      "loss": 1.1545,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8660534497081044e-06,
+      "loss": 1.1255,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865691380521857e-06,
+      "loss": 1.127,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865328839644707e-06,
+      "loss": 1.1771,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864965827168314e-06,
+      "loss": 1.1563,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864602343184457e-06,
+      "loss": 1.1353,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864238387785031e-06,
+      "loss": 1.1099,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863873961062055e-06,
+      "loss": 1.086,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863509063107664e-06,
+      "loss": 1.1429,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8631436940141105e-06,
+      "loss": 1.134,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862777853873773e-06,
+      "loss": 1.1432,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.86241154277914e-06,
+      "loss": 1.1198,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862044760822827e-06,
+      "loss": 1.1156,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861677508097562e-06,
+      "loss": 0.674,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861309784696198e-06,
+      "loss": 1.083,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860941590711703e-06,
+      "loss": 1.1424,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860572926237165e-06,
+      "loss": 1.1056,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860203791365792e-06,
+      "loss": 0.6936,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8598341861909085e-06,
+      "loss": 1.1785,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859464110805962e-06,
+      "loss": 1.1016,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.859093565304513e-06,
+      "loss": 1.1799,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858722549780246e-06,
+      "loss": 1.1133,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858351064326962e-06,
+      "loss": 0.6913,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857979109038582e-06,
+      "loss": 1.1394,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857606684009144e-06,
+      "loss": 1.1068,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857233789332807e-06,
+      "loss": 1.1406,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856860425103847e-06,
+      "loss": 0.7001,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85648659141666e-06,
+      "loss": 1.1349,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856112288365758e-06,
+      "loss": 1.1023,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8557375160457765e-06,
+      "loss": 1.0955,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855362274551465e-06,
+      "loss": 1.1441,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854986563977693e-06,
+      "loss": 1.0978,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854610384419451e-06,
+      "loss": 1.1192,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854233735971844e-06,
+      "loss": 1.1333,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853856618730099e-06,
+      "loss": 1.0937,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85347903278956e-06,
+      "loss": 1.1293,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853100978245688e-06,
+      "loss": 1.1444,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8527224551940665e-06,
+      "loss": 1.1261,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852343463730394e-06,
+      "loss": 1.1205,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851964003950488e-06,
+      "loss": 0.6868,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851584075950286e-06,
+      "loss": 1.1175,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851203679825843e-06,
+      "loss": 1.1376,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85082281567333e-06,
+      "loss": 1.0973,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8504414835890405e-06,
+      "loss": 1.1569,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850059683669383e-06,
+      "loss": 0.693,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849677416010887e-06,
+      "loss": 1.1214,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849294680710198e-06,
+      "loss": 1.1396,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84891147786408e-06,
+      "loss": 1.1356,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8485278075694165e-06,
+      "loss": 1.1285,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848143669923208e-06,
+      "loss": 1.1473,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847759065022573e-06,
+      "loss": 1.1127,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8473739929647505e-06,
+      "loss": 1.092,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846988453847094e-06,
+      "loss": 1.1256,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846602447767079e-06,
+      "loss": 1.161,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846215974822295e-06,
+      "loss": 0.6582,
+      "step": 979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8458290351104515e-06,
+      "loss": 1.1382,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845441628729378e-06,
+      "loss": 1.1091,
+      "step": 981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845053755777018e-06,
+      "loss": 1.1452,
+      "step": 982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8446654163514345e-06,
+      "loss": 1.1143,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844276610550811e-06,
+      "loss": 1.1362,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843887338473446e-06,
+      "loss": 1.121,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843497600217756e-06,
+      "loss": 0.7058,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8431073958822775e-06,
+      "loss": 1.0959,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.842716725565662e-06,
+      "loss": 1.1297,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8423255893666806e-06,
+      "loss": 1.1567,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841933987384221e-06,
+      "loss": 1.1401,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841541919717291e-06,
+      "loss": 1.1215,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841149386465014e-06,
+      "loss": 1.1712,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840756387726631e-06,
+      "loss": 1.1081,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840362923601502e-06,
+      "loss": 0.7135,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.839968994189104e-06,
+      "loss": 1.1034,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.839574599589031e-06,
+      "loss": 1.1092,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8391797399009965e-06,
+      "loss": 0.6545,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.83878441522483e-06,
+      "loss": 1.1409,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.838388625660479e-06,
+      "loss": 0.6792,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837992371308007e-06,
+      "loss": 1.1474,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837595652267598e-06,
+      "loss": 1.0822,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8371984686395525e-06,
+      "loss": 1.1113,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836800820524286e-06,
+      "loss": 1.0809,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836402708022334e-06,
+      "loss": 1.0756,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83600413123435e-06,
+      "loss": 1.1104,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835605090261103e-06,
+      "loss": 1.1804,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835205585203479e-06,
+      "loss": 1.1306,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834805616162482e-06,
+      "loss": 1.1855,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834405183239236e-06,
+      "loss": 1.1608,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834004286534978e-06,
+      "loss": 1.1728,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833602926151066e-06,
+      "loss": 1.1303,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8332011021889705e-06,
+      "loss": 1.0975,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832798814750284e-06,
+      "loss": 1.1297,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832396063936715e-06,
+      "loss": 1.1588,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831992849850087e-06,
+      "loss": 1.1739,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831589172592342e-06,
+      "loss": 0.6739,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8311850322655395e-06,
+      "loss": 1.0912,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830780428971857e-06,
+      "loss": 1.1363,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830375362813586e-06,
+      "loss": 1.1009,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829969833893138e-06,
+      "loss": 1.111,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829563842313039e-06,
+      "loss": 0.6908,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829157388175935e-06,
+      "loss": 1.1529,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828750471584586e-06,
+      "loss": 0.6713,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828343092641869e-06,
+      "loss": 1.116,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827935251450782e-06,
+      "loss": 1.1697,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827526948114435e-06,
+      "loss": 1.1293,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8271181827360575e-06,
+      "loss": 1.0985,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826708955418994e-06,
+      "loss": 1.1128,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826299266266708e-06,
+      "loss": 1.1143,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825889115382777e-06,
+      "loss": 1.1172,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825478502870899e-06,
+      "loss": 1.0998,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825067428834885e-06,
+      "loss": 1.0989,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824655893378665e-06,
+      "loss": 1.18,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824243896606284e-06,
+      "loss": 1.1193,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823831438621906e-06,
+      "loss": 1.1262,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8234185195298086e-06,
+      "loss": 1.1256,
+      "step": 1036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823005139434389e-06,
+      "loss": 1.1075,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822591298440158e-06,
+      "loss": 1.1326,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822176996651745e-06,
+      "loss": 0.6547,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8217622341738954e-06,
+      "loss": 1.1168,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821347011111471e-06,
+      "loss": 1.1384,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820931327569449e-06,
+      "loss": 1.1425,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820515183652927e-06,
+      "loss": 1.1179,
+      "step": 1043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8200985794671125e-06,
+      "loss": 1.1507,
+      "step": 1044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819681515117334e-06,
+      "loss": 1.147,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819263990709037e-06,
+      "loss": 1.1341,
+      "step": 1046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818846006347779e-06,
+      "loss": 1.1383,
+      "step": 1047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818427562139238e-06,
+      "loss": 1.1313,
+      "step": 1048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8180086581892055e-06,
+      "loss": 1.094,
+      "step": 1049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817589294603591e-06,
+      "loss": 1.0856,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.81716947148842e-06,
+      "loss": 1.111,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816749188949833e-06,
+      "loss": 1.1525,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8163284470940865e-06,
+      "loss": 0.6418,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815907246027555e-06,
+      "loss": 1.1399,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815485585856729e-06,
+      "loss": 1.1339,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815063466688213e-06,
+      "loss": 1.0941,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814640888628729e-06,
+      "loss": 1.1122,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8142178517851143e-06,
+      "loss": 1.1535,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8137943562643235e-06,
+      "loss": 1.1217,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813370402173425e-06,
+      "loss": 0.6496,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8129459896196057e-06,
+      "loss": 0.6393,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812521118710166e-06,
+      "loss": 1.1319,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812095789552525e-06,
+      "loss": 1.138,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8116700022542137e-06,
+      "loss": 1.1215,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8112437569228834e-06,
+      "loss": 1.1107,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8108170536662976e-06,
+      "loss": 1.1241,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8103898925923375e-06,
+      "loss": 1.1206,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809962273808999e-06,
+      "loss": 1.0888,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809534197424395e-06,
+      "loss": 1.1466,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809105663546752e-06,
+      "loss": 0.7034,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8086766722844156e-06,
+      "loss": 1.1439,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8082472237458435e-06,
+      "loss": 1.1171,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8078173180396104e-06,
+      "loss": 1.1307,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8073869552744077e-06,
+      "loss": 0.6673,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.80695613555904e-06,
+      "loss": 1.093,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8065248590024297e-06,
+      "loss": 1.0995,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8060931257136124e-06,
+      "loss": 1.1427,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805660935801742e-06,
+      "loss": 1.1237,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8052282893760856e-06,
+      "loss": 1.1327,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8047951865460268e-06,
+      "loss": 0.7119,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8043616274210637e-06,
+      "loss": 0.667,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8039276121108105e-06,
+      "loss": 1.1001,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8034931407249964e-06,
+      "loss": 1.1606,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803058213373467e-06,
+      "loss": 1.142,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8026228301661805e-06,
+      "loss": 1.1245,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802186991213214e-06,
+      "loss": 1.1621,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8017506966247563e-06,
+      "loss": 1.126,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801313946511114e-06,
+      "loss": 1.1053,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800876740982707e-06,
+      "loss": 1.1029,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800439080150072e-06,
+      "loss": 1.1616,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.80000096412386e-06,
+      "loss": 1.165,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7995623930148365e-06,
+      "loss": 1.1376,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799123366933883e-06,
+      "loss": 0.6814,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7986838859919955e-06,
+      "loss": 1.1434,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7982439503002853e-06,
+      "loss": 1.1277,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7978035599699787e-06,
+      "loss": 1.1254,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797362715112416e-06,
+      "loss": 1.0744,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796921415839054e-06,
+      "loss": 1.1005,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796479662261464e-06,
+      "loss": 1.1426,
+      "step": 1099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7960374544913306e-06,
+      "loss": 1.1353,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795594792640456e-06,
+      "loss": 0.6823,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7951516768207535e-06,
+      "loss": 1.1603,
+      "step": 1102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794708107144254e-06,
+      "loss": 1.1176,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794264083723104e-06,
+      "loss": 1.1321,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7938196066695607e-06,
+      "loss": 1.1008,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7933746760959998e-06,
+      "loss": 1.0861,
+      "step": 1106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79292929211491e-06,
+      "loss": 0.6461,
+      "step": 1107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792483454838895e-06,
+      "loss": 1.1151,
+      "step": 1108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792037164380673e-06,
+      "loss": 1.0981,
+      "step": 1109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7915904208530768e-06,
+      "loss": 0.6553,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791143224369053e-06,
+      "loss": 1.1332,
+      "step": 1111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790695575041665e-06,
+      "loss": 1.0682,
+      "step": 1112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790247472984087e-06,
+      "loss": 0.6874,
+      "step": 1113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7897989183096124e-06,
+      "loss": 1.078,
+      "step": 1114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7893499111316444e-06,
+      "loss": 1.1576,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7889004515637024e-06,
+      "loss": 1.1108,
+      "step": 1116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7884505397194224e-06,
+      "loss": 0.6594,
+      "step": 1117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788000175712551e-06,
+      "loss": 1.0859,
+      "step": 1118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.787549359656952e-06,
+      "loss": 1.1163,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7870980916666015e-06,
+      "loss": 1.1304,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786646371855592e-06,
+      "loss": 0.6706,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7861942003381262e-06,
+      "loss": 1.0896,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785741577228527e-06,
+      "loss": 1.1458,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785288502641227e-06,
+      "loss": 1.0567,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7848349766907727e-06,
+      "loss": 0.7211,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7843809994918283e-06,
+      "loss": 1.0962,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7839265711591687e-06,
+      "loss": 1.1066,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783471691807685e-06,
+      "loss": 1.1269,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7830163615523807e-06,
+      "loss": 1.1606,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7825605805083745e-06,
+      "loss": 1.1003,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7821043487908987e-06,
+      "loss": 1.1076,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781647666515299e-06,
+      "loss": 1.1016,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781190533797036e-06,
+      "loss": 1.1396,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780732950751683e-06,
+      "loss": 1.155,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7802749174949286e-06,
+      "loss": 0.6781,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779816434142574e-06,
+      "loss": 1.0488,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779357500810535e-06,
+      "loss": 1.1511,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77889811761484e-06,
+      "loss": 0.6319,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7784382846716324e-06,
+      "loss": 0.6753,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777978002097169e-06,
+      "loss": 1.0828,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77751727000782e-06,
+      "loss": 0.6942,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777056088520069e-06,
+      "loss": 1.1131,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776594457750515e-06,
+      "loss": 1.117,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7761323778158674e-06,
+      "loss": 0.6545,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775669848832952e-06,
+      "loss": 1.1007,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7752068709187063e-06,
+      "loss": 1.0984,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7747434441901827e-06,
+      "loss": 0.6844,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7742795687645464e-06,
+      "loss": 1.1082,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7738152447590754e-06,
+      "loss": 1.154,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7733504722911626e-06,
+      "loss": 0.717,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772885251478313e-06,
+      "loss": 1.1137,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7724195824381458e-06,
+      "loss": 1.0917,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771953465288393e-06,
+      "loss": 1.1513,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7714869001469e-06,
+      "loss": 1.0798,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771019887131625e-06,
+      "loss": 1.1352,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770552426360641e-06,
+      "loss": 1.1326,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770084517952132e-06,
+      "loss": 1.1398,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7696161620243974e-06,
+      "loss": 1.1396,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.769147358695848e-06,
+      "loss": 1.1281,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768678108085008e-06,
+      "loss": 0.6746,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768208410310516e-06,
+      "loss": 1.0723,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7677382654911217e-06,
+      "loss": 1.1562,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76726767374569e-06,
+      "loss": 1.1383,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7667966351931966e-06,
+      "loss": 1.1294,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766325149952732e-06,
+      "loss": 1.0613,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765853218143498e-06,
+      "loss": 1.1034,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7653808398848107e-06,
+      "loss": 0.6637,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7649080152960987e-06,
+      "loss": 1.1103,
+      "step": 1168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764434744496902e-06,
+      "loss": 0.6373,
+      "step": 1169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763961027606876e-06,
+      "loss": 1.1712,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763486864745787e-06,
+      "loss": 1.1343,
+      "step": 1171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763012256033514e-06,
+      "loss": 1.084,
+      "step": 1172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.762537201590051e-06,
+      "loss": 1.1219,
+      "step": 1173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7620617015355013e-06,
+      "loss": 1.097,
+      "step": 1174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761585755990083e-06,
+      "loss": 1.1115,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7611093650741267e-06,
+      "loss": 1.1317,
+      "step": 1176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7606325289080755e-06,
+      "loss": 1.133,
+      "step": 1177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7601552476124844e-06,
+      "loss": 1.1316,
+      "step": 1178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7596775213080213e-06,
+      "loss": 1.1493,
+      "step": 1179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.759199350115467e-06,
+      "loss": 1.1351,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758720734155714e-06,
+      "loss": 1.1312,
+      "step": 1181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7582416735497686e-06,
+      "loss": 1.126,
+      "step": 1182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7577621684187488e-06,
+      "loss": 1.1051,
+      "step": 1183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757282218883883e-06,
+      "loss": 1.1297,
+      "step": 1184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7568018250665157e-06,
+      "loss": 0.7027,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7563209870881015e-06,
+      "loss": 0.6764,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7558397050702064e-06,
+      "loss": 1.0903,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755357979134511e-06,
+      "loss": 1.1215,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754875809402807e-06,
+      "loss": 1.1127,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754393195996998e-06,
+      "loss": 1.0976,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7539101390390997e-06,
+      "loss": 1.0946,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753426638651241e-06,
+      "loss": 1.1442,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752942694955662e-06,
+      "loss": 1.0978,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752458308074715e-06,
+      "loss": 1.1297,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751973478130865e-06,
+      "loss": 1.1176,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7514882052466876e-06,
+      "loss": 0.7022,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751002489544872e-06,
+      "loss": 1.133,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7505163311482176e-06,
+      "loss": 1.1007,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750029730179638e-06,
+      "loss": 1.1309,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.749542686762157e-06,
+      "loss": 1.1073,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7490552010189107e-06,
+      "loss": 1.0918,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748567273073146e-06,
+      "loss": 1.1106,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748078903048224e-06,
+      "loss": 1.0832,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.747590091067616e-06,
+      "loss": 0.6861,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.747100837254905e-06,
+      "loss": 1.0977,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.746611141733785e-06,
+      "loss": 1.1215,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7461210046280643e-06,
+      "loss": 1.0796,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7456304260616604e-06,
+      "loss": 1.1115,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7451394061586026e-06,
+      "loss": 0.6513,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744647945043033e-06,
+      "loss": 1.1018,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7441560428392043e-06,
+      "loss": 1.1036,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743663699671481e-06,
+      "loss": 1.1112,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74317091566434e-06,
+      "loss": 1.1188,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7426776909423677e-06,
+      "loss": 1.1076,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742184025630263e-06,
+      "loss": 1.1465,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7416899198528365e-06,
+      "loss": 1.1319,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7411953737350094e-06,
+      "loss": 1.1268,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7407003874018158e-06,
+      "loss": 1.1384,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7402049609783984e-06,
+      "loss": 1.0908,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739709094590013e-06,
+      "loss": 1.1545,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739212788362028e-06,
+      "loss": 1.1574,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.73871604241992e-06,
+      "loss": 0.6845,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738218856889278e-06,
+      "loss": 0.6797,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7377212318958024e-06,
+      "loss": 1.1032,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7372231675653054e-06,
+      "loss": 1.1255,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736724664023708e-06,
+      "loss": 1.1535,
+      "step": 1226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736225721397045e-06,
+      "loss": 1.1583,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7357263398114597e-06,
+      "loss": 1.1516,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7352265193932088e-06,
+      "loss": 1.0734,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7347262602686576e-06,
+      "loss": 1.1298,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7342255625642844e-06,
+      "loss": 1.132,
+      "step": 1231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.733724426406677e-06,
+      "loss": 1.096,
+      "step": 1232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.733222851922534e-06,
+      "loss": 1.0942,
+      "step": 1233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7327208392386665e-06,
+      "loss": 1.08,
+      "step": 1234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7322183884819937e-06,
+      "loss": 1.0994,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7317154997795488e-06,
+      "loss": 1.1148,
+      "step": 1236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7312121732584716e-06,
+      "loss": 1.123,
+      "step": 1237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7307084090460172e-06,
+      "loss": 0.7036,
+      "step": 1238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7302042072695475e-06,
+      "loss": 1.1487,
+      "step": 1239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729699568056538e-06,
+      "loss": 1.1241,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7291944915345725e-06,
+      "loss": 1.1231,
+      "step": 1241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7286889778313464e-06,
+      "loss": 1.1087,
+      "step": 1242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7281830270746656e-06,
+      "loss": 1.1275,
+      "step": 1243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7276766393924464e-06,
+      "loss": 1.0964,
+      "step": 1244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7271698149127157e-06,
+      "loss": 1.1428,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.726662553763611e-06,
+      "loss": 1.0865,
+      "step": 1246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7261548560733795e-06,
+      "loss": 1.0943,
+      "step": 1247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7256467219703787e-06,
+      "loss": 1.1363,
+      "step": 1248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7251381515830777e-06,
+      "loss": 1.092,
+      "step": 1249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7246291450400554e-06,
+      "loss": 0.6792,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72411970247e-06,
+      "loss": 1.1489,
+      "step": 1251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.723609824001711e-06,
+      "loss": 1.1288,
+      "step": 1252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7230995097640968e-06,
+      "loss": 1.083,
+      "step": 1253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722588759886178e-06,
+      "loss": 1.0788,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722077574497084e-06,
+      "loss": 1.1272,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7215659537260543e-06,
+      "loss": 1.1124,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7210538977024387e-06,
+      "loss": 1.1104,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7205414065556984e-06,
+      "loss": 1.0997,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7200284804154006e-06,
+      "loss": 1.1094,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7195151194112277e-06,
+      "loss": 1.1443,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7190013236729682e-06,
+      "loss": 1.1509,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7184870933305226e-06,
+      "loss": 1.0725,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7179724285138993e-06,
+      "loss": 1.1166,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7174573293532188e-06,
+      "loss": 1.1217,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7169417959787098e-06,
+      "loss": 1.1303,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7164258285207116e-06,
+      "loss": 1.1236,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715909427109673e-06,
+      "loss": 1.1062,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7153925918761518e-06,
+      "loss": 1.1464,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7148753229508176e-06,
+      "loss": 1.0542,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7143576204644468e-06,
+      "loss": 1.1334,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713839484547928e-06,
+      "loss": 1.1413,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7133209153322574e-06,
+      "loss": 1.0848,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7128019129485414e-06,
+      "loss": 1.1177,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7122824775279966e-06,
+      "loss": 1.1288,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7117626092019483e-06,
+      "loss": 1.0789,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7112423081018323e-06,
+      "loss": 0.6675,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7107215743591915e-06,
+      "loss": 1.0973,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710200408105681e-06,
+      "loss": 1.1275,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709678809473063e-06,
+      "loss": 1.1511,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709156778593211e-06,
+      "loss": 1.1298,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7086343155981065e-06,
+      "loss": 1.0748,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7081114206198397e-06,
+      "loss": 1.1061,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7075880937906114e-06,
+      "loss": 1.1425,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7070643352427312e-06,
+      "loss": 1.1304,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706540145108617e-06,
+      "loss": 1.1175,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706015523520797e-06,
+      "loss": 1.1549,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7054904706119075e-06,
+      "loss": 0.6833,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7049649865146953e-06,
+      "loss": 1.1302,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7044390713620143e-06,
+      "loss": 1.151,
+      "step": 1289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703912725286828e-06,
+      "loss": 0.6997,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7033859484222098e-06,
+      "loss": 1.1346,
+      "step": 1291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7028587409013413e-06,
+      "loss": 1.127,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702331102857513e-06,
+      "loss": 1.1088,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701803034424124e-06,
+      "loss": 1.1025,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701274535734683e-06,
+      "loss": 1.1447,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.700745606922806e-06,
+      "loss": 1.1442,
+      "step": 1296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7002162481222195e-06,
+      "loss": 1.1032,
+      "step": 1297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699686459466758e-06,
+      "loss": 1.0892,
+      "step": 1298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699156241090364e-06,
+      "loss": 1.1075,
+      "step": 1299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69862559312709e-06,
+      "loss": 1.1352,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.698094515711096e-06,
+      "loss": 0.6644,
+      "step": 1301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6975630089766506e-06,
+      "loss": 1.0861,
+      "step": 1302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6970310730581318e-06,
+      "loss": 1.12,
+      "step": 1303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696498708090025e-06,
+      "loss": 1.1113,
+      "step": 1304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695965914206925e-06,
+      "loss": 1.0917,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695432691543535e-06,
+      "loss": 1.127,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.694899040234666e-06,
+      "loss": 1.141,
+      "step": 1307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6943649604152373e-06,
+      "loss": 1.1334,
+      "step": 1308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6938304522202777e-06,
+      "loss": 1.1208,
+      "step": 1309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.693295515784923e-06,
+      "loss": 1.1276,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6927601512444177e-06,
+      "loss": 1.1128,
+      "step": 1311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6922243587341146e-06,
+      "loss": 1.1145,
+      "step": 1312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6916881383894748e-06,
+      "loss": 1.1762,
+      "step": 1313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691151490346068e-06,
+      "loss": 1.1039,
+      "step": 1314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6906144147395706e-06,
+      "loss": 1.0908,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690076911705769e-06,
+      "loss": 1.1262,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689538981380556e-06,
+      "loss": 1.0843,
+      "step": 1317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6890006238999335e-06,
+      "loss": 1.1126,
+      "step": 1318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6884618394000106e-06,
+      "loss": 1.1011,
+      "step": 1319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687922628017005e-06,
+      "loss": 1.1259,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6873829898872417e-06,
+      "loss": 1.1412,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.686842925147155e-06,
+      "loss": 1.1136,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6863024339332846e-06,
+      "loss": 1.1346,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685761516382281e-06,
+      "loss": 1.1295,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6852201726309e-06,
+      "loss": 0.6764,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6846784028160056e-06,
+      "loss": 1.0924,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6841362070745713e-06,
+      "loss": 1.147,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6835935855436762e-06,
+      "loss": 1.1166,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.683050538360508e-06,
+      "loss": 1.1128,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6825070656623624e-06,
+      "loss": 1.1001,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681963167586641e-06,
+      "loss": 1.1292,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681418844270855e-06,
+      "loss": 0.6527,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6808740958526224e-06,
+      "loss": 1.0753,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6803289224696677e-06,
+      "loss": 0.6959,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6797833242598244e-06,
+      "loss": 0.6742,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6792373013610317e-06,
+      "loss": 1.1031,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6786908539113374e-06,
+      "loss": 1.1326,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.678143982048897e-06,
+      "loss": 1.1268,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6775966859119723e-06,
+      "loss": 1.0734,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6770489656389324e-06,
+      "loss": 1.1447,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.676500821368254e-06,
+      "loss": 1.139,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.675952253238521e-06,
+      "loss": 1.0576,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6754032613884255e-06,
+      "loss": 1.1093,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.674853845956764e-06,
+      "loss": 0.6682,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6743040070824417e-06,
+      "loss": 1.1365,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6737537449044717e-06,
+      "loss": 1.1085,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6732030595619738e-06,
+      "loss": 1.1137,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.672651951194173e-06,
+      "loss": 1.0959,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6721004199404026e-06,
+      "loss": 1.1144,
+      "step": 1349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.671548465940104e-06,
+      "loss": 1.0839,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670996089332823e-06,
+      "loss": 1.1071,
+      "step": 1351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6704432902582143e-06,
+      "loss": 1.1279,
+      "step": 1352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669890068856038e-06,
+      "loss": 1.0844,
+      "step": 1353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669336425266162e-06,
+      "loss": 0.7105,
+      "step": 1354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.66878235962856e-06,
+      "loss": 1.1091,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6682278720833127e-06,
+      "loss": 1.0902,
+      "step": 1356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6676729627706086e-06,
+      "loss": 1.1231,
+      "step": 1357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6671176318307417e-06,
+      "loss": 1.095,
+      "step": 1358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666561879404112e-06,
+      "loss": 1.1041,
+      "step": 1359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666005705631227e-06,
+      "loss": 0.6581,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6654491106527017e-06,
+      "loss": 1.148,
+      "step": 1361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6648920946092553e-06,
+      "loss": 1.1221,
+      "step": 1362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664334657641715e-06,
+      "loss": 1.1247,
+      "step": 1363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6637767998910145e-06,
+      "loss": 1.0995,
+      "step": 1364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6632185214981927e-06,
+      "loss": 1.1298,
+      "step": 1365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6626598226043955e-06,
+      "loss": 1.1427,
+      "step": 1366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662100703350877e-06,
+      "loss": 1.0852,
+      "step": 1367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6615411638789927e-06,
+      "loss": 1.1245,
+      "step": 1368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.66098120433021e-06,
+      "loss": 1.1102,
+      "step": 1369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.660420824846098e-06,
+      "loss": 1.1263,
+      "step": 1370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659860025568336e-06,
+      "loss": 1.1078,
+      "step": 1371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6592988066387056e-06,
+      "loss": 1.0621,
+      "step": 1372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658737168199097e-06,
+      "loss": 1.1159,
+      "step": 1373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6581751103915055e-06,
+      "loss": 1.092,
+      "step": 1374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6576126333580323e-06,
+      "loss": 1.1271,
+      "step": 1375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6570497372408858e-06,
+      "loss": 1.0916,
+      "step": 1376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6564864221823778e-06,
+      "loss": 1.1054,
+      "step": 1377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6559226883249296e-06,
+      "loss": 1.1281,
+      "step": 1378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655358535811065e-06,
+      "loss": 1.0818,
+      "step": 1379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6547939647834164e-06,
+      "loss": 1.1261,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6542289753847193e-06,
+      "loss": 0.6601,
+      "step": 1381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6536635677578174e-06,
+      "loss": 1.1001,
+      "step": 1382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6530977420456583e-06,
+      "loss": 1.1289,
+      "step": 1383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6525314983912973e-06,
+      "loss": 1.1096,
+      "step": 1384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6519648369378928e-06,
+      "loss": 1.0754,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.651397757828711e-06,
+      "loss": 1.0983,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6508302612071235e-06,
+      "loss": 1.1387,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6502623472166053e-06,
+      "loss": 0.7116,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6496940160007393e-06,
+      "loss": 1.1185,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6491252677032137e-06,
+      "loss": 1.0753,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6485561024678204e-06,
+      "loss": 1.1217,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647986520438459e-06,
+      "loss": 1.118,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647416521759132e-06,
+      "loss": 1.1196,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6468461065739497e-06,
+      "loss": 1.1064,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6462752750271253e-06,
+      "loss": 1.0833,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6457040272629804e-06,
+      "loss": 1.1056,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6451323634259384e-06,
+      "loss": 1.0944,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6445602836605297e-06,
+      "loss": 1.0681,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6439877881113894e-06,
+      "loss": 1.0965,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643414876923259e-06,
+      "loss": 1.1479,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6428415502409832e-06,
+      "loss": 1.0927,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642267808209512e-06,
+      "loss": 1.1475,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6416936509739022e-06,
+      "loss": 1.0728,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6411190786793134e-06,
+      "loss": 1.0873,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6405440914710114e-06,
+      "loss": 1.1071,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6399686894943662e-06,
+      "loss": 1.1031,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639392872894854e-06,
+      "loss": 1.1316,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6388166418180537e-06,
+      "loss": 1.0811,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.638239996409651e-06,
+      "loss": 1.1296,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6376629368154345e-06,
+      "loss": 1.0967,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6370854631813e-06,
+      "loss": 1.108,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6365075756532452e-06,
+      "loss": 0.6719,
+      "step": 1412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6359292743773742e-06,
+      "loss": 1.1276,
+      "step": 1413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.635350559499895e-06,
+      "loss": 1.1337,
+      "step": 1414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634771431167121e-06,
+      "loss": 1.1221,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634191889525469e-06,
+      "loss": 1.1173,
+      "step": 1416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6336119347214613e-06,
+      "loss": 1.1215,
+      "step": 1417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6330315669017235e-06,
+      "loss": 1.1023,
+      "step": 1418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6324507862129864e-06,
+      "loss": 1.1168,
+      "step": 1419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6318695928020866e-06,
+      "loss": 0.6874,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6312879868159605e-06,
+      "loss": 1.1297,
+      "step": 1421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6307059684016553e-06,
+      "loss": 0.6542,
+      "step": 1422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630123537706317e-06,
+      "loss": 1.1533,
+      "step": 1423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6295406948771977e-06,
+      "loss": 1.1287,
+      "step": 1424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6289574400616544e-06,
+      "loss": 1.1399,
+      "step": 1425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6283737734071483e-06,
+      "loss": 1.0965,
+      "step": 1426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627789695061243e-06,
+      "loss": 1.1059,
+      "step": 1427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6272052051716084e-06,
+      "loss": 1.1241,
+      "step": 1428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6266203038860166e-06,
+      "loss": 1.1322,
+      "step": 1429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6260349913523446e-06,
+      "loss": 1.1118,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625449267718573e-06,
+      "loss": 1.0983,
+      "step": 1431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.624863133132788e-06,
+      "loss": 1.1008,
+      "step": 1432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6242765877431758e-06,
+      "loss": 1.12,
+      "step": 1433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6236896316980313e-06,
+      "loss": 1.0583,
+      "step": 1434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62310226514575e-06,
+      "loss": 1.1127,
+      "step": 1435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6225144882348315e-06,
+      "loss": 1.1072,
+      "step": 1436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6219263011138806e-06,
+      "loss": 0.6896,
+      "step": 1437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.621337703931604e-06,
+      "loss": 1.1175,
+      "step": 1438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620748696836814e-06,
+      "loss": 1.1149,
+      "step": 1439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6201592799784246e-06,
+      "loss": 1.1373,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619569453505455e-06,
+      "loss": 1.1409,
+      "step": 1441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618979217567027e-06,
+      "loss": 1.0563,
+      "step": 1442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6183885723123667e-06,
+      "loss": 0.6325,
+      "step": 1443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6177975178908014e-06,
+      "loss": 1.1062,
+      "step": 1444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617206054451766e-06,
+      "loss": 1.0757,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6166141821447954e-06,
+      "loss": 1.079,
+      "step": 1446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6160219011195285e-06,
+      "loss": 0.6485,
+      "step": 1447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6154292115257084e-06,
+      "loss": 1.0476,
+      "step": 1448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6148361135131814e-06,
+      "loss": 1.1076,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.614242607231896e-06,
+      "loss": 1.0908,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613648692831905e-06,
+      "loss": 1.0944,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613054370463364e-06,
+      "loss": 1.1537,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6124596402765315e-06,
+      "loss": 1.0907,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6118645024217693e-06,
+      "loss": 1.14,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6112689570495434e-06,
+      "loss": 1.0788,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.61067300431042e-06,
+      "loss": 1.0586,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6100766443550715e-06,
+      "loss": 1.1124,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609479877334271e-06,
+      "loss": 1.1083,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6088827033988957e-06,
+      "loss": 1.1075,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.608285122699925e-06,
+      "loss": 0.6523,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607687135388442e-06,
+      "loss": 1.0792,
+      "step": 1461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6070887416156315e-06,
+      "loss": 1.098,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6064899415327815e-06,
+      "loss": 1.0963,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6058907352912832e-06,
+      "loss": 1.1258,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6052911230426304e-06,
+      "loss": 1.1053,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.604691104938418e-06,
+      "loss": 1.0791,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.604090681130346e-06,
+      "loss": 1.1016,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6034898517702156e-06,
+      "loss": 1.1391,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.60288861700993e-06,
+      "loss": 1.0845,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6022869770014957e-06,
+      "loss": 1.1292,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6016849318970225e-06,
+      "loss": 1.0947,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6010824818487207e-06,
+      "loss": 1.1114,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.600479627008904e-06,
+      "loss": 1.0734,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5998763675299885e-06,
+      "loss": 1.1186,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5992727035644926e-06,
+      "loss": 1.0916,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598668635265037e-06,
+      "loss": 1.0924,
+      "step": 1476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598064162784343e-06,
+      "loss": 1.1114,
+      "step": 1477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5974592862752376e-06,
+      "loss": 1.1231,
+      "step": 1478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.596854005890647e-06,
+      "loss": 1.1147,
+      "step": 1479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5962483217836e-06,
+      "loss": 1.1065,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5956422341072285e-06,
+      "loss": 0.6943,
+      "step": 1481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595035743014765e-06,
+      "loss": 1.1044,
+      "step": 1482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5944288486595463e-06,
+      "loss": 1.1094,
+      "step": 1483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.593821551195008e-06,
+      "loss": 1.0969,
+      "step": 1484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.59321385077469e-06,
+      "loss": 1.1343,
+      "step": 1485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592605747552233e-06,
+      "loss": 1.101,
+      "step": 1486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5919972416813805e-06,
+      "loss": 1.0749,
+      "step": 1487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5913883333159767e-06,
+      "loss": 1.1005,
+      "step": 1488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5907790226099682e-06,
+      "loss": 1.1346,
+      "step": 1489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5901693097174033e-06,
+      "loss": 0.6627,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5895591947924314e-06,
+      "loss": 1.1099,
+      "step": 1491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5889486779893043e-06,
+      "loss": 1.1122,
+      "step": 1492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.588337759462375e-06,
+      "loss": 1.1499,
+      "step": 1493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587726439366098e-06,
+      "loss": 1.1175,
+      "step": 1494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.58711471785503e-06,
+      "loss": 1.1149,
+      "step": 1495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5865025950838275e-06,
+      "loss": 1.0874,
+      "step": 1496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5858900712072515e-06,
+      "loss": 1.1382,
+      "step": 1497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5852771463801608e-06,
+      "loss": 1.0823,
+      "step": 1498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.584663820757518e-06,
+      "loss": 1.1088,
+      "step": 1499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5840500944943857e-06,
+      "loss": 1.1301,
+      "step": 1500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5834359677459302e-06,
+      "loss": 1.094,
+      "step": 1501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5828214406674155e-06,
+      "loss": 1.1,
+      "step": 1502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.582206513414209e-06,
+      "loss": 0.694,
+      "step": 1503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5815911861417786e-06,
+      "loss": 1.1208,
+      "step": 1504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5809754590056943e-06,
+      "loss": 1.0776,
+      "step": 1505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580359332161626e-06,
+      "loss": 1.0638,
+      "step": 1506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5797428057653455e-06,
+      "loss": 1.0914,
+      "step": 1507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5791258799727247e-06,
+      "loss": 1.1177,
+      "step": 1508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5785085549397378e-06,
+      "loss": 1.0808,
+      "step": 1509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5778908308224576e-06,
+      "loss": 1.1599,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5772727077770615e-06,
+      "loss": 1.1213,
+      "step": 1511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576654185959825e-06,
+      "loss": 1.0995,
+      "step": 1512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576035265527124e-06,
+      "loss": 1.1189,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5754159466354366e-06,
+      "loss": 1.0681,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5747962294413413e-06,
+      "loss": 1.0988,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574176114101518e-06,
+      "loss": 1.127,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5735556007727463e-06,
+      "loss": 1.0909,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572934689611906e-06,
+      "loss": 0.7121,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5723133807759782e-06,
+      "loss": 1.1241,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5716916744220457e-06,
+      "loss": 1.1153,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.571069570707289e-06,
+      "loss": 1.1261,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570447069788991e-06,
+      "loss": 1.0927,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569824171824535e-06,
+      "loss": 1.1221,
+      "step": 1523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5692008769714043e-06,
+      "loss": 1.1046,
+      "step": 1524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.568577185387183e-06,
+      "loss": 1.0776,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5679530972295544e-06,
+      "loss": 1.12,
+      "step": 1526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.567328612656302e-06,
+      "loss": 1.0822,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5667037318253122e-06,
+      "loss": 1.1123,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.566078454894568e-06,
+      "loss": 1.0852,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5654527820221554e-06,
+      "loss": 1.1462,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5648267133662583e-06,
+      "loss": 0.6891,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5642002490851624e-06,
+      "loss": 1.0841,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.563573389337253e-06,
+      "loss": 1.0795,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5629461342810134e-06,
+      "loss": 1.089,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.56231848407503e-06,
+      "loss": 1.0947,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5616904388779877e-06,
+      "loss": 1.1143,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.56106199884867e-06,
+      "loss": 0.6667,
+      "step": 1537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5604331641459623e-06,
+      "loss": 1.1311,
+      "step": 1538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5598039349288486e-06,
+      "loss": 1.1096,
+      "step": 1539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.559174311356413e-06,
+      "loss": 1.1053,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.558544293587839e-06,
+      "loss": 1.115,
+      "step": 1541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5579138817824104e-06,
+      "loss": 0.661,
+      "step": 1542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.55728307609951e-06,
+      "loss": 1.0965,
+      "step": 1543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5566518766986206e-06,
+      "loss": 1.1023,
+      "step": 1544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5560202837393246e-06,
+      "loss": 1.1284,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555388297381302e-06,
+      "loss": 0.6857,
+      "step": 1546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5547559177843355e-06,
+      "loss": 1.1075,
+      "step": 1547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5541231451083054e-06,
+      "loss": 0.6483,
+      "step": 1548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5534899795131915e-06,
+      "loss": 1.1361,
+      "step": 1549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552856421159073e-06,
+      "loss": 1.1206,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552222470206128e-06,
+      "loss": 1.0979,
+      "step": 1551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5515881268146347e-06,
+      "loss": 1.1425,
+      "step": 1552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.55095339114497e-06,
+      "loss": 1.0816,
+      "step": 1553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.550318263357611e-06,
+      "loss": 1.133,
+      "step": 1554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.549682743613131e-06,
+      "loss": 0.6807,
+      "step": 1555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5490468320722065e-06,
+      "loss": 1.0876,
+      "step": 1556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5484105288956093e-06,
+      "loss": 1.1372,
+      "step": 1557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5477738342442135e-06,
+      "loss": 1.1124,
+      "step": 1558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.547136748278989e-06,
+      "loss": 1.1111,
+      "step": 1559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5464992711610077e-06,
+      "loss": 1.0953,
+      "step": 1560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545861403051438e-06,
+      "loss": 0.6689,
+      "step": 1561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5452231441115475e-06,
+      "loss": 1.1313,
+      "step": 1562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.544584494502704e-06,
+      "loss": 1.1232,
+      "step": 1563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5439454543863736e-06,
+      "loss": 1.08,
+      "step": 1564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5433060239241197e-06,
+      "loss": 1.0698,
+      "step": 1565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5426662032776056e-06,
+      "loss": 1.1166,
+      "step": 1566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542025992608594e-06,
+      "loss": 1.089,
+      "step": 1567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5413853920789448e-06,
+      "loss": 1.1237,
+      "step": 1568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5407444018506166e-06,
+      "loss": 1.1028,
+      "step": 1569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540103022085667e-06,
+      "loss": 1.1278,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5394612529462525e-06,
+      "loss": 1.1074,
+      "step": 1571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.538819094594627e-06,
+      "loss": 0.6516,
+      "step": 1572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.538176547193144e-06,
+      "loss": 0.6425,
+      "step": 1573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5375336109042536e-06,
+      "loss": 1.1123,
+      "step": 1574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.536890285890506e-06,
+      "loss": 1.0742,
+      "step": 1575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5362465723145487e-06,
+      "loss": 1.1089,
+      "step": 1576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5356024703391278e-06,
+      "loss": 1.1289,
+      "step": 1577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.534957980127088e-06,
+      "loss": 1.1066,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5343131018413707e-06,
+      "loss": 1.0826,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5336678356450173e-06,
+      "loss": 1.1293,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533022181701166e-06,
+      "loss": 1.1046,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5323761401730528e-06,
+      "loss": 1.1201,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.531729711224014e-06,
+      "loss": 1.1158,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5310828950174796e-06,
+      "loss": 1.108,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530435691716982e-06,
+      "loss": 1.115,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.529788101486149e-06,
+      "loss": 1.1353,
+      "step": 1586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5291401244887063e-06,
+      "loss": 0.7144,
+      "step": 1587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528491760888478e-06,
+      "loss": 1.079,
+      "step": 1588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5278430108493867e-06,
+      "loss": 1.0931,
+      "step": 1589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.52719387453545e-06,
+      "loss": 1.0769,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526544352110787e-06,
+      "loss": 1.1248,
+      "step": 1591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.52589444373961e-06,
+      "loss": 1.1327,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525244149586233e-06,
+      "loss": 1.1251,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5245934698150643e-06,
+      "loss": 1.065,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5239424045906127e-06,
+      "loss": 1.1018,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523290954077482e-06,
+      "loss": 0.7129,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5226391184403747e-06,
+      "loss": 1.133,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521986897844089e-06,
+      "loss": 1.0938,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5213342924535234e-06,
+      "loss": 1.1221,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5206813024336713e-06,
+      "loss": 1.1225,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520027927949624e-06,
+      "loss": 1.0876,
+      "step": 1601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.51937416916657e-06,
+      "loss": 1.1099,
+      "step": 1602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5187200262497946e-06,
+      "loss": 1.1157,
+      "step": 1603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518065499364681e-06,
+      "loss": 1.1091,
+      "step": 1604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5174105886767097e-06,
+      "loss": 1.0853,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.516755294351456e-06,
+      "loss": 1.1001,
+      "step": 1606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5160996165545952e-06,
+      "loss": 1.0855,
+      "step": 1607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5154435554518984e-06,
+      "loss": 1.0838,
+      "step": 1608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514787111209232e-06,
+      "loss": 1.0769,
+      "step": 1609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5141302839925615e-06,
+      "loss": 1.0903,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5134730739679477e-06,
+      "loss": 1.1028,
+      "step": 1611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.51281548130155e-06,
+      "loss": 1.1398,
+      "step": 1612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.512157506159622e-06,
+      "loss": 1.0905,
+      "step": 1613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5114991487085164e-06,
+      "loss": 0.7138,
+      "step": 1614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5108404091146813e-06,
+      "loss": 1.0419,
+      "step": 1615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510181287544661e-06,
+      "loss": 1.138,
+      "step": 1616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5095217841650984e-06,
+      "loss": 1.1262,
+      "step": 1617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.50886189914273e-06,
+      "loss": 1.0946,
+      "step": 1618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5082016326443915e-06,
+      "loss": 1.0786,
+      "step": 1619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5075409848370133e-06,
+      "loss": 1.1025,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506879955887623e-06,
+      "loss": 1.1028,
+      "step": 1621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5062185459633435e-06,
+      "loss": 0.6617,
+      "step": 1622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505556755231396e-06,
+      "loss": 0.6393,
+      "step": 1623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5048945838590973e-06,
+      "loss": 1.1002,
+      "step": 1624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.504232032013858e-06,
+      "loss": 1.1311,
+      "step": 1625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5035690998631884e-06,
+      "loss": 1.0802,
+      "step": 1626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5029057875746935e-06,
+      "loss": 1.0569,
+      "step": 1627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5022420953160738e-06,
+      "loss": 0.646,
+      "step": 1628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.501578023255126e-06,
+      "loss": 1.1009,
+      "step": 1629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5009135715597444e-06,
+      "loss": 1.1335,
+      "step": 1630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500248740397917e-06,
+      "loss": 1.0821,
+      "step": 1631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4995835299377296e-06,
+      "loss": 1.072,
+      "step": 1632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4989179403473622e-06,
+      "loss": 1.0884,
+      "step": 1633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498251971795093e-06,
+      "loss": 1.0998,
+      "step": 1634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497585624449293e-06,
+      "loss": 1.0957,
+      "step": 1635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496918898478432e-06,
+      "loss": 1.0849,
+      "step": 1636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496251794051073e-06,
+      "loss": 1.0769,
+      "step": 1637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.495584311335876e-06,
+      "loss": 1.0762,
+      "step": 1638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4949164505015968e-06,
+      "loss": 1.0777,
+      "step": 1639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4942482117170867e-06,
+      "loss": 1.0804,
+      "step": 1640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493579595151291e-06,
+      "loss": 1.0936,
+      "step": 1641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4929106009732535e-06,
+      "loss": 1.1061,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4922412293521105e-06,
+      "loss": 1.1261,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4915714804570954e-06,
+      "loss": 1.0744,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4909013544575362e-06,
+      "loss": 1.1458,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4902308515228568e-06,
+      "loss": 1.1242,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4895599718225767e-06,
+      "loss": 1.1238,
+      "step": 1647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4888887155263094e-06,
+      "loss": 1.085,
+      "step": 1648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.488217082803765e-06,
+      "loss": 0.6695,
+      "step": 1649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.487545073824748e-06,
+      "loss": 1.104,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486872688759157e-06,
+      "loss": 1.1035,
+      "step": 1651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4861999277769884e-06,
+      "loss": 1.0525,
+      "step": 1652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4855267910483315e-06,
+      "loss": 1.147,
+      "step": 1653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4848532787433714e-06,
+      "loss": 0.6657,
+      "step": 1654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.484179391032388e-06,
+      "loss": 1.1355,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483505128085755e-06,
+      "loss": 1.1198,
+      "step": 1656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4828304900739436e-06,
+      "loss": 1.0815,
+      "step": 1657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482155477167517e-06,
+      "loss": 1.1034,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.481480089537135e-06,
+      "loss": 1.0736,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480804327353551e-06,
+      "loss": 1.1408,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480128190787615e-06,
+      "loss": 1.088,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4794516800102694e-06,
+      "loss": 1.1175,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.478774795192551e-06,
+      "loss": 1.1054,
+      "step": 1663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4780975365055946e-06,
+      "loss": 1.1066,
+      "step": 1664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4774199041206257e-06,
+      "loss": 1.1054,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4767418982089662e-06,
+      "loss": 1.1044,
+      "step": 1666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.476063518942031e-06,
+      "loss": 1.1025,
+      "step": 1667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4753847664913327e-06,
+      "loss": 1.1107,
+      "step": 1668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4747056410284743e-06,
+      "loss": 0.7466,
+      "step": 1669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474026142725155e-06,
+      "loss": 1.0987,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4733462717531687e-06,
+      "loss": 1.1037,
+      "step": 1671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4726660282844017e-06,
+      "loss": 1.126,
+      "step": 1672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.471985412490836e-06,
+      "loss": 1.0736,
+      "step": 1673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4713044245445482e-06,
+      "loss": 1.1096,
+      "step": 1674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4706230646177073e-06,
+      "loss": 1.0741,
+      "step": 1675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.469941332882578e-06,
+      "loss": 1.0983,
+      "step": 1676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4692592295115175e-06,
+      "loss": 1.1035,
+      "step": 1677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4685767546769774e-06,
+      "loss": 1.0746,
+      "step": 1678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4678939085515042e-06,
+      "loss": 1.107,
+      "step": 1679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.467210691307737e-06,
+      "loss": 1.0961,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46652710311841e-06,
+      "loss": 1.1316,
+      "step": 1681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46584314415635e-06,
+      "loss": 1.1111,
+      "step": 1682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465158814594478e-06,
+      "loss": 1.1065,
+      "step": 1683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464474114605809e-06,
+      "loss": 1.106,
+      "step": 1684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4637890443634507e-06,
+      "loss": 1.1106,
+      "step": 1685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4631036040406058e-06,
+      "loss": 1.123,
+      "step": 1686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.462417793810569e-06,
+      "loss": 1.0664,
+      "step": 1687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.46173161384673e-06,
+      "loss": 1.0693,
+      "step": 1688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.461045064322571e-06,
+      "loss": 1.0975,
+      "step": 1689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4603581454116682e-06,
+      "loss": 1.0995,
+      "step": 1690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4596708572876915e-06,
+      "loss": 1.0901,
+      "step": 1691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4589832001244023e-06,
+      "loss": 1.1038,
+      "step": 1692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4582951740956563e-06,
+      "loss": 1.0695,
+      "step": 1693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4576067793754053e-06,
+      "loss": 1.0988,
+      "step": 1694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456918016137689e-06,
+      "loss": 1.0694,
+      "step": 1695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456228884556645e-06,
+      "loss": 1.0668,
+      "step": 1696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4555393848065002e-06,
+      "loss": 1.0748,
+      "step": 1697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4548495170615783e-06,
+      "loss": 0.6815,
+      "step": 1698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.454159281496293e-06,
+      "loss": 1.1093,
+      "step": 1699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4534686782851525e-06,
+      "loss": 1.1213,
+      "step": 1700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452777707602758e-06,
+      "loss": 1.0409,
+      "step": 1701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452086369623802e-06,
+      "loss": 1.0952,
+      "step": 1702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4513946645230727e-06,
+      "loss": 1.1043,
+      "step": 1703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.450702592475449e-06,
+      "loss": 1.0759,
+      "step": 1704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4500101536559024e-06,
+      "loss": 1.1014,
+      "step": 1705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.449317348239498e-06,
+      "loss": 1.0951,
+      "step": 1706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4486241764013944e-06,
+      "loss": 1.0692,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4479306383168403e-06,
+      "loss": 1.0775,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44723673416118e-06,
+      "loss": 1.1028,
+      "step": 1709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.446542464109848e-06,
+      "loss": 1.1168,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4458478283383727e-06,
+      "loss": 1.1234,
+      "step": 1711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4451528270223733e-06,
+      "loss": 1.1077,
+      "step": 1712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.444457460337564e-06,
+      "loss": 1.136,
+      "step": 1713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4437617284597488e-06,
+      "loss": 1.0762,
+      "step": 1714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.443065631564826e-06,
+      "loss": 1.0731,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4423691698287855e-06,
+      "loss": 1.0819,
+      "step": 1716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4416723434277086e-06,
+      "loss": 1.0985,
+      "step": 1717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4409751525377696e-06,
+      "loss": 0.6591,
+      "step": 1718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4402775973352347e-06,
+      "loss": 1.0994,
+      "step": 1719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4395796779964635e-06,
+      "loss": 1.0737,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.438881394697905e-06,
+      "loss": 1.0998,
+      "step": 1721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4381827476161027e-06,
+      "loss": 1.0822,
+      "step": 1722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4374837369276904e-06,
+      "loss": 1.1226,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4367843628093956e-06,
+      "loss": 1.0567,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4360846254380355e-06,
+      "loss": 1.0953,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435384524990521e-06,
+      "loss": 1.124,
+      "step": 1726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4346840616438535e-06,
+      "loss": 0.6755,
+      "step": 1727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433983235575127e-06,
+      "loss": 1.1011,
+      "step": 1728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433282046961527e-06,
+      "loss": 0.6467,
+      "step": 1729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.43258049598033e-06,
+      "loss": 1.0759,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4318785828089056e-06,
+      "loss": 1.0942,
+      "step": 1731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431176307624714e-06,
+      "loss": 1.0927,
+      "step": 1732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4304736706053054e-06,
+      "loss": 1.1,
+      "step": 1733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429770671928325e-06,
+      "loss": 1.1606,
+      "step": 1734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429067311771506e-06,
+      "loss": 1.0978,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4283635903126758e-06,
+      "loss": 1.0863,
+      "step": 1736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4276595077297506e-06,
+      "loss": 1.0536,
+      "step": 1737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4269550642007405e-06,
+      "loss": 1.1135,
+      "step": 1738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.426250259903744e-06,
+      "loss": 1.1212,
+      "step": 1739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4255450950169536e-06,
+      "loss": 1.1396,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4248395697186504e-06,
+      "loss": 1.1002,
+      "step": 1741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.424133684187209e-06,
+      "loss": 1.1004,
+      "step": 1742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4234274386010935e-06,
+      "loss": 1.1098,
+      "step": 1743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.422720833138859e-06,
+      "loss": 1.0894,
+      "step": 1744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4220138679791527e-06,
+      "loss": 1.064,
+      "step": 1745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4213065433007122e-06,
+      "loss": 1.0991,
+      "step": 1746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4205988592823653e-06,
+      "loss": 1.0885,
+      "step": 1747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4198908161030316e-06,
+      "loss": 1.1259,
+      "step": 1748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4191824139417218e-06,
+      "loss": 1.104,
+      "step": 1749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4184736529775354e-06,
+      "loss": 1.0752,
+      "step": 1750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4177645333896653e-06,
+      "loss": 1.0771,
+      "step": 1751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.417055055357393e-06,
+      "loss": 1.1405,
+      "step": 1752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4163452190600907e-06,
+      "loss": 1.1295,
+      "step": 1753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4156350246772227e-06,
+      "loss": 1.0917,
+      "step": 1754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4149244723883425e-06,
+      "loss": 1.0789,
+      "step": 1755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.414213562373095e-06,
+      "loss": 1.1483,
+      "step": 1756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4135022948112145e-06,
+      "loss": 0.6952,
+      "step": 1757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4127906698825267e-06,
+      "loss": 1.1161,
+      "step": 1758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4120786877669475e-06,
+      "loss": 1.1105,
+      "step": 1759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4113663486444815e-06,
+      "loss": 1.1194,
+      "step": 1760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4106536526952267e-06,
+      "loss": 1.0654,
+      "step": 1761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.409940600099368e-06,
+      "loss": 1.1071,
+      "step": 1762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.409227191037182e-06,
+      "loss": 1.1146,
+      "step": 1763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4085134256890367e-06,
+      "loss": 1.0861,
+      "step": 1764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4077993042353877e-06,
+      "loss": 1.1133,
+      "step": 1765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407084826856782e-06,
+      "loss": 1.091,
+      "step": 1766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4063699937338566e-06,
+      "loss": 1.0935,
+      "step": 1767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4056548050473383e-06,
+      "loss": 1.1333,
+      "step": 1768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.404939260978043e-06,
+      "loss": 1.1039,
+      "step": 1769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4042233617068785e-06,
+      "loss": 1.0661,
+      "step": 1770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4035071074148396e-06,
+      "loss": 1.0641,
+      "step": 1771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4027904982830135e-06,
+      "loss": 1.1117,
+      "step": 1772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.402073534492575e-06,
+      "loss": 0.6704,
+      "step": 1773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4013562162247896e-06,
+      "loss": 1.0573,
+      "step": 1774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4006385436610138e-06,
+      "loss": 1.1297,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.39992051698269e-06,
+      "loss": 0.6648,
+      "step": 1776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3992021363713538e-06,
+      "loss": 1.079,
+      "step": 1777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3984834020086278e-06,
+      "loss": 1.0616,
+      "step": 1778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.397764314076227e-06,
+      "loss": 1.0996,
+      "step": 1779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.397044872755951e-06,
+      "loss": 1.0854,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3963250782296934e-06,
+      "loss": 1.095,
+      "step": 1781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3956049306794352e-06,
+      "loss": 1.0399,
+      "step": 1782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3948844302872458e-06,
+      "loss": 1.0884,
+      "step": 1783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394163577235286e-06,
+      "loss": 1.1253,
+      "step": 1784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.393442371705804e-06,
+      "loss": 1.095,
+      "step": 1785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3927208138811368e-06,
+      "loss": 1.0855,
+      "step": 1786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3919989039437127e-06,
+      "loss": 1.1152,
+      "step": 1787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391276642076046e-06,
+      "loss": 0.7014,
+      "step": 1788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3905540284607426e-06,
+      "loss": 1.0935,
+      "step": 1789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3898310632804973e-06,
+      "loss": 1.1311,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389107746718091e-06,
+      "loss": 1.0592,
+      "step": 1791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388384078956395e-06,
+      "loss": 1.0693,
+      "step": 1792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3876600601783717e-06,
+      "loss": 1.0906,
+      "step": 1793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3869356905670684e-06,
+      "loss": 1.0798,
+      "step": 1794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.386210970305624e-06,
+      "loss": 0.6727,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3854858995772638e-06,
+      "loss": 1.0753,
+      "step": 1796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3847604785653036e-06,
+      "loss": 1.046,
+      "step": 1797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3840347074531473e-06,
+      "loss": 1.0722,
+      "step": 1798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3833085864242855e-06,
+      "loss": 1.0531,
+      "step": 1799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3825821156623e-06,
+      "loss": 1.0643,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38185529535086e-06,
+      "loss": 1.0452,
+      "step": 1801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3811281256737224e-06,
+      "loss": 1.0874,
+      "step": 1802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3804006068147323e-06,
+      "loss": 1.1372,
+      "step": 1803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.379672738957825e-06,
+      "loss": 1.1012,
+      "step": 1804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3789445222870212e-06,
+      "loss": 1.0997,
+      "step": 1805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3782159569864325e-06,
+      "loss": 1.1283,
+      "step": 1806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.377487043240257e-06,
+      "loss": 1.0698,
+      "step": 1807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.376757781232781e-06,
+      "loss": 1.1055,
+      "step": 1808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3760281711483804e-06,
+      "loss": 1.1528,
+      "step": 1809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.375298213171517e-06,
+      "loss": 1.0648,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3745679074867403e-06,
+      "loss": 1.092,
+      "step": 1811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3738372542786904e-06,
+      "loss": 1.0965,
+      "step": 1812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3731062537320936e-06,
+      "loss": 1.0493,
+      "step": 1813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.372374906031764e-06,
+      "loss": 1.0695,
+      "step": 1814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.371643211362603e-06,
+      "loss": 1.1035,
+      "step": 1815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3709111699096003e-06,
+      "loss": 1.065,
+      "step": 1816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3701787818578347e-06,
+      "loss": 1.1036,
+      "step": 1817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3694460473924703e-06,
+      "loss": 1.0756,
+      "step": 1818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3687129666987584e-06,
+      "loss": 0.6782,
+      "step": 1819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3679795399620416e-06,
+      "loss": 1.0947,
+      "step": 1820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.367245767367746e-06,
+      "loss": 1.0718,
+      "step": 1821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3665116491013865e-06,
+      "loss": 1.0856,
+      "step": 1822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365777185348567e-06,
+      "loss": 1.1292,
+      "step": 1823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3650423762949754e-06,
+      "loss": 1.1038,
+      "step": 1824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3643072221263903e-06,
+      "loss": 1.1249,
+      "step": 1825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3635717230286753e-06,
+      "loss": 1.08,
+      "step": 1826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3628358791877826e-06,
+      "loss": 1.1412,
+      "step": 1827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3620996907897493e-06,
+      "loss": 1.0999,
+      "step": 1828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3613631580207037e-06,
+      "loss": 1.0332,
+      "step": 1829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3606262810668567e-06,
+      "loss": 1.1192,
+      "step": 1830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.359889060114509e-06,
+      "loss": 1.1119,
+      "step": 1831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3591514953500477e-06,
+      "loss": 1.0828,
+      "step": 1832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.358413586959946e-06,
+      "loss": 1.0786,
+      "step": 1833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3576753351307654e-06,
+      "loss": 1.0885,
+      "step": 1834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356936740049153e-06,
+      "loss": 0.7258,
+      "step": 1835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3561978019018433e-06,
+      "loss": 0.6843,
+      "step": 1836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.355458520875656e-06,
+      "loss": 0.6951,
+      "step": 1837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.354718897157501e-06,
+      "loss": 1.1113,
+      "step": 1838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3539789309343714e-06,
+      "loss": 1.1028,
+      "step": 1839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.353238622393348e-06,
+      "loss": 1.1103,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.352497971721599e-06,
+      "loss": 1.0897,
+      "step": 1841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.351756979106378e-06,
+      "loss": 1.0988,
+      "step": 1842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3510156447350253e-06,
+      "loss": 1.0957,
+      "step": 1843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3502739687949683e-06,
+      "loss": 1.0788,
+      "step": 1844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.34953195147372e-06,
+      "loss": 1.0647,
+      "step": 1845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3487895929588794e-06,
+      "loss": 1.0671,
+      "step": 1846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.348046893438133e-06,
+      "loss": 0.6899,
+      "step": 1847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3473038530992524e-06,
+      "loss": 1.1175,
+      "step": 1848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.346560472130096e-06,
+      "loss": 1.108,
+      "step": 1849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.345816750718608e-06,
+      "loss": 1.1085,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3450726890528187e-06,
+      "loss": 1.1204,
+      "step": 1851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3443282873208443e-06,
+      "loss": 1.1248,
+      "step": 1852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343583545710888e-06,
+      "loss": 1.0756,
+      "step": 1853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.342838464411237e-06,
+      "loss": 0.6902,
+      "step": 1854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.342093043610266e-06,
+      "loss": 1.1242,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.341347283496435e-06,
+      "loss": 1.1022,
+      "step": 1856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3406011842582903e-06,
+      "loss": 1.0895,
+      "step": 1857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3398547460844623e-06,
+      "loss": 1.1319,
+      "step": 1858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3391079691636687e-06,
+      "loss": 1.1117,
+      "step": 1859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.338360853684713e-06,
+      "loss": 0.6821,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3376133998364828e-06,
+      "loss": 0.6738,
+      "step": 1861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3368656078079526e-06,
+      "loss": 1.0553,
+      "step": 1862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3361174777881814e-06,
+      "loss": 1.071,
+      "step": 1863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3353690099663147e-06,
+      "loss": 1.0749,
+      "step": 1864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.334620204531583e-06,
+      "loss": 1.0984,
+      "step": 1865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3338710616733013e-06,
+      "loss": 1.0932,
+      "step": 1866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3331215815808716e-06,
+      "loss": 1.1044,
+      "step": 1867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3323717644437785e-06,
+      "loss": 1.0928,
+      "step": 1868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.331621610451596e-06,
+      "loss": 1.0833,
+      "step": 1869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3308711197939787e-06,
+      "loss": 1.1035,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3301202926606688e-06,
+      "loss": 0.698,
+      "step": 1871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.329369129241494e-06,
+      "loss": 1.1228,
+      "step": 1872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3286176297263656e-06,
+      "loss": 1.1082,
+      "step": 1873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.32786579430528e-06,
+      "loss": 1.0673,
+      "step": 1874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.32711362316832e-06,
+      "loss": 1.0443,
+      "step": 1875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3263611165056516e-06,
+      "loss": 1.0627,
+      "step": 1876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3256082745075258e-06,
+      "loss": 1.0606,
+      "step": 1877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3248550973642803e-06,
+      "loss": 1.1337,
+      "step": 1878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3241015852663345e-06,
+      "loss": 1.1043,
+      "step": 1879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.323347738404195e-06,
+      "loss": 1.109,
+      "step": 1880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3225935569684522e-06,
+      "loss": 1.0947,
+      "step": 1881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3218390411497806e-06,
+      "loss": 1.1012,
+      "step": 1882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.32108419113894e-06,
+      "loss": 0.636,
+      "step": 1883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.320329007126774e-06,
+      "loss": 0.6813,
+      "step": 1884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3195734893042104e-06,
+      "loss": 1.1475,
+      "step": 1885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3188176378622633e-06,
+      "loss": 1.0312,
+      "step": 1886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.318061452992029e-06,
+      "loss": 1.0952,
+      "step": 1887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.317304934884689e-06,
+      "loss": 1.1033,
+      "step": 1888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.316548083731509e-06,
+      "loss": 1.0912,
+      "step": 1889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.315790899723839e-06,
+      "loss": 1.0636,
+      "step": 1890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3150333830531133e-06,
+      "loss": 1.0981,
+      "step": 1891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3142755339108495e-06,
+      "loss": 1.1664,
+      "step": 1892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.31351735248865e-06,
+      "loss": 1.0952,
+      "step": 1893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.312758838978201e-06,
+      "loss": 1.1303,
+      "step": 1894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.311999993571272e-06,
+      "loss": 1.0992,
+      "step": 1895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3112408164597182e-06,
+      "loss": 1.0867,
+      "step": 1896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3104813078354773e-06,
+      "loss": 1.064,
+      "step": 1897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3097214678905703e-06,
+      "loss": 1.0815,
+      "step": 1898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3089612968171032e-06,
+      "loss": 1.079,
+      "step": 1899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.308200794807265e-06,
+      "loss": 0.7152,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.307439962053329e-06,
+      "loss": 1.0756,
+      "step": 1901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3066787987476514e-06,
+      "loss": 1.0788,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3059173050826725e-06,
+      "loss": 1.1321,
+      "step": 1903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3051554812509156e-06,
+      "loss": 1.0948,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304393327444988e-06,
+      "loss": 1.0968,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.30363084385758e-06,
+      "loss": 1.0715,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.302868030681466e-06,
+      "loss": 1.0659,
+      "step": 1907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3021048881095028e-06,
+      "loss": 1.1167,
+      "step": 1908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3013414163346305e-06,
+      "loss": 1.0844,
+      "step": 1909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.300577615549874e-06,
+      "loss": 1.1243,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2998134859483388e-06,
+      "loss": 1.0495,
+      "step": 1911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2990490277232155e-06,
+      "loss": 1.0949,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2982842410677776e-06,
+      "loss": 1.0961,
+      "step": 1913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2975191261753815e-06,
+      "loss": 1.082,
+      "step": 1914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2967536832394653e-06,
+      "loss": 0.6765,
+      "step": 1915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.295987912453552e-06,
+      "loss": 1.1382,
+      "step": 1916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2952218140112456e-06,
+      "loss": 1.0581,
+      "step": 1917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2944553881062344e-06,
+      "loss": 1.1347,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2936886349322893e-06,
+      "loss": 1.0736,
+      "step": 1919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.292921554683263e-06,
+      "loss": 1.0842,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2921541475530924e-06,
+      "loss": 1.0897,
+      "step": 1921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2913864137357956e-06,
+      "loss": 1.0985,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290618353425474e-06,
+      "loss": 1.0809,
+      "step": 1923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2898499668163107e-06,
+      "loss": 1.0949,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289081254102573e-06,
+      "loss": 1.1372,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2883122154786095e-06,
+      "loss": 1.1028,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2875428511388516e-06,
+      "loss": 1.1124,
+      "step": 1927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2867731612778115e-06,
+      "loss": 0.6589,
+      "step": 1928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.286003146090086e-06,
+      "loss": 1.0724,
+      "step": 1929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.285232805770353e-06,
+      "loss": 1.1057,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.284462140513373e-06,
+      "loss": 1.0894,
+      "step": 1931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2836911505139884e-06,
+      "loss": 1.1125,
+      "step": 1932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2829198359671236e-06,
+      "loss": 1.0965,
+      "step": 1933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2821481970677844e-06,
+      "loss": 1.094,
+      "step": 1934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.28137623401106e-06,
+      "loss": 1.0834,
+      "step": 1935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2806039469921213e-06,
+      "loss": 1.0941,
+      "step": 1936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2798313362062196e-06,
+      "loss": 1.0957,
+      "step": 1937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2790584018486913e-06,
+      "loss": 1.0788,
+      "step": 1938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2782851441149495e-06,
+      "loss": 1.0789,
+      "step": 1939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2775115632004945e-06,
+      "loss": 1.1122,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.276737659300905e-06,
+      "loss": 1.1184,
+      "step": 1941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2759634326118424e-06,
+      "loss": 0.7051,
+      "step": 1942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.275188883329048e-06,
+      "loss": 1.0663,
+      "step": 1943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.274414011648349e-06,
+      "loss": 1.0841,
+      "step": 1944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2736388177656487e-06,
+      "loss": 1.0799,
+      "step": 1945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.272863301876936e-06,
+      "loss": 1.1168,
+      "step": 1946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.272087464178279e-06,
+      "loss": 1.0742,
+      "step": 1947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2713113048658285e-06,
+      "loss": 1.1127,
+      "step": 1948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.270534824135815e-06,
+      "loss": 1.0551,
+      "step": 1949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2697580221845515e-06,
+      "loss": 1.0884,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.268980899208432e-06,
+      "loss": 0.6737,
+      "step": 1951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.268203455403932e-06,
+      "loss": 1.0661,
+      "step": 1952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2674256909676065e-06,
+      "loss": 1.0982,
+      "step": 1953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.266647606096094e-06,
+      "loss": 0.6904,
+      "step": 1954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.265869200986112e-06,
+      "loss": 1.1026,
+      "step": 1955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2650904758344597e-06,
+      "loss": 1.091,
+      "step": 1956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2643114308380184e-06,
+      "loss": 0.6819,
+      "step": 1957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2635320661937476e-06,
+      "loss": 1.0551,
+      "step": 1958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2627523820986897e-06,
+      "loss": 1.0543,
+      "step": 1959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2619723787499674e-06,
+      "loss": 1.0519,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.261192056344784e-06,
+      "loss": 1.0638,
+      "step": 1961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.260411415080423e-06,
+      "loss": 1.0959,
+      "step": 1962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.25963045515425e-06,
+      "loss": 1.1068,
+      "step": 1963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.258849176763709e-06,
+      "loss": 1.1144,
+      "step": 1964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.258067580106326e-06,
+      "loss": 1.0763,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.257285665379707e-06,
+      "loss": 1.1175,
+      "step": 1966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2565034327815398e-06,
+      "loss": 1.0907,
+      "step": 1967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.255720882509589e-06,
+      "loss": 1.0817,
+      "step": 1968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2549380147617037e-06,
+      "loss": 1.0791,
+      "step": 1969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.254154829735811e-06,
+      "loss": 1.0716,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.253371327629918e-06,
+      "loss": 0.683,
+      "step": 1971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.252587508642113e-06,
+      "loss": 1.0955,
+      "step": 1972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2518033729705633e-06,
+      "loss": 1.1005,
+      "step": 1973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2510189208135187e-06,
+      "loss": 1.1016,
+      "step": 1974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.250234152369305e-06,
+      "loss": 1.0978,
+      "step": 1975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.249449067836332e-06,
+      "loss": 1.0669,
+      "step": 1976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2486636674130865e-06,
+      "loss": 1.0945,
+      "step": 1977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.247877951298137e-06,
+      "loss": 1.0875,
+      "step": 1978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.247091919690131e-06,
+      "loss": 0.6747,
+      "step": 1979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2463055727877953e-06,
+      "loss": 1.0825,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2455189107899377e-06,
+      "loss": 1.0827,
+      "step": 1981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.244731933895445e-06,
+      "loss": 1.0544,
+      "step": 1982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.243944642303282e-06,
+      "loss": 1.1126,
+      "step": 1983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.243157036212497e-06,
+      "loss": 0.6875,
+      "step": 1984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2423691158222144e-06,
+      "loss": 0.6548,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2415808813316384e-06,
+      "loss": 1.0872,
+      "step": 1986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2407923329400546e-06,
+      "loss": 1.0621,
+      "step": 1987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.240003470846826e-06,
+      "loss": 1.0532,
+      "step": 1988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2392142952513953e-06,
+      "loss": 1.0659,
+      "step": 1989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2384248063532845e-06,
+      "loss": 1.0695,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.237635004352097e-06,
+      "loss": 1.1165,
+      "step": 1991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.236844889447511e-06,
+      "loss": 1.0762,
+      "step": 1992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2360544618392882e-06,
+      "loss": 1.0893,
+      "step": 1993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.235263721727266e-06,
+      "loss": 1.0557,
+      "step": 1994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234472669311363e-06,
+      "loss": 1.1032,
+      "step": 1995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2336813047915747e-06,
+      "loss": 1.0771,
+      "step": 1996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2328896283679787e-06,
+      "loss": 1.089,
+      "step": 1997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2320976402407283e-06,
+      "loss": 1.0961,
+      "step": 1998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.231305340610057e-06,
+      "loss": 1.0924,
+      "step": 1999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2305127296762768e-06,
+      "loss": 1.0417,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.229719807639779e-06,
+      "loss": 1.1172,
+      "step": 2001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.228926574701033e-06,
+      "loss": 1.0574,
+      "step": 2002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.228133031060586e-06,
+      "loss": 1.0791,
+      "step": 2003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227339176919065e-06,
+      "loss": 1.1261,
+      "step": 2004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2265450124771753e-06,
+      "loss": 1.0893,
+      "step": 2005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2257505379357005e-06,
+      "loss": 1.081,
+      "step": 2006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2249557534955024e-06,
+      "loss": 1.1212,
+      "step": 2007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2241606593575213e-06,
+      "loss": 1.1034,
+      "step": 2008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2233652557227755e-06,
+      "loss": 1.1054,
+      "step": 2009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222569542792363e-06,
+      "loss": 1.0941,
+      "step": 2010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2217735207674575e-06,
+      "loss": 1.0818,
+      "step": 2011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220977189849313e-06,
+      "loss": 1.0839,
+      "step": 2012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.22018055023926e-06,
+      "loss": 1.1155,
+      "step": 2013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.219383602138709e-06,
+      "loss": 1.1187,
+      "step": 2014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2185863457491467e-06,
+      "loss": 1.0936,
+      "step": 2015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.217788781272138e-06,
+      "loss": 1.1362,
+      "step": 2016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.216990908909326e-06,
+      "loss": 1.0634,
+      "step": 2017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.216192728862433e-06,
+      "loss": 1.0831,
+      "step": 2018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.215394241333256e-06,
+      "loss": 1.078,
+      "step": 2019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.214595446523673e-06,
+      "loss": 1.0842,
+      "step": 2020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2137963446356384e-06,
+      "loss": 1.0896,
+      "step": 2021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2129969358711823e-06,
+      "loss": 1.0708,
+      "step": 2022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.212197220432415e-06,
+      "loss": 0.6622,
+      "step": 2023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.211397198521524e-06,
+      "loss": 0.6869,
+      "step": 2024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2105968703407733e-06,
+      "loss": 1.1112,
+      "step": 2025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.209796236092505e-06,
+      "loss": 1.0659,
+      "step": 2026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2089952959791376e-06,
+      "loss": 1.1345,
+      "step": 2027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.208194050203169e-06,
+      "loss": 1.1019,
+      "step": 2028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.207392498967173e-06,
+      "loss": 1.1316,
+      "step": 2029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2065906424737985e-06,
+      "loss": 1.0863,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2057884809257765e-06,
+      "loss": 1.1105,
+      "step": 2031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2049860145259114e-06,
+      "loss": 1.1038,
+      "step": 2032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204183243477085e-06,
+      "loss": 1.1014,
+      "step": 2033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.203380167982258e-06,
+      "loss": 1.0775,
+      "step": 2034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.202576788244466e-06,
+      "loss": 1.0791,
+      "step": 2035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.201773104466823e-06,
+      "loss": 1.0601,
+      "step": 2036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2009691168525185e-06,
+      "loss": 1.0935,
+      "step": 2037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.20016482560482e-06,
+      "loss": 1.0921,
+      "step": 2038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1993602309270714e-06,
+      "loss": 1.0691,
+      "step": 2039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.198555333022694e-06,
+      "loss": 1.0823,
+      "step": 2040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1977501320951837e-06,
+      "loss": 1.0214,
+      "step": 2041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.196944628348115e-06,
+      "loss": 1.1106,
+      "step": 2042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.196138821985138e-06,
+      "loss": 1.091,
+      "step": 2043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1953327132099796e-06,
+      "loss": 1.1149,
+      "step": 2044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1945263022264437e-06,
+      "loss": 1.0997,
+      "step": 2045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1937195892384095e-06,
+      "loss": 1.1129,
+      "step": 2046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.192912574449834e-06,
+      "loss": 1.0712,
+      "step": 2047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1921052580647486e-06,
+      "loss": 1.0466,
+      "step": 2048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.191297640287262e-06,
+      "loss": 1.1223,
+      "step": 2049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19048972132156e-06,
+      "loss": 0.6906,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.189681501371903e-06,
+      "loss": 1.0595,
+      "step": 2051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1888729806426273e-06,
+      "loss": 1.1177,
+      "step": 2052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1880641593381475e-06,
+      "loss": 1.107,
+      "step": 2053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1872550376629526e-06,
+      "loss": 1.1115,
+      "step": 2054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186445615821607e-06,
+      "loss": 1.1025,
+      "step": 2055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1856358940187516e-06,
+      "loss": 1.1067,
+      "step": 2056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184825872459104e-06,
+      "loss": 1.0747,
+      "step": 2057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184015551347456e-06,
+      "loss": 0.6717,
+      "step": 2058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.183204930888677e-06,
+      "loss": 1.0697,
+      "step": 2059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1823940112877098e-06,
+      "loss": 1.0774,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1815827927495746e-06,
+      "loss": 1.1125,
+      "step": 2061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1807712754793666e-06,
+      "loss": 0.6863,
+      "step": 2062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1799594596822567e-06,
+      "loss": 0.676,
+      "step": 2063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.179147345563491e-06,
+      "loss": 0.6585,
+      "step": 2064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178334933328391e-06,
+      "loss": 1.0944,
+      "step": 2065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.177522223182354e-06,
+      "loss": 1.0648,
+      "step": 2066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1767092153308524e-06,
+      "loss": 1.1095,
+      "step": 2067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1758959099794338e-06,
+      "loss": 1.0755,
+      "step": 2068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1750823073337206e-06,
+      "loss": 1.1027,
+      "step": 2069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.174268407599411e-06,
+      "loss": 1.0812,
+      "step": 2070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.173454210982278e-06,
+      "loss": 1.111,
+      "step": 2071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.17263971768817e-06,
+      "loss": 1.0527,
+      "step": 2072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1718249279230103e-06,
+      "loss": 1.1173,
+      "step": 2073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1710098418927967e-06,
+      "loss": 1.0593,
+      "step": 2074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1701944598036027e-06,
+      "loss": 1.0999,
+      "step": 2075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1693787818615753e-06,
+      "loss": 1.1103,
+      "step": 2076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.168562808272938e-06,
+      "loss": 1.0736,
+      "step": 2077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.167746539243988e-06,
+      "loss": 1.1148,
+      "step": 2078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1669299749810973e-06,
+      "loss": 0.686,
+      "step": 2079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166113115690712e-06,
+      "loss": 1.0898,
+      "step": 2080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.165295961579355e-06,
+      "loss": 0.6702,
+      "step": 2081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1644785128536207e-06,
+      "loss": 0.6799,
+      "step": 2082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1636607697201806e-06,
+      "loss": 1.0788,
+      "step": 2083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.162842732385779e-06,
+      "loss": 1.076,
+      "step": 2084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1620244010572343e-06,
+      "loss": 1.1165,
+      "step": 2085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1612057759414413e-06,
+      "loss": 0.6832,
+      "step": 2086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.160386857245367e-06,
+      "loss": 1.1089,
+      "step": 2087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1595676451760537e-06,
+      "loss": 1.1002,
+      "step": 2088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158748139940618e-06,
+      "loss": 1.0977,
+      "step": 2089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.15792834174625e-06,
+      "loss": 1.0986,
+      "step": 2090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1571082508002127e-06,
+      "loss": 1.093,
+      "step": 2091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1562878673098467e-06,
+      "loss": 0.6716,
+      "step": 2092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.155467191482563e-06,
+      "loss": 1.0919,
+      "step": 2093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1546462235258477e-06,
+      "loss": 1.0724,
+      "step": 2094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1538249636472617e-06,
+      "loss": 1.0611,
+      "step": 2095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.153003412054438e-06,
+      "loss": 1.0504,
+      "step": 2096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1521815689550853e-06,
+      "loss": 1.1155,
+      "step": 2097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1513594345569843e-06,
+      "loss": 0.6718,
+      "step": 2098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1505370090679905e-06,
+      "loss": 1.0499,
+      "step": 2099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1497142926960322e-06,
+      "loss": 0.7106,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.148891285649111e-06,
+      "loss": 1.0487,
+      "step": 2101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1480679881353037e-06,
+      "loss": 1.0205,
+      "step": 2102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.147244400362759e-06,
+      "loss": 1.0973,
+      "step": 2103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.146420522539699e-06,
+      "loss": 0.6626,
+      "step": 2104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1455963548744198e-06,
+      "loss": 1.0941,
+      "step": 2105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.14477189757529e-06,
+      "loss": 1.1115,
+      "step": 2106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.143947150850752e-06,
+      "loss": 1.1064,
+      "step": 2107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.143122114909322e-06,
+      "loss": 1.0909,
+      "step": 2108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1422967899595885e-06,
+      "loss": 1.1036,
+      "step": 2109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141471176210212e-06,
+      "loss": 1.1171,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.140645273869928e-06,
+      "loss": 1.0606,
+      "step": 2111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1398190831475445e-06,
+      "loss": 1.082,
+      "step": 2112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1389926042519417e-06,
+      "loss": 1.0836,
+      "step": 2113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.138165837392073e-06,
+      "loss": 1.1049,
+      "step": 2114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1373387827769637e-06,
+      "loss": 1.0862,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1365114406157146e-06,
+      "loss": 1.0429,
+      "step": 2116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1356838111174963e-06,
+      "loss": 1.1088,
+      "step": 2117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1348558944915533e-06,
+      "loss": 1.0706,
+      "step": 2118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.134027690947202e-06,
+      "loss": 1.0865,
+      "step": 2119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.133199200693832e-06,
+      "loss": 1.0572,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1323704239409065e-06,
+      "loss": 1.127,
+      "step": 2121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131541360897958e-06,
+      "loss": 1.058,
+      "step": 2122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1307120117745947e-06,
+      "loss": 1.0953,
+      "step": 2123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.129882376780495e-06,
+      "loss": 1.1128,
+      "step": 2124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1290524561254097e-06,
+      "loss": 1.0946,
+      "step": 2125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.128222250019164e-06,
+      "loss": 1.0899,
+      "step": 2126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1273917586716517e-06,
+      "loss": 1.1068,
+      "step": 2127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1265609822928426e-06,
+      "loss": 0.6924,
+      "step": 2128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.125729921092775e-06,
+      "loss": 1.1166,
+      "step": 2129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.124898575281561e-06,
+      "loss": 1.0924,
+      "step": 2130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1240669450693853e-06,
+      "loss": 1.0337,
+      "step": 2131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1232350306665034e-06,
+      "loss": 0.7074,
+      "step": 2132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.122402832283243e-06,
+      "loss": 0.6798,
+      "step": 2133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.121570350130003e-06,
+      "loss": 1.0473,
+      "step": 2134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1207375844172548e-06,
+      "loss": 1.0687,
+      "step": 2135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1199045353555417e-06,
+      "loss": 0.6912,
+      "step": 2136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.119071203155478e-06,
+      "loss": 1.0282,
+      "step": 2137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1182375880277493e-06,
+      "loss": 1.0638,
+      "step": 2138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.117403690183114e-06,
+      "loss": 1.0575,
+      "step": 2139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1165695098324002e-06,
+      "loss": 1.0975,
+      "step": 2140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1157350471865097e-06,
+      "loss": 1.0885,
+      "step": 2141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114900302456413e-06,
+      "loss": 1.0655,
+      "step": 2142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1140652758531543e-06,
+      "loss": 1.0874,
+      "step": 2143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1132299675878486e-06,
+      "loss": 1.097,
+      "step": 2144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1123943778716798e-06,
+      "loss": 1.127,
+      "step": 2145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111558506915906e-06,
+      "loss": 1.0962,
+      "step": 2146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.110722354931855e-06,
+      "loss": 1.0399,
+      "step": 2147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1098859221309255e-06,
+      "loss": 1.1109,
+      "step": 2148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1090492087245876e-06,
+      "loss": 1.0786,
+      "step": 2149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.108212214924382e-06,
+      "loss": 1.1176,
+      "step": 2150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1073749409419216e-06,
+      "loss": 1.1118,
+      "step": 2151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.106537386988888e-06,
+      "loss": 0.6688,
+      "step": 2152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1056995532770346e-06,
+      "loss": 1.1114,
+      "step": 2153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.104861440018186e-06,
+      "loss": 1.081,
+      "step": 2154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1040230474242366e-06,
+      "loss": 0.6834,
+      "step": 2155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.103184375707153e-06,
+      "loss": 1.1026,
+      "step": 2156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1023454250789696e-06,
+      "loss": 1.0554,
+      "step": 2157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1015061957517947e-06,
+      "loss": 1.1003,
+      "step": 2158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.100666687937804e-06,
+      "loss": 1.0242,
+      "step": 2159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0998269018492454e-06,
+      "loss": 1.064,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0989868376984366e-06,
+      "loss": 1.0519,
+      "step": 2161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.098146495697766e-06,
+      "loss": 1.0839,
+      "step": 2162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.097305876059692e-06,
+      "loss": 1.0423,
+      "step": 2163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0964649789967423e-06,
+      "loss": 1.038,
+      "step": 2164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.095623804721516e-06,
+      "loss": 0.6938,
+      "step": 2165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0947823534466814e-06,
+      "loss": 1.1094,
+      "step": 2166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0939406253849786e-06,
+      "loss": 1.0875,
+      "step": 2167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0930986207492148e-06,
+      "loss": 1.099,
+      "step": 2168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0922563397522704e-06,
+      "loss": 1.0884,
+      "step": 2169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.091413782607093e-06,
+      "loss": 1.0908,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0905709495267004e-06,
+      "loss": 1.1018,
+      "step": 2171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.089727840724181e-06,
+      "loss": 1.0641,
+      "step": 2172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0888844564126936e-06,
+      "loss": 1.0891,
+      "step": 2173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0880407968054645e-06,
+      "loss": 1.0615,
+      "step": 2174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0871968621157916e-06,
+      "loss": 1.0636,
+      "step": 2175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0863526525570415e-06,
+      "loss": 0.7009,
+      "step": 2176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0855081683426497e-06,
+      "loss": 1.0563,
+      "step": 2177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.084663409686122e-06,
+      "loss": 1.0843,
+      "step": 2178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.083818376801034e-06,
+      "loss": 1.1009,
+      "step": 2179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0829730699010296e-06,
+      "loss": 0.6768,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0821274891998225e-06,
+      "loss": 1.0673,
+      "step": 2181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0812816349111954e-06,
+      "loss": 1.123,
+      "step": 2182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0804355072489997e-06,
+      "loss": 1.0943,
+      "step": 2183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0795891064271572e-06,
+      "loss": 1.0775,
+      "step": 2184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0787424326596574e-06,
+      "loss": 1.0769,
+      "step": 2185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0778954861605607e-06,
+      "loss": 1.1021,
+      "step": 2186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0770482671439945e-06,
+      "loss": 1.11,
+      "step": 2187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0762007758241553e-06,
+      "loss": 0.6781,
+      "step": 2188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0753530124153083e-06,
+      "loss": 1.0883,
+      "step": 2189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0745049771317904e-06,
+      "loss": 1.0847,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0736566701880037e-06,
+      "loss": 0.6475,
+      "step": 2191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.07280809179842e-06,
+      "loss": 1.0786,
+      "step": 2192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.07195924217758e-06,
+      "loss": 1.0552,
+      "step": 2193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0711101215400944e-06,
+      "loss": 1.1055,
+      "step": 2194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0702607301006387e-06,
+      "loss": 1.0984,
+      "step": 2195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0694110680739605e-06,
+      "loss": 1.1421,
+      "step": 2196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0685611356748746e-06,
+      "loss": 1.0407,
+      "step": 2197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067710933118263e-06,
+      "loss": 1.127,
+      "step": 2198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0668604606190785e-06,
+      "loss": 1.1034,
+      "step": 2199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0660097183923396e-06,
+      "loss": 1.0795,
+      "step": 2200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0651587066531334e-06,
+      "loss": 1.084,
+      "step": 2201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0643074256166173e-06,
+      "loss": 1.0968,
+      "step": 2202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.063455875498015e-06,
+      "loss": 1.0589,
+      "step": 2203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.062604056512618e-06,
+      "loss": 1.1144,
+      "step": 2204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.061751968875786e-06,
+      "loss": 0.6509,
+      "step": 2205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0608996128029475e-06,
+      "loss": 1.0928,
+      "step": 2206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.060046988509598e-06,
+      "loss": 1.061,
+      "step": 2207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0591940962113017e-06,
+      "loss": 1.0661,
+      "step": 2208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.058340936123689e-06,
+      "loss": 1.0619,
+      "step": 2209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.057487508462459e-06,
+      "loss": 1.0596,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.056633813443379e-06,
+      "loss": 0.6551,
+      "step": 2211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.055779851282283e-06,
+      "loss": 1.045,
+      "step": 2212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0549256221950735e-06,
+      "loss": 1.0842,
+      "step": 2213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0540711263977194e-06,
+      "loss": 1.0873,
+      "step": 2214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.053216364106256e-06,
+      "loss": 1.0505,
+      "step": 2215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0523613355367896e-06,
+      "loss": 0.688,
+      "step": 2216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0515060409054903e-06,
+      "loss": 1.0882,
+      "step": 2217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.050650480428597e-06,
+      "loss": 1.0989,
+      "step": 2218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.049794654322417e-06,
+      "loss": 1.0956,
+      "step": 2219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.048938562803322e-06,
+      "loss": 1.0754,
+      "step": 2220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.048082206087751e-06,
+      "loss": 1.059,
+      "step": 2221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.047225584392214e-06,
+      "loss": 1.0363,
+      "step": 2222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.046368697933283e-06,
+      "loss": 1.103,
+      "step": 2223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0455115469276002e-06,
+      "loss": 0.6743,
+      "step": 2224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0446541315918736e-06,
+      "loss": 1.0737,
+      "step": 2225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.043796452142878e-06,
+      "loss": 1.0955,
+      "step": 2226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0429385087974553e-06,
+      "loss": 1.0578,
+      "step": 2227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0420803017725128e-06,
+      "loss": 1.0559,
+      "step": 2228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.041221831285027e-06,
+      "loss": 0.7258,
+      "step": 2229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0403630975520385e-06,
+      "loss": 1.0925,
+      "step": 2230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.039504100790656e-06,
+      "loss": 0.6835,
+      "step": 2231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0386448412180537e-06,
+      "loss": 1.0567,
+      "step": 2232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0377853190514735e-06,
+      "loss": 1.0626,
+      "step": 2233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0369255345082216e-06,
+      "loss": 1.0837,
+      "step": 2234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.036065487805674e-06,
+      "loss": 1.101,
+      "step": 2235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0352051791612683e-06,
+      "loss": 1.0951,
+      "step": 2236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.034344608792513e-06,
+      "loss": 1.0714,
+      "step": 2237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.03348377691698e-06,
+      "loss": 1.079,
+      "step": 2238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.032622683752307e-06,
+      "loss": 1.1103,
+      "step": 2239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0317613295162e-06,
+      "loss": 1.0827,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0308997144264287e-06,
+      "loss": 1.0696,
+      "step": 2241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0300378387008304e-06,
+      "loss": 1.0553,
+      "step": 2242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0291757025573078e-06,
+      "loss": 1.1119,
+      "step": 2243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.028313306213829e-06,
+      "loss": 1.0996,
+      "step": 2244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.027450649888428e-06,
+      "loss": 1.0447,
+      "step": 2245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0265877337992055e-06,
+      "loss": 1.0839,
+      "step": 2246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0257245581643265e-06,
+      "loss": 1.1115,
+      "step": 2247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0248611232020217e-06,
+      "loss": 1.0729,
+      "step": 2248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0239974291305893e-06,
+      "loss": 1.079,
+      "step": 2249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0231334761683895e-06,
+      "loss": 1.1004,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.022269264533852e-06,
+      "loss": 0.6395,
+      "step": 2251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.021404794445469e-06,
+      "loss": 1.1249,
+      "step": 2252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.020540066121799e-06,
+      "loss": 1.0834,
+      "step": 2253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0196750797814662e-06,
+      "loss": 1.1033,
+      "step": 2254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0188098356431593e-06,
+      "loss": 1.0792,
+      "step": 2255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0179443339256327e-06,
+      "loss": 1.0819,
+      "step": 2256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.017078574847705e-06,
+      "loss": 1.0454,
+      "step": 2257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.016212558628262e-06,
+      "loss": 0.6517,
+      "step": 2258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0153462854862514e-06,
+      "loss": 1.1173,
+      "step": 2259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.014479755640689e-06,
+      "loss": 1.0603,
+      "step": 2260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0136129693106533e-06,
+      "loss": 1.0791,
+      "step": 2261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.012745926715288e-06,
+      "loss": 1.1055,
+      "step": 2262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.011878628073802e-06,
+      "loss": 1.0793,
+      "step": 2263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.01101107360547e-06,
+      "loss": 1.0813,
+      "step": 2264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0101432635296295e-06,
+      "loss": 1.0931,
+      "step": 2265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0092751980656833e-06,
+      "loss": 0.6664,
+      "step": 2266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.008406877433099e-06,
+      "loss": 1.0626,
+      "step": 2267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0075383018514095e-06,
+      "loss": 1.0621,
+      "step": 2268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.00666947154021e-06,
+      "loss": 1.0725,
+      "step": 2269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.005800386719161e-06,
+      "loss": 1.0732,
+      "step": 2270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0049310476079895e-06,
+      "loss": 1.0574,
+      "step": 2271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0040614544264835e-06,
+      "loss": 1.0832,
+      "step": 2272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0031916073944976e-06,
+      "loss": 1.0654,
+      "step": 2273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.002321506731949e-06,
+      "loss": 1.031,
+      "step": 2274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0014511526588195e-06,
+      "loss": 1.09,
+      "step": 2275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0005805453951563e-06,
+      "loss": 1.0701,
+      "step": 2276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.999709685161069e-06,
+      "loss": 1.0861,
+      "step": 2277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9988385721767315e-06,
+      "loss": 1.0877,
+      "step": 2278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9979672066623812e-06,
+      "loss": 0.6606,
+      "step": 2279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9970955888383214e-06,
+      "loss": 1.0908,
+      "step": 2280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9962237189249164e-06,
+      "loss": 0.663,
+      "step": 2281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.995351597142596e-06,
+      "loss": 1.0873,
+      "step": 2282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.994479223711853e-06,
+      "loss": 1.0163,
+      "step": 2283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.993606598853244e-06,
+      "loss": 0.6552,
+      "step": 2284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9927337227873903e-06,
+      "loss": 1.1132,
+      "step": 2285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.991860595734974e-06,
+      "loss": 1.0543,
+      "step": 2286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9909872179167433e-06,
+      "loss": 0.6564,
+      "step": 2287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9901135895535085e-06,
+      "loss": 1.0691,
+      "step": 2288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9892397108661432e-06,
+      "loss": 1.0981,
+      "step": 2289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.988365582075585e-06,
+      "loss": 1.0683,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.987491203402834e-06,
+      "loss": 1.0827,
+      "step": 2291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.986616575068955e-06,
+      "loss": 1.0293,
+      "step": 2292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.985741697295073e-06,
+      "loss": 0.629,
+      "step": 2293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9848665703023777e-06,
+      "loss": 1.079,
+      "step": 2294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9839911943121238e-06,
+      "loss": 1.0889,
+      "step": 2295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.983115569545626e-06,
+      "loss": 1.0877,
+      "step": 2296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9822396962242626e-06,
+      "loss": 1.0889,
+      "step": 2297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.981363574569475e-06,
+      "loss": 0.6999,
+      "step": 2298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9804872048027685e-06,
+      "loss": 1.0388,
+      "step": 2299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9796105871457094e-06,
+      "loss": 0.6977,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9787337218199272e-06,
+      "loss": 1.1021,
+      "step": 2301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.977856609047115e-06,
+      "loss": 0.6674,
+      "step": 2302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9769792490490267e-06,
+      "loss": 1.0964,
+      "step": 2303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97610164204748e-06,
+      "loss": 1.0616,
+      "step": 2304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975223788264354e-06,
+      "loss": 1.0692,
+      "step": 2305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.974345687921593e-06,
+      "loss": 1.0546,
+      "step": 2306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9734673412412e-06,
+      "loss": 1.1166,
+      "step": 2307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.972588748445241e-06,
+      "loss": 1.0688,
+      "step": 2308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.971709909755847e-06,
+      "loss": 1.0706,
+      "step": 2309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9708308253952085e-06,
+      "loss": 0.6824,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.969951495585578e-06,
+      "loss": 1.0736,
+      "step": 2311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.969071920549271e-06,
+      "loss": 1.0918,
+      "step": 2312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9681921005086653e-06,
+      "loss": 1.122,
+      "step": 2313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.967312035686201e-06,
+      "loss": 1.0608,
+      "step": 2314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9664317263043785e-06,
+      "loss": 1.045,
+      "step": 2315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96555117258576e-06,
+      "loss": 1.0844,
+      "step": 2316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9646703747529717e-06,
+      "loss": 1.1044,
+      "step": 2317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9637893330286997e-06,
+      "loss": 1.061,
+      "step": 2318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.962908047635692e-06,
+      "loss": 1.0946,
+      "step": 2319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.962026518796758e-06,
+      "loss": 1.0776,
+      "step": 2320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96114474673477e-06,
+      "loss": 1.0885,
+      "step": 2321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96026273167266e-06,
+      "loss": 0.704,
+      "step": 2322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9593804738334226e-06,
+      "loss": 1.0657,
+      "step": 2323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9584979734401135e-06,
+      "loss": 1.0349,
+      "step": 2324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9576152307158494e-06,
+      "loss": 1.0764,
+      "step": 2325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9567322458838094e-06,
+      "loss": 1.1279,
+      "step": 2326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9558490191672316e-06,
+      "loss": 1.104,
+      "step": 2327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.954965550789417e-06,
+      "loss": 1.0664,
+      "step": 2328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9540818409737287e-06,
+      "loss": 0.6718,
+      "step": 2329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9531978899435873e-06,
+      "loss": 1.073,
+      "step": 2330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9523136979224783e-06,
+      "loss": 1.0688,
+      "step": 2331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.951429265133945e-06,
+      "loss": 1.0715,
+      "step": 2332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9505445918015935e-06,
+      "loss": 1.0461,
+      "step": 2333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.94965967814909e-06,
+      "loss": 1.071,
+      "step": 2334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9487745244001625e-06,
+      "loss": 1.0705,
+      "step": 2335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9478891307785973e-06,
+      "loss": 1.0537,
+      "step": 2336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.947003497508244e-06,
+      "loss": 1.087,
+      "step": 2337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.946117624813011e-06,
+      "loss": 1.1007,
+      "step": 2338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9452315129168684e-06,
+      "loss": 1.0394,
+      "step": 2339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.944345162043846e-06,
+      "loss": 1.0655,
+      "step": 2340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.943458572418034e-06,
+      "loss": 1.0867,
+      "step": 2341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.942571744263583e-06,
+      "loss": 1.102,
+      "step": 2342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9416846778047054e-06,
+      "loss": 1.0764,
+      "step": 2343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9407973732656714e-06,
+      "loss": 1.1101,
+      "step": 2344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9399098308708124e-06,
+      "loss": 1.0927,
+      "step": 2345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.939022050844521e-06,
+      "loss": 1.0689,
+      "step": 2346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9381340334112484e-06,
+      "loss": 1.0678,
+      "step": 2347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9372457787955074e-06,
+      "loss": 1.0414,
+      "step": 2348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9363572872218686e-06,
+      "loss": 0.6935,
+      "step": 2349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.935468558914963e-06,
+      "loss": 0.6626,
+      "step": 2350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.934579594099485e-06,
+      "loss": 1.0522,
+      "step": 2351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9336903930001827e-06,
+      "loss": 1.0859,
+      "step": 2352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.932800955841869e-06,
+      "loss": 1.0816,
+      "step": 2353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9319112828494143e-06,
+      "loss": 1.0631,
+      "step": 2354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.931021374247749e-06,
+      "loss": 1.0694,
+      "step": 2355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.930131230261863e-06,
+      "loss": 1.0972,
+      "step": 2356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9292408511168053e-06,
+      "loss": 1.0803,
+      "step": 2357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9283502370376853e-06,
+      "loss": 1.0892,
+      "step": 2358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9274593882496723e-06,
+      "loss": 1.092,
+      "step": 2359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9265683049779917e-06,
+      "loss": 1.0768,
+      "step": 2360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9256769874479316e-06,
+      "loss": 1.1079,
+      "step": 2361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9247854358848383e-06,
+      "loss": 1.075,
+      "step": 2362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.923893650514117e-06,
+      "loss": 1.0777,
+      "step": 2363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9230016315612316e-06,
+      "loss": 1.061,
+      "step": 2364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9221093792517066e-06,
+      "loss": 1.0844,
+      "step": 2365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.921216893811124e-06,
+      "loss": 1.095,
+      "step": 2366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9203241754651244e-06,
+      "loss": 1.0541,
+      "step": 2367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.91943122443941e-06,
+      "loss": 1.0412,
+      "step": 2368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.918538040959738e-06,
+      "loss": 1.0033,
+      "step": 2369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9176446252519273e-06,
+      "loss": 1.1159,
+      "step": 2370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9167509775418543e-06,
+      "loss": 0.6917,
+      "step": 2371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9158570980554547e-06,
+      "loss": 1.1033,
+      "step": 2372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9149629870187217e-06,
+      "loss": 1.0641,
+      "step": 2373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.914068644657708e-06,
+      "loss": 1.0715,
+      "step": 2374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9131740711985246e-06,
+      "loss": 1.0661,
+      "step": 2375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9122792668673406e-06,
+      "loss": 1.0933,
+      "step": 2376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.911384231890383e-06,
+      "loss": 0.6585,
+      "step": 2377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.91048896649394e-06,
+      "loss": 1.0921,
+      "step": 2378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9095934709043537e-06,
+      "loss": 1.0766,
+      "step": 2379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9086977453480276e-06,
+      "loss": 1.0974,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.907801790051422e-06,
+      "loss": 1.0218,
+      "step": 2381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.906905605241056e-06,
+      "loss": 1.1247,
+      "step": 2382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9060091911435054e-06,
+      "loss": 1.0676,
+      "step": 2383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.905112547985406e-06,
+      "loss": 1.0876,
+      "step": 2384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9042156759934492e-06,
+      "loss": 1.0956,
+      "step": 2385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.903318575394386e-06,
+      "loss": 1.0713,
+      "step": 2386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9024212464150255e-06,
+      "loss": 1.1139,
+      "step": 2387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.901523689282232e-06,
+      "loss": 1.101,
+      "step": 2388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9006259042229308e-06,
+      "loss": 1.0597,
+      "step": 2389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.899727891464103e-06,
+      "loss": 1.1188,
+      "step": 2390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.898829651232786e-06,
+      "loss": 1.0547,
+      "step": 2391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8979311837560772e-06,
+      "loss": 1.0914,
+      "step": 2392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8970324892611305e-06,
+      "loss": 1.0598,
+      "step": 2393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.896133567975157e-06,
+      "loss": 1.1206,
+      "step": 2394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.895234420125425e-06,
+      "loss": 1.0788,
+      "step": 2395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.894335045939261e-06,
+      "loss": 1.1096,
+      "step": 2396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8934354456440473e-06,
+      "loss": 1.089,
+      "step": 2397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8925356194672246e-06,
+      "loss": 1.0736,
+      "step": 2398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.89163556763629e-06,
+      "loss": 1.0739,
+      "step": 2399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8907352903787974e-06,
+      "loss": 0.7138,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8898347879223595e-06,
+      "loss": 1.0933,
+      "step": 2401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.888934060494644e-06,
+      "loss": 1.0813,
+      "step": 2402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.888033108323375e-06,
+      "loss": 1.1086,
+      "step": 2403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.887131931636337e-06,
+      "loss": 1.0787,
+      "step": 2404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8862305306613663e-06,
+      "loss": 1.0897,
+      "step": 2405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.885328905626359e-06,
+      "loss": 1.0713,
+      "step": 2406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.884427056759268e-06,
+      "loss": 1.0937,
+      "step": 2407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8835249842881017e-06,
+      "loss": 0.6799,
+      "step": 2408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8826226884409254e-06,
+      "loss": 1.0891,
+      "step": 2409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8817201694458604e-06,
+      "loss": 1.0824,
+      "step": 2410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8808174275310852e-06,
+      "loss": 1.1039,
+      "step": 2411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.879914462924834e-06,
+      "loss": 1.026,
+      "step": 2412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.879011275855398e-06,
+      "loss": 1.0898,
+      "step": 2413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8781078665511245e-06,
+      "loss": 1.0838,
+      "step": 2414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8772042352404163e-06,
+      "loss": 1.1079,
+      "step": 2415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.876300382151732e-06,
+      "loss": 1.0775,
+      "step": 2416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.875396307513588e-06,
+      "loss": 1.0812,
+      "step": 2417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.874492011554556e-06,
+      "loss": 1.0494,
+      "step": 2418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8735874945032626e-06,
+      "loss": 1.0581,
+      "step": 2419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.872682756588392e-06,
+      "loss": 1.0714,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8717777980386826e-06,
+      "loss": 1.0216,
+      "step": 2421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.870872619082929e-06,
+      "loss": 1.0576,
+      "step": 2422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8699672199499833e-06,
+      "loss": 1.073,
+      "step": 2423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8690616008687507e-06,
+      "loss": 1.1197,
+      "step": 2424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8681557620681927e-06,
+      "loss": 1.0901,
+      "step": 2425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.867249703777328e-06,
+      "loss": 1.0666,
+      "step": 2426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8663434262252288e-06,
+      "loss": 1.087,
+      "step": 2427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8654369296410233e-06,
+      "loss": 1.0717,
+      "step": 2428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8645302142538956e-06,
+      "loss": 1.0484,
+      "step": 2429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.863623280293085e-06,
+      "loss": 1.0687,
+      "step": 2430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8627161279878864e-06,
+      "loss": 1.0543,
+      "step": 2431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8618087575676474e-06,
+      "loss": 1.0891,
+      "step": 2432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8609011692617746e-06,
+      "loss": 1.0536,
+      "step": 2433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.859993363299727e-06,
+      "loss": 1.0924,
+      "step": 2434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8590853399110197e-06,
+      "loss": 1.0713,
+      "step": 2435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.858177099325222e-06,
+      "loss": 1.0759,
+      "step": 2436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.857268641771959e-06,
+      "loss": 1.1128,
+      "step": 2437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8563599674809106e-06,
+      "loss": 1.0496,
+      "step": 2438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8554510766818103e-06,
+      "loss": 1.0901,
+      "step": 2439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8545419696044474e-06,
+      "loss": 0.6627,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.853632646478666e-06,
+      "loss": 0.7041,
+      "step": 2441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.852723107534365e-06,
+      "loss": 1.0695,
+      "step": 2442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.851813353001496e-06,
+      "loss": 1.0519,
+      "step": 2443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8509033831100675e-06,
+      "loss": 1.0841,
+      "step": 2444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.849993198090141e-06,
+      "loss": 1.1084,
+      "step": 2445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8490827981718325e-06,
+      "loss": 1.082,
+      "step": 2446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8481721835853134e-06,
+      "loss": 0.6633,
+      "step": 2447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8472613545608083e-06,
+      "loss": 1.0979,
+      "step": 2448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8463503113285957e-06,
+      "loss": 1.0225,
+      "step": 2449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.84543905411901e-06,
+      "loss": 0.6796,
+      "step": 2450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8445275831624375e-06,
+      "loss": 1.1138,
+      "step": 2451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.843615898689321e-06,
+      "loss": 1.0192,
+      "step": 2452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.842704000930154e-06,
+      "loss": 0.6722,
+      "step": 2453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.841791890115487e-06,
+      "loss": 0.6769,
+      "step": 2454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8408795664759232e-06,
+      "loss": 1.0852,
+      "step": 2455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8399670302421197e-06,
+      "loss": 1.0827,
+      "step": 2456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8390542816447866e-06,
+      "loss": 1.0687,
+      "step": 2457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8381413209146884e-06,
+      "loss": 1.0605,
+      "step": 2458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8372281482826436e-06,
+      "loss": 1.049,
+      "step": 2459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8363147639795234e-06,
+      "loss": 1.0446,
+      "step": 2460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8354011682362536e-06,
+      "loss": 1.0606,
+      "step": 2461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8344873612838123e-06,
+      "loss": 1.0851,
+      "step": 2462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.833573343353231e-06,
+      "loss": 1.0857,
+      "step": 2463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.832659114675597e-06,
+      "loss": 1.0875,
+      "step": 2464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8317446754820468e-06,
+      "loss": 1.0536,
+      "step": 2465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.830830026003773e-06,
+      "loss": 1.0797,
+      "step": 2466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8299151664720205e-06,
+      "loss": 1.0809,
+      "step": 2467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8290000971180884e-06,
+      "loss": 1.0514,
+      "step": 2468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.828084818173326e-06,
+      "loss": 1.1012,
+      "step": 2469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.827169329869139e-06,
+      "loss": 1.0849,
+      "step": 2470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8262536324369838e-06,
+      "loss": 1.0552,
+      "step": 2471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8253377261083712e-06,
+      "loss": 1.0987,
+      "step": 2472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8244216111148625e-06,
+      "loss": 1.1008,
+      "step": 2473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.823505287688075e-06,
+      "loss": 1.0746,
+      "step": 2474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8225887560596758e-06,
+      "loss": 1.0819,
+      "step": 2475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8216720164613858e-06,
+      "loss": 1.0354,
+      "step": 2476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8207550691249786e-06,
+      "loss": 1.0626,
+      "step": 2477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8198379142822803e-06,
+      "loss": 1.0759,
+      "step": 2478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8189205521651687e-06,
+      "loss": 0.6753,
+      "step": 2479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8180029830055763e-06,
+      "loss": 0.6905,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8170852070354847e-06,
+      "loss": 1.0688,
+      "step": 2481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8161672244869295e-06,
+      "loss": 1.062,
+      "step": 2482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8152490355919996e-06,
+      "loss": 1.0828,
+      "step": 2483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8143306405828333e-06,
+      "loss": 1.1213,
+      "step": 2484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8134120396916233e-06,
+      "loss": 1.0582,
+      "step": 2485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.812493233150614e-06,
+      "loss": 1.0777,
+      "step": 2486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8115742211921006e-06,
+      "loss": 1.0519,
+      "step": 2487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.810655004048432e-06,
+      "loss": 1.0161,
+      "step": 2488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8097355819520068e-06,
+      "loss": 1.0725,
+      "step": 2489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8088159551352776e-06,
+      "loss": 1.0825,
+      "step": 2490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8078961238307477e-06,
+      "loss": 1.0944,
+      "step": 2491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8069760882709722e-06,
+      "loss": 1.0746,
+      "step": 2492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8060558486885582e-06,
+      "loss": 1.0596,
+      "step": 2493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.805135405316163e-06,
+      "loss": 1.085,
+      "step": 2494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.804214758386497e-06,
+      "loss": 1.027,
+      "step": 2495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8032939081323216e-06,
+      "loss": 1.0993,
+      "step": 2496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8023728547864494e-06,
+      "loss": 1.0936,
+      "step": 2497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.801451598581745e-06,
+      "loss": 1.0501,
+      "step": 2498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.800530139751123e-06,
+      "loss": 1.0634,
+      "step": 2499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79960847852755e-06,
+      "loss": 1.0903,
+      "step": 2500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.798686615144044e-06,
+      "loss": 1.0986,
+      "step": 2501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7977645498336733e-06,
+      "loss": 1.0665,
+      "step": 2502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79684228282956e-06,
+      "loss": 1.0901,
+      "step": 2503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7959198143648724e-06,
+      "loss": 1.0702,
+      "step": 2504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.794997144672834e-06,
+      "loss": 1.0595,
+      "step": 2505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.794074273986716e-06,
+      "loss": 1.049,
+      "step": 2506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.793151202539843e-06,
+      "loss": 1.0614,
+      "step": 2507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7922279305655893e-06,
+      "loss": 1.0434,
+      "step": 2508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79130445829738e-06,
+      "loss": 1.0481,
+      "step": 2509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79038078596869e-06,
+      "loss": 1.0622,
+      "step": 2510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7894569138130453e-06,
+      "loss": 1.0662,
+      "step": 2511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7885328420640242e-06,
+      "loss": 0.6902,
+      "step": 2512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.787608570955252e-06,
+      "loss": 1.0652,
+      "step": 2513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7866841007204086e-06,
+      "loss": 1.0876,
+      "step": 2514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7857594315932195e-06,
+      "loss": 1.0634,
+      "step": 2515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7848345638074648e-06,
+      "loss": 1.0844,
+      "step": 2516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7839094975969716e-06,
+      "loss": 1.0231,
+      "step": 2517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.782984233195619e-06,
+      "loss": 1.0456,
+      "step": 2518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.782058770837336e-06,
+      "loss": 1.0429,
+      "step": 2519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7811331107561005e-06,
+      "loss": 1.0715,
+      "step": 2520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.780207253185943e-06,
+      "loss": 1.0541,
+      "step": 2521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7792811983609402e-06,
+      "loss": 1.0897,
+      "step": 2522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.778354946515222e-06,
+      "loss": 1.0543,
+      "step": 2523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.777428497882966e-06,
+      "loss": 1.0422,
+      "step": 2524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.776501852698401e-06,
+      "loss": 1.0791,
+      "step": 2525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.775575011195804e-06,
+      "loss": 1.0999,
+      "step": 2526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7746479736095037e-06,
+      "loss": 1.0767,
+      "step": 2527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7737207401738756e-06,
+      "loss": 1.0694,
+      "step": 2528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7727933111233474e-06,
+      "loss": 1.0744,
+      "step": 2529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7718656866923943e-06,
+      "loss": 1.0546,
+      "step": 2530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.770937867115542e-06,
+      "loss": 1.1116,
+      "step": 2531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.770009852627366e-06,
+      "loss": 1.0982,
+      "step": 2532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.769081643462489e-06,
+      "loss": 0.9857,
+      "step": 2533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.768153239855585e-06,
+      "loss": 1.0615,
+      "step": 2534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.767224642041376e-06,
+      "loss": 1.0548,
+      "step": 2535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7662958502546347e-06,
+      "loss": 1.0979,
+      "step": 2536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7653668647301796e-06,
+      "loss": 1.0638,
+      "step": 2537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.764437685702882e-06,
+      "loss": 0.6936,
+      "step": 2538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7635083134076586e-06,
+      "loss": 1.1011,
+      "step": 2539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7625787480794784e-06,
+      "loss": 1.0905,
+      "step": 2540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.761648989953356e-06,
+      "loss": 1.0821,
+      "step": 2541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.760719039264358e-06,
+      "loss": 1.0991,
+      "step": 2542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7597888962475964e-06,
+      "loss": 1.0456,
+      "step": 2543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.758858561138234e-06,
+      "loss": 0.6891,
+      "step": 2544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.757928034171481e-06,
+      "loss": 1.0568,
+      "step": 2545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.756997315582598e-06,
+      "loss": 1.0871,
+      "step": 2546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7560664056068912e-06,
+      "loss": 1.0795,
+      "step": 2547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.755135304479717e-06,
+      "loss": 1.0724,
+      "step": 2548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7542040124364807e-06,
+      "loss": 1.0913,
+      "step": 2549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.753272529712633e-06,
+      "loss": 1.0606,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7523408565436764e-06,
+      "loss": 1.0961,
+      "step": 2551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7514089931651595e-06,
+      "loss": 1.0788,
+      "step": 2552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7504769398126795e-06,
+      "loss": 1.1106,
+      "step": 2553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7495446967218813e-06,
+      "loss": 1.0671,
+      "step": 2554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7486122641284583e-06,
+      "loss": 0.7006,
+      "step": 2555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.747679642268151e-06,
+      "loss": 1.0673,
+      "step": 2556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7467468313767484e-06,
+      "loss": 1.076,
+      "step": 2557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7458138316900882e-06,
+      "loss": 1.038,
+      "step": 2558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.744880643444053e-06,
+      "loss": 0.6725,
+      "step": 2559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.743947266874576e-06,
+      "loss": 1.1206,
+      "step": 2560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.743013702217637e-06,
+      "loss": 1.112,
+      "step": 2561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.742079949709262e-06,
+      "loss": 1.057,
+      "step": 2562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7411460095855274e-06,
+      "loss": 1.0983,
+      "step": 2563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7402118820825545e-06,
+      "loss": 0.6802,
+      "step": 2564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7392775674365123e-06,
+      "loss": 1.0724,
+      "step": 2565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7383430658836188e-06,
+      "loss": 1.0656,
+      "step": 2566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7374083776601373e-06,
+      "loss": 1.0515,
+      "step": 2567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7364735030023786e-06,
+      "loss": 1.0711,
+      "step": 2568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7355384421467025e-06,
+      "loss": 1.08,
+      "step": 2569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7346031953295135e-06,
+      "loss": 1.0735,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7336677627872645e-06,
+      "loss": 1.0331,
+      "step": 2571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7327321447564545e-06,
+      "loss": 1.0527,
+      "step": 2572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7317963414736308e-06,
+      "loss": 1.047,
+      "step": 2573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.730860353175386e-06,
+      "loss": 1.1017,
+      "step": 2574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7299241800983595e-06,
+      "loss": 0.6528,
+      "step": 2575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7289878224792387e-06,
+      "loss": 1.1094,
+      "step": 2576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.728051280554758e-06,
+      "loss": 0.688,
+      "step": 2577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.727114554561695e-06,
+      "loss": 1.0695,
+      "step": 2578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7261776447368772e-06,
+      "loss": 1.0759,
+      "step": 2579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7252405513171787e-06,
+      "loss": 1.0461,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7243032745395173e-06,
+      "loss": 1.0979,
+      "step": 2581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7233658146408595e-06,
+      "loss": 1.0872,
+      "step": 2582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7224281718582165e-06,
+      "loss": 1.0571,
+      "step": 2583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7214903464286483e-06,
+      "loss": 1.0454,
+      "step": 2584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.720552338589258e-06,
+      "loss": 1.0705,
+      "step": 2585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.719614148577196e-06,
+      "loss": 1.0683,
+      "step": 2586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.71867577662966e-06,
+      "loss": 0.6449,
+      "step": 2587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.717737222983892e-06,
+      "loss": 1.0758,
+      "step": 2588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.71679848787718e-06,
+      "loss": 0.6757,
+      "step": 2589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7158595715468586e-06,
+      "loss": 0.6452,
+      "step": 2590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7149204742303095e-06,
+      "loss": 1.0392,
+      "step": 2591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7139811961649573e-06,
+      "loss": 1.0527,
+      "step": 2592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.713041737588274e-06,
+      "loss": 1.1025,
+      "step": 2593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7121020987377767e-06,
+      "loss": 1.0476,
+      "step": 2594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.711162279851029e-06,
+      "loss": 1.116,
+      "step": 2595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.710222281165639e-06,
+      "loss": 1.0817,
+      "step": 2596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7092821029192604e-06,
+      "loss": 1.0517,
+      "step": 2597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7083417453495926e-06,
+      "loss": 1.0832,
+      "step": 2598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7074012086943805e-06,
+      "loss": 1.0489,
+      "step": 2599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7064604931914138e-06,
+      "loss": 1.0722,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7055195990785274e-06,
+      "loss": 1.0448,
+      "step": 2601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.704578526593602e-06,
+      "loss": 1.0431,
+      "step": 2602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7036372759745627e-06,
+      "loss": 1.0597,
+      "step": 2603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.70269584745938e-06,
+      "loss": 0.6622,
+      "step": 2604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7017542412860693e-06,
+      "loss": 0.6907,
+      "step": 2605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7008124576926906e-06,
+      "loss": 1.0211,
+      "step": 2606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6998704969173505e-06,
+      "loss": 1.0544,
+      "step": 2607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.698928359198197e-06,
+      "loss": 1.0555,
+      "step": 2608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.697986044773425e-06,
+      "loss": 1.0762,
+      "step": 2609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6970435538812756e-06,
+      "loss": 0.678,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.696100886760031e-06,
+      "loss": 1.0496,
+      "step": 2611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.695158043648021e-06,
+      "loss": 1.0668,
+      "step": 2612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6942150247836175e-06,
+      "loss": 1.0729,
+      "step": 2613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6932718304052386e-06,
+      "loss": 1.0671,
+      "step": 2614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6923284607513465e-06,
+      "loss": 1.032,
+      "step": 2615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.691384916060446e-06,
+      "loss": 1.0591,
+      "step": 2616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6904411965710896e-06,
+      "loss": 1.0543,
+      "step": 2617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6894973025218705e-06,
+      "loss": 1.0511,
+      "step": 2618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6885532341514274e-06,
+      "loss": 1.0866,
+      "step": 2619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.687608991698443e-06,
+      "loss": 0.6446,
+      "step": 2620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.686664575401645e-06,
+      "loss": 1.0875,
+      "step": 2621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6857199854998036e-06,
+      "loss": 1.0763,
+      "step": 2622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6847752222317336e-06,
+      "loss": 1.1137,
+      "step": 2623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6838302858362935e-06,
+      "loss": 0.6594,
+      "step": 2624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.682885176552385e-06,
+      "loss": 0.7081,
+      "step": 2625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6819398946189555e-06,
+      "loss": 1.0907,
+      "step": 2626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6809944402749935e-06,
+      "loss": 1.0888,
+      "step": 2627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.680048813759532e-06,
+      "loss": 0.9964,
+      "step": 2628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6791030153116483e-06,
+      "loss": 1.0676,
+      "step": 2629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6781570451704624e-06,
+      "loss": 1.0576,
+      "step": 2630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.677210903575138e-06,
+      "loss": 1.0205,
+      "step": 2631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.676264590764883e-06,
+      "loss": 1.0641,
+      "step": 2632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.675318106978946e-06,
+      "loss": 0.6549,
+      "step": 2633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6743714524566206e-06,
+      "loss": 1.0931,
+      "step": 2634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.673424627437245e-06,
+      "loss": 0.6412,
+      "step": 2635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6724776321601976e-06,
+      "loss": 1.1195,
+      "step": 2636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.671530466864902e-06,
+      "loss": 1.0499,
+      "step": 2637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6705831317908236e-06,
+      "loss": 1.0604,
+      "step": 2638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.669635627177471e-06,
+      "loss": 0.6948,
+      "step": 2639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.668687953264396e-06,
+      "loss": 1.0664,
+      "step": 2640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.667740110291193e-06,
+      "loss": 1.0988,
+      "step": 2641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6667920984974984e-06,
+      "loss": 1.0926,
+      "step": 2642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6658439181229932e-06,
+      "loss": 1.0262,
+      "step": 2643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6648955694073987e-06,
+      "loss": 1.0665,
+      "step": 2644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6639470525904807e-06,
+      "loss": 1.0528,
+      "step": 2645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.662998367912047e-06,
+      "loss": 1.0133,
+      "step": 2646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.662049515611946e-06,
+      "loss": 1.0347,
+      "step": 2647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6611004959300714e-06,
+      "loss": 1.083,
+      "step": 2648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.660151309106357e-06,
+      "loss": 1.0616,
+      "step": 2649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6592019553807806e-06,
+      "loss": 1.0997,
+      "step": 2650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6582524349933605e-06,
+      "loss": 1.0793,
+      "step": 2651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.657302748184158e-06,
+      "loss": 1.0903,
+      "step": 2652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.656352895193276e-06,
+      "loss": 1.0452,
+      "step": 2653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6554028762608606e-06,
+      "loss": 1.0392,
+      "step": 2654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6544526916270985e-06,
+      "loss": 1.0827,
+      "step": 2655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.653502341532219e-06,
+      "loss": 1.0819,
+      "step": 2656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.652551826216494e-06,
+      "loss": 1.0717,
+      "step": 2657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6516011459202348e-06,
+      "loss": 1.0884,
+      "step": 2658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6506503008837967e-06,
+      "loss": 1.0505,
+      "step": 2659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6496992913475756e-06,
+      "loss": 1.0819,
+      "step": 2660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.648748117552009e-06,
+      "loss": 1.0783,
+      "step": 2661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6477967797375765e-06,
+      "loss": 0.6884,
+      "step": 2662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6468452781447985e-06,
+      "loss": 1.1083,
+      "step": 2663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6458936130142365e-06,
+      "loss": 1.0413,
+      "step": 2664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6449417845864956e-06,
+      "loss": 1.0647,
+      "step": 2665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.643989793102219e-06,
+      "loss": 1.0729,
+      "step": 2666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6430376388020935e-06,
+      "loss": 1.0335,
+      "step": 2667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.642085321926846e-06,
+      "loss": 1.0547,
+      "step": 2668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.641132842717245e-06,
+      "loss": 1.0333,
+      "step": 2669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.640180201414099e-06,
+      "loss": 1.0718,
+      "step": 2670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6392273982582592e-06,
+      "loss": 1.0705,
+      "step": 2671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6382744334906156e-06,
+      "loss": 1.0546,
+      "step": 2672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6373213073521015e-06,
+      "loss": 1.0599,
+      "step": 2673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6363680200836893e-06,
+      "loss": 1.0489,
+      "step": 2674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6354145719263917e-06,
+      "loss": 0.695,
+      "step": 2675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.634460963121264e-06,
+      "loss": 1.1089,
+      "step": 2676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.633507193909401e-06,
+      "loss": 1.0573,
+      "step": 2677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6325532645319374e-06,
+      "loss": 0.6336,
+      "step": 2678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.63159917523005e-06,
+      "loss": 1.0481,
+      "step": 2679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6306449262449543e-06,
+      "loss": 1.0843,
+      "step": 2680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.629690517817907e-06,
+      "loss": 1.0838,
+      "step": 2681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.628735950190207e-06,
+      "loss": 1.0948,
+      "step": 2682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6277812236031885e-06,
+      "loss": 1.058,
+      "step": 2683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.626826338298231e-06,
+      "loss": 1.0648,
+      "step": 2684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6258712945167515e-06,
+      "loss": 0.6714,
+      "step": 2685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.624916092500208e-06,
+      "loss": 1.1002,
+      "step": 2686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.623960732490098e-06,
+      "loss": 1.0494,
+      "step": 2687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.623005214727959e-06,
+      "loss": 1.0318,
+      "step": 2688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.622049539455368e-06,
+      "loss": 0.6355,
+      "step": 2689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.621093706913944e-06,
+      "loss": 1.0573,
+      "step": 2690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6201377173453417e-06,
+      "loss": 1.0886,
+      "step": 2691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.61918157099126e-06,
+      "loss": 1.083,
+      "step": 2692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6182252680934336e-06,
+      "loss": 1.056,
+      "step": 2693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6172688088936395e-06,
+      "loss": 1.0885,
+      "step": 2694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6163121936336932e-06,
+      "loss": 0.6805,
+      "step": 2695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6153554225554493e-06,
+      "loss": 1.0226,
+      "step": 2696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6143984959008022e-06,
+      "loss": 1.0558,
+      "step": 2697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.613441413911686e-06,
+      "loss": 1.0515,
+      "step": 2698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.612484176830073e-06,
+      "loss": 1.017,
+      "step": 2699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6115267848979757e-06,
+      "loss": 1.1218,
+      "step": 2700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.610569238357446e-06,
+      "loss": 1.0729,
+      "step": 2701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.609611537450573e-06,
+      "loss": 1.0728,
+      "step": 2702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.608653682419487e-06,
+      "loss": 1.0628,
+      "step": 2703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.607695673506357e-06,
+      "loss": 0.6798,
+      "step": 2704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6067375109533892e-06,
+      "loss": 1.0674,
+      "step": 2705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6057791950028296e-06,
+      "loss": 1.0605,
+      "step": 2706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.604820725896964e-06,
+      "loss": 1.0523,
+      "step": 2707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6038621038781157e-06,
+      "loss": 1.0811,
+      "step": 2708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.602903329188647e-06,
+      "loss": 1.0342,
+      "step": 2709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6019444020709597e-06,
+      "loss": 1.0761,
+      "step": 2710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6009853227674915e-06,
+      "loss": 1.0439,
+      "step": 2711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6000260915207215e-06,
+      "loss": 1.0828,
+      "step": 2712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5990667085731658e-06,
+      "loss": 1.0479,
+      "step": 2713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5981071741673794e-06,
+      "loss": 1.0439,
+      "step": 2714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5971474885459546e-06,
+      "loss": 1.0448,
+      "step": 2715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5961876519515234e-06,
+      "loss": 1.0836,
+      "step": 2716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.595227664626754e-06,
+      "loss": 1.0658,
+      "step": 2717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.594267526814355e-06,
+      "loss": 1.0724,
+      "step": 2718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5933072387570717e-06,
+      "loss": 1.0801,
+      "step": 2719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.592346800697688e-06,
+      "loss": 1.0667,
+      "step": 2720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5913862128790244e-06,
+      "loss": 1.0737,
+      "step": 2721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.590425475543941e-06,
+      "loss": 1.0575,
+      "step": 2722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5894645889353347e-06,
+      "loss": 1.0551,
+      "step": 2723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.58850355329614e-06,
+      "loss": 1.0842,
+      "step": 2724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5875423688693296e-06,
+      "loss": 1.0952,
+      "step": 2725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.586581035897914e-06,
+      "loss": 1.0655,
+      "step": 2726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5856195546249408e-06,
+      "loss": 1.0893,
+      "step": 2727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5846579252934954e-06,
+      "loss": 1.0623,
+      "step": 2728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5836961481466997e-06,
+      "loss": 1.0723,
+      "step": 2729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5827342234277144e-06,
+      "loss": 1.0482,
+      "step": 2730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.581772151379737e-06,
+      "loss": 0.6848,
+      "step": 2731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5808099322460008e-06,
+      "loss": 1.0585,
+      "step": 2732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5798475662697785e-06,
+      "loss": 0.6478,
+      "step": 2733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5788850536943797e-06,
+      "loss": 0.6802,
+      "step": 2734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5779223947631488e-06,
+      "loss": 1.092,
+      "step": 2735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5769595897194693e-06,
+      "loss": 1.0449,
+      "step": 2736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5759966388067615e-06,
+      "loss": 1.0925,
+      "step": 2737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5750335422684817e-06,
+      "loss": 1.091,
+      "step": 2738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5740703003481245e-06,
+      "loss": 1.1091,
+      "step": 2739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5731069132892185e-06,
+      "loss": 1.08,
+      "step": 2740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.572143381335332e-06,
+      "loss": 1.0642,
+      "step": 2741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5711797047300684e-06,
+      "loss": 1.0761,
+      "step": 2742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5702158837170677e-06,
+      "loss": 1.0415,
+      "step": 2743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5692519185400064e-06,
+      "loss": 1.0536,
+      "step": 2744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5682878094425986e-06,
+      "loss": 1.0738,
+      "step": 2745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5673235566685935e-06,
+      "loss": 1.0583,
+      "step": 2746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5663591604617763e-06,
+      "loss": 1.0718,
+      "step": 2747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5653946210659702e-06,
+      "loss": 1.0647,
+      "step": 2748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5644299387250326e-06,
+      "loss": 1.0715,
+      "step": 2749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5634651136828594e-06,
+      "loss": 1.0545,
+      "step": 2750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.56250014618338e-06,
+      "loss": 0.6724,
+      "step": 2751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5615350364705607e-06,
+      "loss": 1.0489,
+      "step": 2752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5605697847884053e-06,
+      "loss": 1.0054,
+      "step": 2753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5596043913809518e-06,
+      "loss": 1.0761,
+      "step": 2754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.558638856492274e-06,
+      "loss": 0.6568,
+      "step": 2755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.557673180366482e-06,
+      "loss": 1.0905,
+      "step": 2756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5567073632477217e-06,
+      "loss": 0.6838,
+      "step": 2757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5557414053801748e-06,
+      "loss": 1.0826,
+      "step": 2758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5547753070080582e-06,
+      "loss": 0.6395,
+      "step": 2759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5538090683756243e-06,
+      "loss": 1.048,
+      "step": 2760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5528426897271614e-06,
+      "loss": 1.088,
+      "step": 2761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5518761713069924e-06,
+      "loss": 1.0641,
+      "step": 2762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5509095133594762e-06,
+      "loss": 1.1122,
+      "step": 2763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.549942716129007e-06,
+      "loss": 1.074,
+      "step": 2764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5489757798600133e-06,
+      "loss": 0.6552,
+      "step": 2765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.54800870479696e-06,
+      "loss": 1.1088,
+      "step": 2766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5470414911843463e-06,
+      "loss": 1.0798,
+      "step": 2767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5460741392667073e-06,
+      "loss": 1.0879,
+      "step": 2768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5451066492886124e-06,
+      "loss": 1.0565,
+      "step": 2769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.544139021494665e-06,
+      "loss": 1.1155,
+      "step": 2770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.543171256129505e-06,
+      "loss": 1.0657,
+      "step": 2771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5422033534378065e-06,
+      "loss": 1.0582,
+      "step": 2772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5412353136642775e-06,
+      "loss": 1.0763,
+      "step": 2773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.540267137053662e-06,
+      "loss": 1.0629,
+      "step": 2774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5392988238507375e-06,
+      "loss": 1.0994,
+      "step": 2775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5383303743003166e-06,
+      "loss": 1.0885,
+      "step": 2776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.537361788647246e-06,
+      "loss": 1.0594,
+      "step": 2777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5363930671364082e-06,
+      "loss": 0.6439,
+      "step": 2778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.535424210012717e-06,
+      "loss": 1.0733,
+      "step": 2779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5344552175211244e-06,
+      "loss": 1.0509,
+      "step": 2780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.533486089906614e-06,
+      "loss": 1.0384,
+      "step": 2781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.532516827414203e-06,
+      "loss": 1.0864,
+      "step": 2782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5315474302889455e-06,
+      "loss": 1.0437,
+      "step": 2783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.530577898775927e-06,
+      "loss": 1.0329,
+      "step": 2784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.529608233120268e-06,
+      "loss": 1.0771,
+      "step": 2785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5286384335671235e-06,
+      "loss": 1.0752,
+      "step": 2786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.527668500361681e-06,
+      "loss": 1.0799,
+      "step": 2787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5266984337491634e-06,
+      "loss": 1.0438,
+      "step": 2788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.525728233974826e-06,
+      "loss": 1.0582,
+      "step": 2789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.524757901283958e-06,
+      "loss": 1.0554,
+      "step": 2790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.523787435921883e-06,
+      "loss": 1.0476,
+      "step": 2791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5228168381339575e-06,
+      "loss": 0.6907,
+      "step": 2792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5218461081655713e-06,
+      "loss": 1.0716,
+      "step": 2793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5208752462621487e-06,
+      "loss": 1.0415,
+      "step": 2794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.519904252669145e-06,
+      "loss": 1.0667,
+      "step": 2795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.518933127632051e-06,
+      "loss": 1.076,
+      "step": 2796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5179618713963912e-06,
+      "loss": 1.0816,
+      "step": 2797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.516990484207721e-06,
+      "loss": 1.0566,
+      "step": 2798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.51601896631163e-06,
+      "loss": 0.6786,
+      "step": 2799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.515047317953741e-06,
+      "loss": 1.0241,
+      "step": 2800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.51407553937971e-06,
+      "loss": 0.7194,
+      "step": 2801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.513103630835226e-06,
+      "loss": 1.0621,
+      "step": 2802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5121315925660097e-06,
+      "loss": 1.0567,
+      "step": 2803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5111594248178154e-06,
+      "loss": 1.08,
+      "step": 2804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5101871278364303e-06,
+      "loss": 1.0497,
+      "step": 2805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.509214701867674e-06,
+      "loss": 1.0487,
+      "step": 2806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.508242147157399e-06,
+      "loss": 1.1033,
+      "step": 2807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5072694639514903e-06,
+      "loss": 1.0593,
+      "step": 2808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5062966524958647e-06,
+      "loss": 1.0691,
+      "step": 2809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.505323713036472e-06,
+      "loss": 0.6538,
+      "step": 2810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.504350645819295e-06,
+      "loss": 1.0743,
+      "step": 2811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.503377451090347e-06,
+      "loss": 1.0613,
+      "step": 2812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5024041290956753e-06,
+      "loss": 1.0596,
+      "step": 2813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5014306800813586e-06,
+      "loss": 1.0362,
+      "step": 2814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5004571042935078e-06,
+      "loss": 1.0428,
+      "step": 2815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4994834019782657e-06,
+      "loss": 1.0117,
+      "step": 2816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.498509573381808e-06,
+      "loss": 1.0413,
+      "step": 2817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4975356187503412e-06,
+      "loss": 1.0587,
+      "step": 2818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.496561538330103e-06,
+      "loss": 1.026,
+      "step": 2819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.495587332367366e-06,
+      "loss": 0.7146,
+      "step": 2820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4946130011084306e-06,
+      "loss": 1.0288,
+      "step": 2821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4936385447996323e-06,
+      "loss": 1.0489,
+      "step": 2822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4926639636873364e-06,
+      "loss": 1.0989,
+      "step": 2823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.491689258017939e-06,
+      "loss": 1.0753,
+      "step": 2824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.490714428037869e-06,
+      "loss": 1.0371,
+      "step": 2825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4897394739935874e-06,
+      "loss": 1.0523,
+      "step": 2826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.488764396131585e-06,
+      "loss": 1.0905,
+      "step": 2827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4877891946983847e-06,
+      "loss": 0.6293,
+      "step": 2828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4868138699405406e-06,
+      "loss": 1.0798,
+      "step": 2829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4858384221046376e-06,
+      "loss": 1.0981,
+      "step": 2830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.484862851437292e-06,
+      "loss": 1.0336,
+      "step": 2831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.483887158185151e-06,
+      "loss": 1.0623,
+      "step": 2832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4829113425948935e-06,
+      "loss": 1.0629,
+      "step": 2833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.481935404913229e-06,
+      "loss": 1.0288,
+      "step": 2834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4809593453868965e-06,
+      "loss": 0.6344,
+      "step": 2835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4799831642626673e-06,
+      "loss": 1.0295,
+      "step": 2836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.479006861787343e-06,
+      "loss": 1.0541,
+      "step": 2837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4780304382077568e-06,
+      "loss": 1.0806,
+      "step": 2838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4770538937707707e-06,
+      "loss": 1.0382,
+      "step": 2839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4760772287232788e-06,
+      "loss": 1.081,
+      "step": 2840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4751004433122047e-06,
+      "loss": 1.0427,
+      "step": 2841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4741235377845036e-06,
+      "loss": 1.0727,
+      "step": 2842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.473146512387159e-06,
+      "loss": 1.0417,
+      "step": 2843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4721693673671873e-06,
+      "loss": 1.0882,
+      "step": 2844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.471192102971634e-06,
+      "loss": 1.0695,
+      "step": 2845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4702147194475737e-06,
+      "loss": 1.0463,
+      "step": 2846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4692372170421117e-06,
+      "loss": 1.0833,
+      "step": 2847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.468259596002385e-06,
+      "loss": 1.0217,
+      "step": 2848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4672818565755584e-06,
+      "loss": 0.6814,
+      "step": 2849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.466303999008828e-06,
+      "loss": 1.0862,
+      "step": 2850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.46532602354942e-06,
+      "loss": 1.0118,
+      "step": 2851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.464347930444588e-06,
+      "loss": 1.0864,
+      "step": 2852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.463369719941618e-06,
+      "loss": 0.6649,
+      "step": 2853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.462391392287825e-06,
+      "loss": 1.0778,
+      "step": 2854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4614129477305527e-06,
+      "loss": 1.0711,
+      "step": 2855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4604343865171765e-06,
+      "loss": 0.6803,
+      "step": 2856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4594557088950984e-06,
+      "loss": 0.6587,
+      "step": 2857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4584769151117512e-06,
+      "loss": 1.087,
+      "step": 2858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4574980054145977e-06,
+      "loss": 1.0478,
+      "step": 2859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.456518980051129e-06,
+      "loss": 1.0644,
+      "step": 2860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4555398392688663e-06,
+      "loss": 1.0626,
+      "step": 2861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4545605833153597e-06,
+      "loss": 1.0783,
+      "step": 2862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4535812124381876e-06,
+      "loss": 1.0653,
+      "step": 2863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4526017268849586e-06,
+      "loss": 1.042,
+      "step": 2864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.451622126903309e-06,
+      "loss": 1.0543,
+      "step": 2865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.450642412740906e-06,
+      "loss": 1.0569,
+      "step": 2866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.449662584645445e-06,
+      "loss": 1.0713,
+      "step": 2867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4486826428646473e-06,
+      "loss": 1.0202,
+      "step": 2868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.447702587646267e-06,
+      "loss": 1.0587,
+      "step": 2869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.446722419238085e-06,
+      "loss": 1.0872,
+      "step": 2870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4457421378879116e-06,
+      "loss": 1.0158,
+      "step": 2871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.444761743843584e-06,
+      "loss": 1.0561,
+      "step": 2872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.44378123735297e-06,
+      "loss": 0.6672,
+      "step": 2873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4428006186639646e-06,
+      "loss": 1.1136,
+      "step": 2874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4418198880244905e-06,
+      "loss": 1.0107,
+      "step": 2875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4408390456825003e-06,
+      "loss": 1.0452,
+      "step": 2876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4398580918859743e-06,
+      "loss": 1.0589,
+      "step": 2877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4388770268829213e-06,
+      "loss": 1.0812,
+      "step": 2878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4378958509213766e-06,
+      "loss": 1.0548,
+      "step": 2879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.436914564249405e-06,
+      "loss": 1.0667,
+      "step": 2880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4359331671150995e-06,
+      "loss": 1.0697,
+      "step": 2881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4349516597665797e-06,
+      "loss": 1.0436,
+      "step": 2882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4339700424519944e-06,
+      "loss": 1.0648,
+      "step": 2883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43298831541952e-06,
+      "loss": 1.0682,
+      "step": 2884
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4320064789173593e-06,
+      "loss": 1.0305,
+      "step": 2885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.431024533193745e-06,
+      "loss": 1.0878,
+      "step": 2886
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.430042478496935e-06,
+      "loss": 1.0982,
+      "step": 2887
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.429060315075217e-06,
+      "loss": 1.0323,
+      "step": 2888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4280780431769043e-06,
+      "loss": 1.0315,
+      "step": 2889
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.427095663050339e-06,
+      "loss": 1.0814,
+      "step": 2890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.42611317494389e-06,
+      "loss": 1.0396,
+      "step": 2891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.425130579105953e-06,
+      "loss": 1.051,
+      "step": 2892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.424147875784952e-06,
+      "loss": 1.0473,
+      "step": 2893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.423165065229338e-06,
+      "loss": 1.0759,
+      "step": 2894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4221821476875887e-06,
+      "loss": 1.0499,
+      "step": 2895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4211991234082073e-06,
+      "loss": 1.0577,
+      "step": 2896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.420215992639727e-06,
+      "loss": 1.0784,
+      "step": 2897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4192327556307062e-06,
+      "loss": 1.0416,
+      "step": 2898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4182494126297305e-06,
+      "loss": 1.0176,
+      "step": 2899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.417265963885413e-06,
+      "loss": 1.0899,
+      "step": 2900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4162824096463913e-06,
+      "loss": 1.0699,
+      "step": 2901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.415298750161332e-06,
+      "loss": 1.0661,
+      "step": 2902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4143149856789278e-06,
+      "loss": 1.0556,
+      "step": 2903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.413331116447897e-06,
+      "loss": 1.0637,
+      "step": 2904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.412347142716985e-06,
+      "loss": 1.0762,
+      "step": 2905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.411363064734965e-06,
+      "loss": 1.0868,
+      "step": 2906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.410378882750633e-06,
+      "loss": 1.0148,
+      "step": 2907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4093945970128147e-06,
+      "loss": 1.0881,
+      "step": 2908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4084102077703608e-06,
+      "loss": 1.0257,
+      "step": 2909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.407425715272148e-06,
+      "loss": 1.0792,
+      "step": 2910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.40644111976708e-06,
+      "loss": 1.0545,
+      "step": 2911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.405456421504084e-06,
+      "loss": 1.0498,
+      "step": 2912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.404471620732117e-06,
+      "loss": 1.0635,
+      "step": 2913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4034867177001586e-06,
+      "loss": 1.0562,
+      "step": 2914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4025017126572164e-06,
+      "loss": 0.6696,
+      "step": 2915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4015166058523224e-06,
+      "loss": 1.0084,
+      "step": 2916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.400531397534536e-06,
+      "loss": 1.011,
+      "step": 2917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3995460879529397e-06,
+      "loss": 1.092,
+      "step": 2918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.398560677356643e-06,
+      "loss": 1.0556,
+      "step": 2919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.397575165994782e-06,
+      "loss": 1.0703,
+      "step": 2920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.396589554116517e-06,
+      "loss": 1.0494,
+      "step": 2921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3956038419710346e-06,
+      "loss": 1.0429,
+      "step": 2922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.394618029807545e-06,
+      "loss": 1.0786,
+      "step": 2923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.393632117875285e-06,
+      "loss": 1.0511,
+      "step": 2924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3926461064235173e-06,
+      "loss": 1.0652,
+      "step": 2925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.391659995701528e-06,
+      "loss": 1.0899,
+      "step": 2926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.390673785958629e-06,
+      "loss": 1.0729,
+      "step": 2927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.38968747744416e-06,
+      "loss": 1.069,
+      "step": 2928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3887010704074796e-06,
+      "loss": 1.0559,
+      "step": 2929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3877145650979765e-06,
+      "loss": 1.0613,
+      "step": 2930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3867279617650627e-06,
+      "loss": 1.0642,
+      "step": 2931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.385741260658174e-06,
+      "loss": 1.0697,
+      "step": 2932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3847544620267732e-06,
+      "loss": 0.6808,
+      "step": 2933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.383767566120345e-06,
+      "loss": 1.0905,
+      "step": 2934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3827805731884005e-06,
+      "loss": 1.0481,
+      "step": 2935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.381793483480475e-06,
+      "loss": 1.0804,
+      "step": 2936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.380806297246128e-06,
+      "loss": 1.0017,
+      "step": 2937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.379819014734943e-06,
+      "loss": 1.0437,
+      "step": 2938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3788316361965303e-06,
+      "loss": 1.0401,
+      "step": 2939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3778441618805203e-06,
+      "loss": 1.087,
+      "step": 2940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.376856592036571e-06,
+      "loss": 1.0587,
+      "step": 2941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3758689269143625e-06,
+      "loss": 1.051,
+      "step": 2942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3748811667636e-06,
+      "loss": 1.039,
+      "step": 2943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3738933118340147e-06,
+      "loss": 1.0758,
+      "step": 2944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3729053623753575e-06,
+      "loss": 1.0156,
+      "step": 2945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3719173186374055e-06,
+      "loss": 1.0822,
+      "step": 2946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.37092918086996e-06,
+      "loss": 1.0696,
+      "step": 2947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3699409493228463e-06,
+      "loss": 1.0496,
+      "step": 2948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3689526242459116e-06,
+      "loss": 1.015,
+      "step": 2949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.367964205889029e-06,
+      "loss": 1.0296,
+      "step": 2950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.366975694502093e-06,
+      "loss": 1.0566,
+      "step": 2951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.365987090335022e-06,
+      "loss": 1.0667,
+      "step": 2952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3649983936377605e-06,
+      "loss": 0.6489,
+      "step": 2953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3640096046602726e-06,
+      "loss": 1.0273,
+      "step": 2954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.363020723652549e-06,
+      "loss": 1.0172,
+      "step": 2955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3620317508646014e-06,
+      "loss": 1.0726,
+      "step": 2956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3610426865464654e-06,
+      "loss": 1.0768,
+      "step": 2957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3600535309482003e-06,
+      "loss": 1.0701,
+      "step": 2958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3590642843198875e-06,
+      "loss": 1.0488,
+      "step": 2959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3580749469116322e-06,
+      "loss": 1.0373,
+      "step": 2960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.357085518973563e-06,
+      "loss": 1.0262,
+      "step": 2961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.35609600075583e-06,
+      "loss": 1.0918,
+      "step": 2962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.355106392508607e-06,
+      "loss": 1.0721,
+      "step": 2963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3541166944820895e-06,
+      "loss": 1.0683,
+      "step": 2964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3531269069264973e-06,
+      "loss": 1.0411,
+      "step": 2965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3521370300920727e-06,
+      "loss": 1.0466,
+      "step": 2966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.351147064229079e-06,
+      "loss": 1.0657,
+      "step": 2967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3501570095878034e-06,
+      "loss": 1.0685,
+      "step": 2968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3491668664185543e-06,
+      "loss": 1.0863,
+      "step": 2969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.348176634971665e-06,
+      "loss": 1.058,
+      "step": 2970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3471863154974876e-06,
+      "loss": 1.0563,
+      "step": 2971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3461959082464003e-06,
+      "loss": 1.0633,
+      "step": 2972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3452054134687998e-06,
+      "loss": 1.0182,
+      "step": 2973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3442148314151068e-06,
+      "loss": 1.0592,
+      "step": 2974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3432241623357648e-06,
+      "loss": 1.0795,
+      "step": 2975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.342233406481237e-06,
+      "loss": 1.0965,
+      "step": 2976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3412425641020124e-06,
+      "loss": 1.0462,
+      "step": 2977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3402516354485965e-06,
+      "loss": 1.0507,
+      "step": 2978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3392606207715205e-06,
+      "loss": 1.0498,
+      "step": 2979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.338269520321337e-06,
+      "loss": 1.0918,
+      "step": 2980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.337278334348619e-06,
+      "loss": 0.6794,
+      "step": 2981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3362870631039616e-06,
+      "loss": 1.0371,
+      "step": 2982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3352957068379826e-06,
+      "loss": 1.0695,
+      "step": 2983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.334304265801319e-06,
+      "loss": 1.0589,
+      "step": 2984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3333127402446314e-06,
+      "loss": 1.0693,
+      "step": 2985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3323211304186006e-06,
+      "loss": 1.0731,
+      "step": 2986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3313294365739293e-06,
+      "loss": 1.043,
+      "step": 2987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3303376589613413e-06,
+      "loss": 1.0626,
+      "step": 2988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3293457978315812e-06,
+      "loss": 1.0406,
+      "step": 2989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.328353853435414e-06,
+      "loss": 1.0453,
+      "step": 2990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3273618260236277e-06,
+      "loss": 0.6523,
+      "step": 2991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3263697158470302e-06,
+      "loss": 1.0391,
+      "step": 2992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3253775231564505e-06,
+      "loss": 1.0755,
+      "step": 2993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3243852482027386e-06,
+      "loss": 1.0996,
+      "step": 2994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.323392891236764e-06,
+      "loss": 1.0326,
+      "step": 2995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.322400452509419e-06,
+      "loss": 1.0865,
+      "step": 2996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.321407932271615e-06,
+      "loss": 1.0658,
+      "step": 2997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3204153307742853e-06,
+      "loss": 0.6669,
+      "step": 2998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.319422648268383e-06,
+      "loss": 1.0218,
+      "step": 2999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.318429885004881e-06,
+      "loss": 1.0368,
+      "step": 3000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3174370412347734e-06,
+      "loss": 0.7099,
+      "step": 3001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3164441172090746e-06,
+      "loss": 1.0708,
+      "step": 3002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3154511131788198e-06,
+      "loss": 0.6753,
+      "step": 3003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.314458029395063e-06,
+      "loss": 1.0264,
+      "step": 3004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.313464866108881e-06,
+      "loss": 1.0304,
+      "step": 3005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3124716235713673e-06,
+      "loss": 1.0278,
+      "step": 3006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.311478302033638e-06,
+      "loss": 1.033,
+      "step": 3007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.310484901746827e-06,
+      "loss": 1.0622,
+      "step": 3008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.309491422962091e-06,
+      "loss": 1.087,
+      "step": 3009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.308497865930605e-06,
+      "loss": 1.0223,
+      "step": 3010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.307504230903562e-06,
+      "loss": 1.004,
+      "step": 3011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3065105181321775e-06,
+      "loss": 1.0623,
+      "step": 3012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3055167278676853e-06,
+      "loss": 1.0694,
+      "step": 3013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3045228603613392e-06,
+      "loss": 1.0943,
+      "step": 3014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.303528915864413e-06,
+      "loss": 1.0585,
+      "step": 3015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3025348946281987e-06,
+      "loss": 1.0483,
+      "step": 3016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3015407969040083e-06,
+      "loss": 1.1003,
+      "step": 3017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.300546622943173e-06,
+      "loss": 1.0442,
+      "step": 3018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2995523729970446e-06,
+      "loss": 0.6838,
+      "step": 3019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2985580473169918e-06,
+      "loss": 1.0571,
+      "step": 3020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2975636461544042e-06,
+      "loss": 1.0552,
+      "step": 3021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2965691697606915e-06,
+      "loss": 1.0586,
+      "step": 3022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2955746183872776e-06,
+      "loss": 0.6529,
+      "step": 3023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2945799922856105e-06,
+      "loss": 1.0586,
+      "step": 3024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2935852917071544e-06,
+      "loss": 1.0579,
+      "step": 3025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2925905169033943e-06,
+      "loss": 1.042,
+      "step": 3026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.291595668125833e-06,
+      "loss": 1.0554,
+      "step": 3027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2906007456259897e-06,
+      "loss": 1.0063,
+      "step": 3028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2896057496554058e-06,
+      "loss": 1.0942,
+      "step": 3029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.28861068046564e-06,
+      "loss": 1.0822,
+      "step": 3030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2876155383082685e-06,
+      "loss": 1.0317,
+      "step": 3031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.286620323434888e-06,
+      "loss": 1.036,
+      "step": 3032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2856250360971114e-06,
+      "loss": 1.0871,
+      "step": 3033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2846296765465706e-06,
+      "loss": 0.6753,
+      "step": 3034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2836342450349165e-06,
+      "loss": 1.0398,
+      "step": 3035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2826387418138176e-06,
+      "loss": 1.0653,
+      "step": 3036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2816431671349606e-06,
+      "loss": 0.6137,
+      "step": 3037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2806475212500515e-06,
+      "loss": 1.0372,
+      "step": 3038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2796518044108114e-06,
+      "loss": 1.0827,
+      "step": 3039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2786560168689813e-06,
+      "loss": 1.0487,
+      "step": 3040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2776601588763207e-06,
+      "loss": 1.0272,
+      "step": 3041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.276664230684606e-06,
+      "loss": 1.0794,
+      "step": 3042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.275668232545631e-06,
+      "loss": 1.0286,
+      "step": 3043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.274672164711208e-06,
+      "loss": 1.0381,
+      "step": 3044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2736760274331653e-06,
+      "loss": 1.109,
+      "step": 3045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2726798209633517e-06,
+      "loss": 1.086,
+      "step": 3046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.27168354555363e-06,
+      "loss": 1.0375,
+      "step": 3047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2706872014558835e-06,
+      "loss": 1.0323,
+      "step": 3048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2696907889220116e-06,
+      "loss": 1.0475,
+      "step": 3049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.268694308203929e-06,
+      "loss": 1.0379,
+      "step": 3050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.267697759553572e-06,
+      "loss": 0.7232,
+      "step": 3051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2667011432228906e-06,
+      "loss": 1.0531,
+      "step": 3052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.265704459463853e-06,
+      "loss": 1.0565,
+      "step": 3053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.264707708528445e-06,
+      "loss": 1.0903,
+      "step": 3054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2637108906686676e-06,
+      "loss": 1.0641,
+      "step": 3055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2627140061365414e-06,
+      "loss": 1.0625,
+      "step": 3056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2617170551841005e-06,
+      "loss": 1.0333,
+      "step": 3057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2607200380633994e-06,
+      "loss": 1.031,
+      "step": 3058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2597229550265076e-06,
+      "loss": 1.0595,
+      "step": 3059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.258725806325511e-06,
+      "loss": 0.6592,
+      "step": 3060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.257728592212511e-06,
+      "loss": 1.0363,
+      "step": 3061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.256731312939628e-06,
+      "loss": 1.0309,
+      "step": 3062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.255733968758998e-06,
+      "loss": 1.0753,
+      "step": 3063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.254736559922773e-06,
+      "loss": 1.0444,
+      "step": 3064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2537390866831213e-06,
+      "loss": 1.0794,
+      "step": 3065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2527415492922284e-06,
+      "loss": 1.0557,
+      "step": 3066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2517439480022943e-06,
+      "loss": 1.0456,
+      "step": 3067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2507462830655372e-06,
+      "loss": 1.0536,
+      "step": 3068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2497485547341895e-06,
+      "loss": 1.0549,
+      "step": 3069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.248750763260501e-06,
+      "loss": 1.0456,
+      "step": 3070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.247752908896738e-06,
+      "loss": 0.6711,
+      "step": 3071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2467549918951793e-06,
+      "loss": 0.6393,
+      "step": 3072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2457570125081237e-06,
+      "loss": 1.0398,
+      "step": 3073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.244758970987883e-06,
+      "loss": 0.6707,
+      "step": 3074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.243760867586787e-06,
+      "loss": 1.0832,
+      "step": 3075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.242762702557179e-06,
+      "loss": 1.0585,
+      "step": 3076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2417644761514187e-06,
+      "loss": 1.0564,
+      "step": 3077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2407661886218814e-06,
+      "loss": 1.0392,
+      "step": 3078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.239767840220958e-06,
+      "loss": 1.0488,
+      "step": 3079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2387694312010537e-06,
+      "loss": 1.0512,
+      "step": 3080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.237770961814591e-06,
+      "loss": 0.691,
+      "step": 3081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2367724323140073e-06,
+      "loss": 1.0397,
+      "step": 3082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2357738429517525e-06,
+      "loss": 1.067,
+      "step": 3083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.234775193980294e-06,
+      "loss": 1.0661,
+      "step": 3084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.233776485652115e-06,
+      "loss": 0.6556,
+      "step": 3085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2327777182197113e-06,
+      "loss": 1.0551,
+      "step": 3086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.231778891935596e-06,
+      "loss": 1.0552,
+      "step": 3087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2307800070522957e-06,
+      "loss": 1.0758,
+      "step": 3088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.229781063822352e-06,
+      "loss": 1.0747,
+      "step": 3089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.228782062498321e-06,
+      "loss": 1.0331,
+      "step": 3090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2277830033327746e-06,
+      "loss": 1.0534,
+      "step": 3091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2267838865782976e-06,
+      "loss": 1.0458,
+      "step": 3092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2257847124874924e-06,
+      "loss": 1.0778,
+      "step": 3093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2247854813129715e-06,
+      "loss": 1.0524,
+      "step": 3094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2237861933073645e-06,
+      "loss": 1.0399,
+      "step": 3095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.222786848723316e-06,
+      "loss": 1.0619,
+      "step": 3096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2217874478134836e-06,
+      "loss": 1.0397,
+      "step": 3097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2207879908305393e-06,
+      "loss": 1.0552,
+      "step": 3098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.21978847802717e-06,
+      "loss": 1.0844,
+      "step": 3099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.218788909656076e-06,
+      "loss": 1.0697,
+      "step": 3100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2177892859699706e-06,
+      "loss": 1.0685,
+      "step": 3101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2167896072215836e-06,
+      "loss": 1.0381,
+      "step": 3102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.215789873663657e-06,
+      "loss": 1.0757,
+      "step": 3103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2147900855489484e-06,
+      "loss": 1.0972,
+      "step": 3104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.213790243130226e-06,
+      "loss": 1.0879,
+      "step": 3105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2127903466602738e-06,
+      "loss": 1.0568,
+      "step": 3106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.21179039639189e-06,
+      "loss": 1.0566,
+      "step": 3107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2107903925778856e-06,
+      "loss": 1.0623,
+      "step": 3108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2097903354710848e-06,
+      "loss": 1.0808,
+      "step": 3109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2087902253243267e-06,
+      "loss": 1.0813,
+      "step": 3110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2077900623904614e-06,
+      "loss": 1.0792,
+      "step": 3111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.206789846922354e-06,
+      "loss": 1.0446,
+      "step": 3112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2057895791728834e-06,
+      "loss": 1.0424,
+      "step": 3113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2047892593949407e-06,
+      "loss": 1.0712,
+      "step": 3114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2037888878414307e-06,
+      "loss": 1.0563,
+      "step": 3115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.20278846476527e-06,
+      "loss": 1.0295,
+      "step": 3116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.20178799041939e-06,
+      "loss": 1.0548,
+      "step": 3117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2007874650567333e-06,
+      "loss": 1.0955,
+      "step": 3118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1997868889302577e-06,
+      "loss": 1.0406,
+      "step": 3119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.198786262292932e-06,
+      "loss": 1.0584,
+      "step": 3120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1977855853977384e-06,
+      "loss": 1.0988,
+      "step": 3121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1967848584976706e-06,
+      "loss": 1.0539,
+      "step": 3122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1957840818457374e-06,
+      "loss": 1.0728,
+      "step": 3123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.194783255694958e-06,
+      "loss": 0.6678,
+      "step": 3124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1937823802983653e-06,
+      "loss": 1.0754,
+      "step": 3125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1927814559090046e-06,
+      "loss": 1.0174,
+      "step": 3126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1917804827799323e-06,
+      "loss": 1.0541,
+      "step": 3127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1907794611642187e-06,
+      "loss": 1.0679,
+      "step": 3128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.189778391314946e-06,
+      "loss": 1.0416,
+      "step": 3129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.188777273485208e-06,
+      "loss": 1.0459,
+      "step": 3130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.187776107928111e-06,
+      "loss": 0.6634,
+      "step": 3131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.186774894896773e-06,
+      "loss": 1.0115,
+      "step": 3132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.185773634644324e-06,
+      "loss": 1.0668,
+      "step": 3133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.184772327423908e-06,
+      "loss": 1.0909,
+      "step": 3134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1837709734886772e-06,
+      "loss": 1.0415,
+      "step": 3135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1827695730917986e-06,
+      "loss": 1.063,
+      "step": 3136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.18176812648645e-06,
+      "loss": 1.0502,
+      "step": 3137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1807666339258205e-06,
+      "loss": 1.0597,
+      "step": 3138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1797650956631103e-06,
+      "loss": 1.0152,
+      "step": 3139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1787635119515336e-06,
+      "loss": 1.052,
+      "step": 3140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.177761883044313e-06,
+      "loss": 1.0666,
+      "step": 3141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1767602091946846e-06,
+      "loss": 1.0086,
+      "step": 3142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1757584906558957e-06,
+      "loss": 1.067,
+      "step": 3143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1747567276812035e-06,
+      "loss": 1.0563,
+      "step": 3144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1737549205238782e-06,
+      "loss": 1.0363,
+      "step": 3145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.172753069437199e-06,
+      "loss": 1.07,
+      "step": 3146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.171751174674459e-06,
+      "loss": 1.0418,
+      "step": 3147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.170749236488961e-06,
+      "loss": 1.029,
+      "step": 3148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1697472551340173e-06,
+      "loss": 1.0296,
+      "step": 3149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1687452308629534e-06,
+      "loss": 1.0656,
+      "step": 3150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.167743163929104e-06,
+      "loss": 1.0644,
+      "step": 3151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.166741054585817e-06,
+      "loss": 1.057,
+      "step": 3152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.165738903086447e-06,
+      "loss": 1.0274,
+      "step": 3153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1647367096843644e-06,
+      "loss": 1.078,
+      "step": 3154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1637344746329446e-06,
+      "loss": 1.0783,
+      "step": 3155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.162732198185578e-06,
+      "loss": 1.0259,
+      "step": 3156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1617298805956627e-06,
+      "loss": 1.0458,
+      "step": 3157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.160727522116609e-06,
+      "loss": 1.0475,
+      "step": 3158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1597251230018372e-06,
+      "loss": 1.0246,
+      "step": 3159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.158722683504777e-06,
+      "loss": 1.0796,
+      "step": 3160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1577202038788685e-06,
+      "loss": 1.0444,
+      "step": 3161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.156717684377563e-06,
+      "loss": 1.0753,
+      "step": 3162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.15571512525432e-06,
+      "loss": 1.0207,
+      "step": 3163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1547125267626115e-06,
+      "loss": 1.0672,
+      "step": 3164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.153709889155918e-06,
+      "loss": 1.0556,
+      "step": 3165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.152707212687729e-06,
+      "loss": 1.044,
+      "step": 3166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.151704497611545e-06,
+      "loss": 1.0515,
+      "step": 3167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1507017441808767e-06,
+      "loss": 1.031,
+      "step": 3168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.149698952649243e-06,
+      "loss": 1.0663,
+      "step": 3169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.148696123270175e-06,
+      "loss": 1.0626,
+      "step": 3170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1476932562972098e-06,
+      "loss": 1.0486,
+      "step": 3171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1466903519838965e-06,
+      "loss": 1.0335,
+      "step": 3172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1456874105837937e-06,
+      "loss": 0.6743,
+      "step": 3173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.144684432350468e-06,
+      "loss": 1.0332,
+      "step": 3174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.143681417537496e-06,
+      "loss": 1.0618,
+      "step": 3175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.142678366398465e-06,
+      "loss": 1.0283,
+      "step": 3176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.141675279186968e-06,
+      "loss": 0.6738,
+      "step": 3177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1406721561566095e-06,
+      "loss": 1.0411,
+      "step": 3178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.139668997561004e-06,
+      "loss": 1.0416,
+      "step": 3179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1386658036537723e-06,
+      "loss": 1.035,
+      "step": 3180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1376625746885466e-06,
+      "loss": 1.061,
+      "step": 3181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.136659310918966e-06,
+      "loss": 1.0209,
+      "step": 3182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.13565601259868e-06,
+      "loss": 1.0545,
+      "step": 3183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.134652679981346e-06,
+      "loss": 1.0812,
+      "step": 3184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.13364931332063e-06,
+      "loss": 1.051,
+      "step": 3185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1326459128702076e-06,
+      "loss": 1.0625,
+      "step": 3186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1316424788837618e-06,
+      "loss": 1.0623,
+      "step": 3187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1306390116149833e-06,
+      "loss": 1.0592,
+      "step": 3188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1296355113175735e-06,
+      "loss": 0.6694,
+      "step": 3189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1286319782452406e-06,
+      "loss": 1.0398,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1276284126517015e-06,
+      "loss": 1.0266,
+      "step": 3191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1266248147906825e-06,
+      "loss": 1.0082,
+      "step": 3192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1256211849159147e-06,
+      "loss": 0.6685,
+      "step": 3193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.124617523281141e-06,
+      "loss": 1.0558,
+      "step": 3194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.12361383014011e-06,
+      "loss": 1.075,
+      "step": 3195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.12261010574658e-06,
+      "loss": 1.0426,
+      "step": 3196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1216063503543153e-06,
+      "loss": 1.0385,
+      "step": 3197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1206025642170904e-06,
+      "loss": 1.0332,
+      "step": 3198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1195987475886835e-06,
+      "loss": 1.0477,
+      "step": 3199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.118594900722886e-06,
+      "loss": 1.0705,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1175910238734926e-06,
+      "loss": 1.0261,
+      "step": 3201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1165871172943075e-06,
+      "loss": 1.0211,
+      "step": 3202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1155831812391424e-06,
+      "loss": 1.0666,
+      "step": 3203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1145792159618154e-06,
+      "loss": 1.0159,
+      "step": 3204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.113575221716153e-06,
+      "loss": 1.0091,
+      "step": 3205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1125711987559886e-06,
+      "loss": 1.0525,
+      "step": 3206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1115671473351627e-06,
+      "loss": 1.1,
+      "step": 3207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.110563067707524e-06,
+      "loss": 1.0596,
+      "step": 3208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1095589601269273e-06,
+      "loss": 1.0668,
+      "step": 3209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.108554824847234e-06,
+      "loss": 1.0316,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.107550662122314e-06,
+      "loss": 1.1049,
+      "step": 3211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.106546472206044e-06,
+      "loss": 1.0216,
+      "step": 3212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1055422553523052e-06,
+      "loss": 1.0736,
+      "step": 3213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.104538011814989e-06,
+      "loss": 0.6766,
+      "step": 3214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1035337418479907e-06,
+      "loss": 1.1085,
+      "step": 3215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.102529445705214e-06,
+      "loss": 1.0978,
+      "step": 3216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.101525123640569e-06,
+      "loss": 1.0265,
+      "step": 3217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1005207759079716e-06,
+      "loss": 1.1191,
+      "step": 3218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0995164027613445e-06,
+      "loss": 1.0743,
+      "step": 3219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0985120044546185e-06,
+      "loss": 1.0353,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0975075812417272e-06,
+      "loss": 1.0711,
+      "step": 3221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0965031333766137e-06,
+      "loss": 1.0488,
+      "step": 3222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.095498661113226e-06,
+      "loss": 1.0596,
+      "step": 3223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.094494164705518e-06,
+      "loss": 1.0618,
+      "step": 3224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0934896444074517e-06,
+      "loss": 1.055,
+      "step": 3225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0924851004729913e-06,
+      "loss": 1.0283,
+      "step": 3226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0914805331561105e-06,
+      "loss": 1.0398,
+      "step": 3227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.090475942710787e-06,
+      "loss": 1.0747,
+      "step": 3228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0894713293910066e-06,
+      "loss": 1.0792,
+      "step": 3229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0884666934507577e-06,
+      "loss": 1.0161,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.087462035144037e-06,
+      "loss": 1.0408,
+      "step": 3231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0864573547248446e-06,
+      "loss": 1.0395,
+      "step": 3232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0854526524471893e-06,
+      "loss": 1.0405,
+      "step": 3233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.084447928565082e-06,
+      "loss": 1.05,
+      "step": 3234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.083443183332541e-06,
+      "loss": 1.0427,
+      "step": 3235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0824384170035915e-06,
+      "loss": 1.055,
+      "step": 3236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0814336298322594e-06,
+      "loss": 1.0758,
+      "step": 3237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0804288220725795e-06,
+      "loss": 1.046,
+      "step": 3238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.079423993978592e-06,
+      "loss": 1.0312,
+      "step": 3239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.07841914580434e-06,
+      "loss": 1.0361,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0774142778038743e-06,
+      "loss": 1.0508,
+      "step": 3241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0764093902312486e-06,
+      "loss": 1.049,
+      "step": 3242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.075404483340521e-06,
+      "loss": 1.0629,
+      "step": 3243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.074399557385757e-06,
+      "loss": 1.0407,
+      "step": 3244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0733946126210258e-06,
+      "loss": 1.0581,
+      "step": 3245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0723896493004005e-06,
+      "loss": 1.0088,
+      "step": 3246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0713846676779615e-06,
+      "loss": 1.0349,
+      "step": 3247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.070379668007789e-06,
+      "loss": 1.0539,
+      "step": 3248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.069374650543972e-06,
+      "loss": 1.0189,
+      "step": 3249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0683696155406026e-06,
+      "loss": 1.0262,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.067364563251778e-06,
+      "loss": 1.0378,
+      "step": 3251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.066359493931598e-06,
+      "loss": 1.0663,
+      "step": 3252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0653544078341695e-06,
+      "loss": 1.0028,
+      "step": 3253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.064349305213601e-06,
+      "loss": 1.0785,
+      "step": 3254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0633441863240064e-06,
+      "loss": 1.045,
+      "step": 3255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.062339051419503e-06,
+      "loss": 1.0392,
+      "step": 3256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.061333900754213e-06,
+      "loss": 1.0744,
+      "step": 3257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.060328734582264e-06,
+      "loss": 1.0734,
+      "step": 3258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.059323553157783e-06,
+      "loss": 1.041,
+      "step": 3259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0583183567349048e-06,
+      "loss": 1.0473,
+      "step": 3260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.057313145567767e-06,
+      "loss": 1.0239,
+      "step": 3261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0563079199105106e-06,
+      "loss": 1.0318,
+      "step": 3262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0553026800172803e-06,
+      "loss": 1.0708,
+      "step": 3263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0542974261422255e-06,
+      "loss": 1.0753,
+      "step": 3264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0532921585394965e-06,
+      "loss": 1.0507,
+      "step": 3265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0522868774632495e-06,
+      "loss": 1.0378,
+      "step": 3266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0512815831676434e-06,
+      "loss": 1.0449,
+      "step": 3267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0502762759068408e-06,
+      "loss": 1.0519,
+      "step": 3268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.049270955935007e-06,
+      "loss": 1.0151,
+      "step": 3269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0482656235063096e-06,
+      "loss": 1.0306,
+      "step": 3270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.047260278874921e-06,
+      "loss": 1.0264,
+      "step": 3271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0462549222950165e-06,
+      "loss": 1.0737,
+      "step": 3272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.045249554020774e-06,
+      "loss": 1.0568,
+      "step": 3273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0442441743063737e-06,
+      "loss": 1.0697,
+      "step": 3274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0432387834060006e-06,
+      "loss": 1.0425,
+      "step": 3275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.04223338157384e-06,
+      "loss": 1.0502,
+      "step": 3276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0412279690640816e-06,
+      "loss": 0.6882,
+      "step": 3277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0402225461309176e-06,
+      "loss": 1.0347,
+      "step": 3278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0392171130285433e-06,
+      "loss": 1.0382,
+      "step": 3279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0382116700111554e-06,
+      "loss": 1.025,
+      "step": 3280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0372062173329536e-06,
+      "loss": 1.0508,
+      "step": 3281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.03620075524814e-06,
+      "loss": 1.0537,
+      "step": 3282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.035195284010919e-06,
+      "loss": 1.0854,
+      "step": 3283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0341898038754982e-06,
+      "loss": 1.0675,
+      "step": 3284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0331843150960864e-06,
+      "loss": 0.6872,
+      "step": 3285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.032178817926895e-06,
+      "loss": 1.0441,
+      "step": 3286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.031173312622138e-06,
+      "loss": 1.0783,
+      "step": 3287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0301677994360295e-06,
+      "loss": 1.0442,
+      "step": 3288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.029162278622788e-06,
+      "loss": 1.0506,
+      "step": 3289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0281567504366327e-06,
+      "loss": 1.0399,
+      "step": 3290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.027151215131786e-06,
+      "loss": 1.0625,
+      "step": 3291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.026145672962469e-06,
+      "loss": 1.0101,
+      "step": 3292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0251401241829076e-06,
+      "loss": 1.0745,
+      "step": 3293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.024134569047329e-06,
+      "loss": 1.0592,
+      "step": 3294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0231290078099605e-06,
+      "loss": 1.0888,
+      "step": 3295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0221234407250314e-06,
+      "loss": 1.0681,
+      "step": 3296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0211178680467736e-06,
+      "loss": 1.0891,
+      "step": 3297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.02011229002942e-06,
+      "loss": 1.0878,
+      "step": 3298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0191067069272023e-06,
+      "loss": 1.0167,
+      "step": 3299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.018101118994358e-06,
+      "loss": 1.0236,
+      "step": 3300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0170955264851227e-06,
+      "loss": 1.0613,
+      "step": 3301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0160899296537344e-06,
+      "loss": 1.0137,
+      "step": 3302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.015084328754431e-06,
+      "loss": 1.0327,
+      "step": 3303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0140787240414523e-06,
+      "loss": 1.0575,
+      "step": 3304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.01307311576904e-06,
+      "loss": 1.0211,
+      "step": 3305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0120675041914337e-06,
+      "loss": 1.0252,
+      "step": 3306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.011061889562878e-06,
+      "loss": 1.0293,
+      "step": 3307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0100562721376156e-06,
+      "loss": 1.0354,
+      "step": 3308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0090506521698886e-06,
+      "loss": 0.667,
+      "step": 3309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.008045029913943e-06,
+      "loss": 1.0549,
+      "step": 3310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0070394056240233e-06,
+      "loss": 1.044,
+      "step": 3311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0060337795543756e-06,
+      "loss": 0.6825,
+      "step": 3312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.005028151959246e-06,
+      "loss": 1.0813,
+      "step": 3313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0040225230928804e-06,
+      "loss": 1.0479,
+      "step": 3314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0030168932095258e-06,
+      "loss": 1.075,
+      "step": 3315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0020112625634294e-06,
+      "loss": 1.0934,
+      "step": 3316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.001005631408838e-06,
+      "loss": 1.048,
+      "step": 3317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999999999999995e-06,
+      "loss": 1.0798,
+      "step": 3318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.998994368591162e-06,
+      "loss": 1.0382,
+      "step": 3319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9979887374365704e-06,
+      "loss": 1.0629,
+      "step": 3320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.996983106790474e-06,
+      "loss": 1.0506,
+      "step": 3321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9959774769071194e-06,
+      "loss": 1.0326,
+      "step": 3322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9949718480407536e-06,
+      "loss": 1.0118,
+      "step": 3323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9939662204456246e-06,
+      "loss": 1.0871,
+      "step": 3324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9929605943759765e-06,
+      "loss": 1.0613,
+      "step": 3325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9919549700860574e-06,
+      "loss": 0.6843,
+      "step": 3326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.990949347830111e-06,
+      "loss": 1.0768,
+      "step": 3327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9899437278623847e-06,
+      "loss": 0.6921,
+      "step": 3328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988938110437121e-06,
+      "loss": 1.0638,
+      "step": 3329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.987932495808566e-06,
+      "loss": 1.0203,
+      "step": 3330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.98692688423096e-06,
+      "loss": 1.0818,
+      "step": 3331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9859212759585475e-06,
+      "loss": 1.0448,
+      "step": 3332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.984915671245569e-06,
+      "loss": 1.0641,
+      "step": 3333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.983910070346266e-06,
+      "loss": 1.0358,
+      "step": 3334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9829044735148776e-06,
+      "loss": 1.0431,
+      "step": 3335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.981898881005642e-06,
+      "loss": 1.0566,
+      "step": 3336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.980893293072798e-06,
+      "loss": 1.0754,
+      "step": 3337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.979887709970581e-06,
+      "loss": 1.0594,
+      "step": 3338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.978882131953226e-06,
+      "loss": 1.0781,
+      "step": 3339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9778765592749684e-06,
+      "loss": 1.0378,
+      "step": 3340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9768709921900398e-06,
+      "loss": 1.0289,
+      "step": 3341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9758654309526713e-06,
+      "loss": 1.0208,
+      "step": 3342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.974859875817092e-06,
+      "loss": 1.0667,
+      "step": 3343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.973854327037531e-06,
+      "loss": 0.6723,
+      "step": 3344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972848784868214e-06,
+      "loss": 1.0267,
+      "step": 3345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971843249563367e-06,
+      "loss": 1.0492,
+      "step": 3346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970837721377212e-06,
+      "loss": 1.0556,
+      "step": 3347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9698322005639707e-06,
+      "loss": 1.0423,
+      "step": 3348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9688266873778627e-06,
+      "loss": 1.0634,
+      "step": 3349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.967821182073105e-06,
+      "loss": 1.0527,
+      "step": 3350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9668156849039134e-06,
+      "loss": 0.9892,
+      "step": 3351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9658101961245024e-06,
+      "loss": 0.6364,
+      "step": 3352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9648047159890813e-06,
+      "loss": 1.099,
+      "step": 3353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9637992447518607e-06,
+      "loss": 1.0469,
+      "step": 3354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.962793782667047e-06,
+      "loss": 1.0421,
+      "step": 3355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.961788329988845e-06,
+      "loss": 1.0521,
+      "step": 3356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.960782886971457e-06,
+      "loss": 1.067,
+      "step": 3357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.959777453869082e-06,
+      "loss": 1.045,
+      "step": 3358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9587720309359186e-06,
+      "loss": 1.0614,
+      "step": 3359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9577666184261604e-06,
+      "loss": 1.03,
+      "step": 3360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9567612165939996e-06,
+      "loss": 1.0629,
+      "step": 3361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.955755825693626e-06,
+      "loss": 1.0505,
+      "step": 3362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9547504459792264e-06,
+      "loss": 0.6513,
+      "step": 3363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.953745077704984e-06,
+      "loss": 1.0714,
+      "step": 3364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.952739721125079e-06,
+      "loss": 1.0406,
+      "step": 3365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.951734376493691e-06,
+      "loss": 1.0897,
+      "step": 3366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.950729044064994e-06,
+      "loss": 1.0474,
+      "step": 3367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9497237240931595e-06,
+      "loss": 0.6444,
+      "step": 3368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948718416832357e-06,
+      "loss": 1.0098,
+      "step": 3369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9477131225367508e-06,
+      "loss": 1.0423,
+      "step": 3370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9467078414605038e-06,
+      "loss": 1.0528,
+      "step": 3371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9457025738577748e-06,
+      "loss": 1.0506,
+      "step": 3372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.944697319982719e-06,
+      "loss": 1.0619,
+      "step": 3373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9436920800894897e-06,
+      "loss": 0.6325,
+      "step": 3374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.942686854432233e-06,
+      "loss": 1.0618,
+      "step": 3375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9416816432650955e-06,
+      "loss": 1.0462,
+      "step": 3376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9406764468422174e-06,
+      "loss": 1.0417,
+      "step": 3377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9396712654177364e-06,
+      "loss": 1.0472,
+      "step": 3378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9386660992457866e-06,
+      "loss": 1.0481,
+      "step": 3379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.937660948580497e-06,
+      "loss": 1.0492,
+      "step": 3380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9366558136759943e-06,
+      "loss": 1.0653,
+      "step": 3381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.935650694786399e-06,
+      "loss": 1.0744,
+      "step": 3382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9346455921658303e-06,
+      "loss": 1.0555,
+      "step": 3383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9336405060684015e-06,
+      "loss": 1.0231,
+      "step": 3384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9326354367482225e-06,
+      "loss": 1.0234,
+      "step": 3385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.931630384459397e-06,
+      "loss": 1.0636,
+      "step": 3386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9306253494560283e-06,
+      "loss": 1.056,
+      "step": 3387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9296203319922113e-06,
+      "loss": 1.0436,
+      "step": 3388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9286153323220387e-06,
+      "loss": 1.0462,
+      "step": 3389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9276103506995993e-06,
+      "loss": 1.027,
+      "step": 3390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.926605387378974e-06,
+      "loss": 1.0706,
+      "step": 3391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.925600442614243e-06,
+      "loss": 1.0621,
+      "step": 3392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.924595516659479e-06,
+      "loss": 1.0513,
+      "step": 3393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9235906097687517e-06,
+      "loss": 1.0635,
+      "step": 3394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.922585722196125e-06,
+      "loss": 1.0302,
+      "step": 3395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9215808541956597e-06,
+      "loss": 1.0616,
+      "step": 3396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.920576006021408e-06,
+      "loss": 1.0495,
+      "step": 3397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9195711779274204e-06,
+      "loss": 1.0672,
+      "step": 3398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.918566370167741e-06,
+      "loss": 1.0489,
+      "step": 3399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9175615829964087e-06,
+      "loss": 1.0837,
+      "step": 3400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9165568166674586e-06,
+      "loss": 1.0392,
+      "step": 3401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915552071434918e-06,
+      "loss": 1.0681,
+      "step": 3402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.914547347552811e-06,
+      "loss": 0.6441,
+      "step": 3403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9135426452751552e-06,
+      "loss": 1.0581,
+      "step": 3404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.912537964855963e-06,
+      "loss": 1.0705,
+      "step": 3405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.911533306549242e-06,
+      "loss": 1.0547,
+      "step": 3406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9105286706089936e-06,
+      "loss": 1.0615,
+      "step": 3407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9095240572892126e-06,
+      "loss": 1.0204,
+      "step": 3408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9085194668438897e-06,
+      "loss": 1.042,
+      "step": 3409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9075148995270085e-06,
+      "loss": 1.0576,
+      "step": 3410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9065103555925485e-06,
+      "loss": 1.0603,
+      "step": 3411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9055058352944821e-06,
+      "loss": 1.0842,
+      "step": 3412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9045013388867744e-06,
+      "loss": 1.0539,
+      "step": 3413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9034968666233863e-06,
+      "loss": 1.0849,
+      "step": 3414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9024924187582728e-06,
+      "loss": 1.0352,
+      "step": 3415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9014879955453817e-06,
+      "loss": 1.0441,
+      "step": 3416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9004835972386549e-06,
+      "loss": 1.0587,
+      "step": 3417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8994792240920284e-06,
+      "loss": 1.0592,
+      "step": 3418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.898474876359431e-06,
+      "loss": 1.0523,
+      "step": 3419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.897470554294786e-06,
+      "loss": 0.6515,
+      "step": 3420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8964662581520093e-06,
+      "loss": 1.017,
+      "step": 3421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8954619881850111e-06,
+      "loss": 1.0195,
+      "step": 3422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8944577446476952e-06,
+      "loss": 0.6446,
+      "step": 3423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8934535277939566e-06,
+      "loss": 1.0864,
+      "step": 3424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8924493378776858e-06,
+      "loss": 1.0558,
+      "step": 3425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914451751527659e-06,
+      "loss": 1.059,
+      "step": 3426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890441039873073e-06,
+      "loss": 1.0173,
+      "step": 3427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894369322924756e-06,
+      "loss": 1.0932,
+      "step": 3428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884328526648372e-06,
+      "loss": 1.0857,
+      "step": 3429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874288012440117e-06,
+      "loss": 0.67,
+      "step": 3430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8864247782838472e-06,
+      "loss": 1.0238,
+      "step": 3431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8854207840381847e-06,
+      "loss": 1.0496,
+      "step": 3432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8844168187608576e-06,
+      "loss": 1.0577,
+      "step": 3433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.883412882705693e-06,
+      "loss": 1.0483,
+      "step": 3434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.882408976126508e-06,
+      "loss": 1.0463,
+      "step": 3435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8814050992771147e-06,
+      "loss": 1.0798,
+      "step": 3436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8804012524113166e-06,
+      "loss": 1.0716,
+      "step": 3437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8793974357829105e-06,
+      "loss": 1.0528,
+      "step": 3438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8783936496456848e-06,
+      "loss": 1.0469,
+      "step": 3439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8773898942534206e-06,
+      "loss": 1.0414,
+      "step": 3440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8763861698598901e-06,
+      "loss": 1.075,
+      "step": 3441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8753824767188593e-06,
+      "loss": 1.0506,
+      "step": 3442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8743788150840853e-06,
+      "loss": 1.0648,
+      "step": 3443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8733751852093177e-06,
+      "loss": 1.0547,
+      "step": 3444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8723715873482983e-06,
+      "loss": 1.0431,
+      "step": 3445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8713680217547594e-06,
+      "loss": 1.0109,
+      "step": 3446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.870364488682427e-06,
+      "loss": 1.0736,
+      "step": 3447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.869360988385017e-06,
+      "loss": 1.0464,
+      "step": 3448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8683575211162385e-06,
+      "loss": 1.0624,
+      "step": 3449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8673540871297922e-06,
+      "loss": 0.6606,
+      "step": 3450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8663506866793696e-06,
+      "loss": 1.0689,
+      "step": 3451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8653473200186539e-06,
+      "loss": 0.644,
+      "step": 3452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8643439874013199e-06,
+      "loss": 1.0294,
+      "step": 3453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8633406890810336e-06,
+      "loss": 1.0297,
+      "step": 3454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8623374253114534e-06,
+      "loss": 1.0698,
+      "step": 3455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.861334196346228e-06,
+      "loss": 1.0577,
+      "step": 3456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8603310024389964e-06,
+      "loss": 1.0613,
+      "step": 3457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8593278438433903e-06,
+      "loss": 1.0518,
+      "step": 3458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8583247208130321e-06,
+      "loss": 1.0554,
+      "step": 3459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8573216336015353e-06,
+      "loss": 1.034,
+      "step": 3460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8563185824625035e-06,
+      "loss": 1.0103,
+      "step": 3461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.855315567649532e-06,
+      "loss": 1.0608,
+      "step": 3462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8543125894162064e-06,
+      "loss": 1.0246,
+      "step": 3463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8533096480161033e-06,
+      "loss": 1.0459,
+      "step": 3464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8523067437027903e-06,
+      "loss": 1.0599,
+      "step": 3465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.851303876729825e-06,
+      "loss": 1.0547,
+      "step": 3466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8503010473507567e-06,
+      "loss": 1.0299,
+      "step": 3467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8492982558191234e-06,
+      "loss": 1.0658,
+      "step": 3468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.848295502388455e-06,
+      "loss": 1.056,
+      "step": 3469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8472927873122713e-06,
+      "loss": 1.0157,
+      "step": 3470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.846290110844082e-06,
+      "loss": 1.0375,
+      "step": 3471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8452874732373877e-06,
+      "loss": 1.0203,
+      "step": 3472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8442848747456798e-06,
+      "loss": 1.0262,
+      "step": 3473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8432823156224374e-06,
+      "loss": 1.0401,
+      "step": 3474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8422797961211315e-06,
+      "loss": 1.0263,
+      "step": 3475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.841277316495223e-06,
+      "loss": 1.0512,
+      "step": 3476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8402748769981626e-06,
+      "loss": 1.0702,
+      "step": 3477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8392724778833907e-06,
+      "loss": 1.0788,
+      "step": 3478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8382701194043373e-06,
+      "loss": 0.6734,
+      "step": 3479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8372678018144223e-06,
+      "loss": 1.0746,
+      "step": 3480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8362655253670555e-06,
+      "loss": 1.0462,
+      "step": 3481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8352632903156357e-06,
+      "loss": 1.0356,
+      "step": 3482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.834261096913552e-06,
+      "loss": 1.0588,
+      "step": 3483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8332589454141834e-06,
+      "loss": 1.0653,
+      "step": 3484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8322568360708957e-06,
+      "loss": 1.0535,
+      "step": 3485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8312547691370468e-06,
+      "loss": 1.0716,
+      "step": 3486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8302527448659827e-06,
+      "loss": 1.0706,
+      "step": 3487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.829250763511039e-06,
+      "loss": 0.9851,
+      "step": 3488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.828248825325541e-06,
+      "loss": 1.0849,
+      "step": 3489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827246930562801e-06,
+      "loss": 1.0763,
+      "step": 3490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8262450794761222e-06,
+      "loss": 1.0888,
+      "step": 3491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8252432723187963e-06,
+      "loss": 0.6418,
+      "step": 3492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8242415093441041e-06,
+      "loss": 1.0589,
+      "step": 3493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8232397908053148e-06,
+      "loss": 1.0264,
+      "step": 3494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8222381169556872e-06,
+      "loss": 0.6495,
+      "step": 3495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8212364880484668e-06,
+      "loss": 1.0366,
+      "step": 3496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8202349043368897e-06,
+      "loss": 1.0774,
+      "step": 3497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8192333660741798e-06,
+      "loss": 1.0111,
+      "step": 3498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8182318735135498e-06,
+      "loss": 1.029,
+      "step": 3499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8172304269082014e-06,
+      "loss": 1.0519,
+      "step": 3500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8162290265113228e-06,
+      "loss": 1.0398,
+      "step": 3501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8152276725760923e-06,
+      "loss": 1.0473,
+      "step": 3502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8142263653556754e-06,
+      "loss": 1.0486,
+      "step": 3503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.813225105103227e-06,
+      "loss": 1.0456,
+      "step": 3504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.812223892071889e-06,
+      "loss": 1.0697,
+      "step": 3505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8112227265147925e-06,
+      "loss": 1.0439,
+      "step": 3506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.810221608685054e-06,
+      "loss": 1.0707,
+      "step": 3507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8092205388357811e-06,
+      "loss": 1.0107,
+      "step": 3508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8082195172200677e-06,
+      "loss": 1.0423,
+      "step": 3509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8072185440909954e-06,
+      "loss": 1.047,
+      "step": 3510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8062176197016347e-06,
+      "loss": 1.0424,
+      "step": 3511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.805216744305042e-06,
+      "loss": 1.0305,
+      "step": 3512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8042159181542629e-06,
+      "loss": 1.0564,
+      "step": 3513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8032151415023294e-06,
+      "loss": 1.054,
+      "step": 3514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8022144146022619e-06,
+      "loss": 0.6772,
+      "step": 3515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8012137377070677e-06,
+      "loss": 1.0927,
+      "step": 3516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8002131110697427e-06,
+      "loss": 1.0216,
+      "step": 3517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7992125349432667e-06,
+      "loss": 1.0184,
+      "step": 3518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7982120095806106e-06,
+      "loss": 1.012,
+      "step": 3519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7972115352347301e-06,
+      "loss": 1.0561,
+      "step": 3520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7962111121585696e-06,
+      "loss": 1.0298,
+      "step": 3521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795210740605059e-06,
+      "loss": 1.0215,
+      "step": 3522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7942104208271165e-06,
+      "loss": 1.0716,
+      "step": 3523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7932101530776457e-06,
+      "loss": 1.0292,
+      "step": 3524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7922099376095388e-06,
+      "loss": 1.0476,
+      "step": 3525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7912097746756735e-06,
+      "loss": 1.0498,
+      "step": 3526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7902096645289148e-06,
+      "loss": 1.0516,
+      "step": 3527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7892096074221146e-06,
+      "loss": 1.0394,
+      "step": 3528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.78820960360811e-06,
+      "loss": 1.0575,
+      "step": 3529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7872096533397265e-06,
+      "loss": 1.0579,
+      "step": 3530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7862097568697747e-06,
+      "loss": 1.0642,
+      "step": 3531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7852099144510519e-06,
+      "loss": 1.0468,
+      "step": 3532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7842101263363423e-06,
+      "loss": 1.075,
+      "step": 3533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7832103927784164e-06,
+      "loss": 1.0378,
+      "step": 3534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7822107140300296e-06,
+      "loss": 1.0756,
+      "step": 3535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7812110903439246e-06,
+      "loss": 1.0224,
+      "step": 3536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.78021152197283e-06,
+      "loss": 1.0357,
+      "step": 3537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7792120091694603e-06,
+      "loss": 1.0478,
+      "step": 3538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7782125521865166e-06,
+      "loss": 1.0211,
+      "step": 3539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.777213151276684e-06,
+      "loss": 1.0542,
+      "step": 3540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7762138066926355e-06,
+      "loss": 1.0375,
+      "step": 3541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.775214518687029e-06,
+      "loss": 1.0391,
+      "step": 3542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7742152875125083e-06,
+      "loss": 1.0586,
+      "step": 3543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7732161134217018e-06,
+      "loss": 1.0817,
+      "step": 3544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7722169966672258e-06,
+      "loss": 1.0143,
+      "step": 3545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7712179375016792e-06,
+      "loss": 1.012,
+      "step": 3546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.770218936177648e-06,
+      "loss": 1.0395,
+      "step": 3547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7692199929477043e-06,
+      "loss": 1.0903,
+      "step": 3548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7682211080644037e-06,
+      "loss": 1.0346,
+      "step": 3549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7672222817802887e-06,
+      "loss": 1.0451,
+      "step": 3550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7662235143478854e-06,
+      "loss": 1.0664,
+      "step": 3551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7652248060197058e-06,
+      "loss": 1.0414,
+      "step": 3552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7642261570482477e-06,
+      "loss": 1.0419,
+      "step": 3553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.763227567685993e-06,
+      "loss": 1.0599,
+      "step": 3554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7622290381854082e-06,
+      "loss": 1.0583,
+      "step": 3555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7612305687989464e-06,
+      "loss": 1.0241,
+      "step": 3556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7602321597790424e-06,
+      "loss": 0.6078,
+      "step": 3557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7592338113781186e-06,
+      "loss": 1.0459,
+      "step": 3558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7582355238485813e-06,
+      "loss": 1.0568,
+      "step": 3559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.757237297442821e-06,
+      "loss": 1.0615,
+      "step": 3560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7562391324132128e-06,
+      "loss": 1.0585,
+      "step": 3561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7552410290121166e-06,
+      "loss": 1.0596,
+      "step": 3562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7542429874918763e-06,
+      "loss": 0.6678,
+      "step": 3563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7532450081048208e-06,
+      "loss": 1.0217,
+      "step": 3564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7522470911032624e-06,
+      "loss": 1.0258,
+      "step": 3565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7512492367394982e-06,
+      "loss": 1.0846,
+      "step": 3566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7502514452658107e-06,
+      "loss": 0.9957,
+      "step": 3567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7492537169344632e-06,
+      "loss": 1.0617,
+      "step": 3568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7482560519977057e-06,
+      "loss": 1.0475,
+      "step": 3569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7472584507077719e-06,
+      "loss": 1.0068,
+      "step": 3570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7462609133168785e-06,
+      "loss": 1.0428,
+      "step": 3571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7452634400772272e-06,
+      "loss": 1.0199,
+      "step": 3572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.744266031241002e-06,
+      "loss": 1.05,
+      "step": 3573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.743268687060372e-06,
+      "loss": 1.0498,
+      "step": 3574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7422714077874892e-06,
+      "loss": 1.031,
+      "step": 3575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7412741936744892e-06,
+      "loss": 0.6706,
+      "step": 3576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7402770449734918e-06,
+      "loss": 1.0382,
+      "step": 3577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7392799619366004e-06,
+      "loss": 1.0567,
+      "step": 3578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7382829448158993e-06,
+      "loss": 1.0268,
+      "step": 3579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.737285993863459e-06,
+      "loss": 1.0345,
+      "step": 3580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7362891093313322e-06,
+      "loss": 1.03,
+      "step": 3581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7352922914715551e-06,
+      "loss": 1.0401,
+      "step": 3582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.734295540536147e-06,
+      "loss": 0.6691,
+      "step": 3583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7332988567771094e-06,
+      "loss": 1.0302,
+      "step": 3584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7323022404464279e-06,
+      "loss": 1.0587,
+      "step": 3585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7313056917960706e-06,
+      "loss": 1.0535,
+      "step": 3586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.730309211077989e-06,
+      "loss": 1.0531,
+      "step": 3587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7293127985441163e-06,
+      "loss": 1.0537,
+      "step": 3588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7283164544463705e-06,
+      "loss": 0.6642,
+      "step": 3589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.727320179036649e-06,
+      "loss": 1.0832,
+      "step": 3590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.726323972566835e-06,
+      "loss": 0.67,
+      "step": 3591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7253278352887924e-06,
+      "loss": 1.0157,
+      "step": 3592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7243317674543687e-06,
+      "loss": 1.0603,
+      "step": 3593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7233357693153941e-06,
+      "loss": 1.0412,
+      "step": 3594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7223398411236789e-06,
+      "loss": 0.9999,
+      "step": 3595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7213439831310183e-06,
+      "loss": 0.9895,
+      "step": 3596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7203481955891888e-06,
+      "loss": 0.6481,
+      "step": 3597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7193524787499486e-06,
+      "loss": 1.0595,
+      "step": 3598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7183568328650386e-06,
+      "loss": 0.6234,
+      "step": 3599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7173612581861828e-06,
+      "loss": 1.0723,
+      "step": 3600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.716365754965084e-06,
+      "loss": 1.0308,
+      "step": 3601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7153703234534298e-06,
+      "loss": 1.0701,
+      "step": 3602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7143749639028892e-06,
+      "loss": 1.028,
+      "step": 3603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7133796765651122e-06,
+      "loss": 1.0383,
+      "step": 3604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7123844616917315e-06,
+      "loss": 0.6635,
+      "step": 3605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7113893195343602e-06,
+      "loss": 0.6805,
+      "step": 3606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.710394250344594e-06,
+      "loss": 1.0279,
+      "step": 3607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7093992543740103e-06,
+      "loss": 1.0305,
+      "step": 3608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7084043318741674e-06,
+      "loss": 1.0536,
+      "step": 3609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.707409483096605e-06,
+      "loss": 1.0316,
+      "step": 3610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7064147082928452e-06,
+      "loss": 1.0557,
+      "step": 3611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7054200077143896e-06,
+      "loss": 1.0475,
+      "step": 3612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7044253816127229e-06,
+      "loss": 0.6273,
+      "step": 3613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7034308302393091e-06,
+      "loss": 1.0397,
+      "step": 3614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.702436353845595e-06,
+      "loss": 1.0803,
+      "step": 3615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7014419526830083e-06,
+      "loss": 1.0957,
+      "step": 3616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7004476270029554e-06,
+      "loss": 1.0336,
+      "step": 3617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6994533770568268e-06,
+      "loss": 1.0115,
+      "step": 3618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.698459203095992e-06,
+      "loss": 1.0644,
+      "step": 3619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6974651053718015e-06,
+      "loss": 1.0665,
+      "step": 3620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.696471084135587e-06,
+      "loss": 1.0444,
+      "step": 3621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6954771396386606e-06,
+      "loss": 1.0468,
+      "step": 3622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6944832721323147e-06,
+      "loss": 0.6812,
+      "step": 3623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6934894818678226e-06,
+      "loss": 1.0474,
+      "step": 3624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6924957690964382e-06,
+      "loss": 1.0598,
+      "step": 3625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6915021340693954e-06,
+      "loss": 1.0387,
+      "step": 3626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6905085770379088e-06,
+      "loss": 1.0146,
+      "step": 3627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6895150982531727e-06,
+      "loss": 0.676,
+      "step": 3628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6885216979663622e-06,
+      "loss": 1.0279,
+      "step": 3629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6875283764286325e-06,
+      "loss": 1.025,
+      "step": 3630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6865351338911186e-06,
+      "loss": 1.0019,
+      "step": 3631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685541970604936e-06,
+      "loss": 1.0835,
+      "step": 3632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6845488868211802e-06,
+      "loss": 1.0134,
+      "step": 3633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6835558827909254e-06,
+      "loss": 1.004,
+      "step": 3634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6825629587652268e-06,
+      "loss": 1.0453,
+      "step": 3635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6815701149951192e-06,
+      "loss": 1.0402,
+      "step": 3636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6805773517316168e-06,
+      "loss": 1.0287,
+      "step": 3637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.679584669225715e-06,
+      "loss": 1.0305,
+      "step": 3638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6785920677283848e-06,
+      "loss": 0.9956,
+      "step": 3639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6775995474905811e-06,
+      "loss": 1.047,
+      "step": 3640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676607108763236e-06,
+      "loss": 0.9992,
+      "step": 3641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6756147517972616e-06,
+      "loss": 1.0306,
+      "step": 3642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6746224768435493e-06,
+      "loss": 1.0059,
+      "step": 3643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6736302841529696e-06,
+      "loss": 1.0425,
+      "step": 3644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6726381739763725e-06,
+      "loss": 1.0759,
+      "step": 3645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6716461465645861e-06,
+      "loss": 1.0291,
+      "step": 3646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6706542021684192e-06,
+      "loss": 1.037,
+      "step": 3647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6696623410386583e-06,
+      "loss": 1.0568,
+      "step": 3648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6686705634260708e-06,
+      "loss": 1.0288,
+      "step": 3649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6676788695813992e-06,
+      "loss": 1.0198,
+      "step": 3650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6666872597553684e-06,
+      "loss": 1.0111,
+      "step": 3651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6656957341986809e-06,
+      "loss": 1.0106,
+      "step": 3652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6647042931620174e-06,
+      "loss": 1.0384,
+      "step": 3653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6637129368960378e-06,
+      "loss": 1.0545,
+      "step": 3654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6627216656513812e-06,
+      "loss": 1.0048,
+      "step": 3655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6617304796786632e-06,
+      "loss": 1.0725,
+      "step": 3656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6607393792284791e-06,
+      "loss": 1.1024,
+      "step": 3657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6597483645514035e-06,
+      "loss": 1.0314,
+      "step": 3658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6587574358979875e-06,
+      "loss": 1.0494,
+      "step": 3659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6577665935187627e-06,
+      "loss": 1.0504,
+      "step": 3660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6567758376642353e-06,
+      "loss": 1.0512,
+      "step": 3661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6557851685848928e-06,
+      "loss": 1.0528,
+      "step": 3662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6547945865312003e-06,
+      "loss": 0.6818,
+      "step": 3663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6538040917535998e-06,
+      "loss": 0.6625,
+      "step": 3664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6528136845025118e-06,
+      "loss": 1.0283,
+      "step": 3665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6518233650283353e-06,
+      "loss": 1.0124,
+      "step": 3666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6508331335814455e-06,
+      "loss": 1.0074,
+      "step": 3667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.649842990412197e-06,
+      "loss": 1.0098,
+      "step": 3668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6488529357709214e-06,
+      "loss": 1.0481,
+      "step": 3669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6478629699079274e-06,
+      "loss": 1.0003,
+      "step": 3670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6468730930735032e-06,
+      "loss": 1.0331,
+      "step": 3671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6458833055179112e-06,
+      "loss": 1.0221,
+      "step": 3672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6448936074913938e-06,
+      "loss": 1.0373,
+      "step": 3673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6439039992441702e-06,
+      "loss": 1.0455,
+      "step": 3674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6429144810264368e-06,
+      "loss": 1.0275,
+      "step": 3675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.641925053088367e-06,
+      "loss": 1.027,
+      "step": 3676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6409357156801125e-06,
+      "loss": 1.0397,
+      "step": 3677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6399464690518e-06,
+      "loss": 1.0577,
+      "step": 3678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6389573134535346e-06,
+      "loss": 0.6893,
+      "step": 3679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6379682491353988e-06,
+      "loss": 1.0302,
+      "step": 3680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6369792763474507e-06,
+      "loss": 0.6601,
+      "step": 3681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6359903953397278e-06,
+      "loss": 1.0627,
+      "step": 3682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6350016063622403e-06,
+      "loss": 1.0573,
+      "step": 3683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6340129096649783e-06,
+      "loss": 1.0268,
+      "step": 3684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6330243054979078e-06,
+      "loss": 1.0585,
+      "step": 3685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6320357941109715e-06,
+      "loss": 1.0287,
+      "step": 3686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6310473757540882e-06,
+      "loss": 1.0737,
+      "step": 3687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6300590506771539e-06,
+      "loss": 1.0991,
+      "step": 3688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6290708191300398e-06,
+      "loss": 1.0264,
+      "step": 3689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6280826813625946e-06,
+      "loss": 1.0339,
+      "step": 3690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6270946376246427e-06,
+      "loss": 1.061,
+      "step": 3691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6261066881659853e-06,
+      "loss": 1.0797,
+      "step": 3692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6251188332363994e-06,
+      "loss": 0.6819,
+      "step": 3693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6241310730856378e-06,
+      "loss": 1.0621,
+      "step": 3694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6231434079634299e-06,
+      "loss": 1.0463,
+      "step": 3695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6221558381194801e-06,
+      "loss": 1.0278,
+      "step": 3696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.62116836380347e-06,
+      "loss": 1.0375,
+      "step": 3697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6201809852650562e-06,
+      "loss": 1.0608,
+      "step": 3698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.619193702753872e-06,
+      "loss": 1.0593,
+      "step": 3699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.618206516519525e-06,
+      "loss": 1.0273,
+      "step": 3700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6172194268115996e-06,
+      "loss": 1.0453,
+      "step": 3701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6162324338796548e-06,
+      "loss": 1.0505,
+      "step": 3702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6152455379732268e-06,
+      "loss": 1.0793,
+      "step": 3703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.614258739341826e-06,
+      "loss": 1.0644,
+      "step": 3704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6132720382349376e-06,
+      "loss": 1.0421,
+      "step": 3705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6122854349020237e-06,
+      "loss": 1.0578,
+      "step": 3706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6112989295925205e-06,
+      "loss": 1.0141,
+      "step": 3707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6103125225558403e-06,
+      "loss": 0.9933,
+      "step": 3708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6093262140413702e-06,
+      "loss": 1.0562,
+      "step": 3709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6083400042984724e-06,
+      "loss": 0.9857,
+      "step": 3710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.607353893576483e-06,
+      "loss": 1.066,
+      "step": 3711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.606367882124715e-06,
+      "loss": 1.0519,
+      "step": 3712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6053819701924551e-06,
+      "loss": 1.0044,
+      "step": 3713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6043961580289652e-06,
+      "loss": 1.0173,
+      "step": 3714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6034104458834827e-06,
+      "loss": 1.0669,
+      "step": 3715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6024248340052177e-06,
+      "loss": 1.0728,
+      "step": 3716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6014393226433568e-06,
+      "loss": 1.049,
+      "step": 3717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6004539120470608e-06,
+      "loss": 1.0495,
+      "step": 3718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5994686024654642e-06,
+      "loss": 1.0104,
+      "step": 3719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.598483394147677e-06,
+      "loss": 1.04,
+      "step": 3720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5974982873427838e-06,
+      "loss": 1.0508,
+      "step": 3721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5965132822998416e-06,
+      "loss": 1.0173,
+      "step": 3722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.595528379267883e-06,
+      "loss": 1.0032,
+      "step": 3723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5945435784959158e-06,
+      "loss": 1.0654,
+      "step": 3724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5935588802329203e-06,
+      "loss": 1.0437,
+      "step": 3725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5925742847278521e-06,
+      "loss": 1.0103,
+      "step": 3726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.591589792229639e-06,
+      "loss": 0.9856,
+      "step": 3727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5906054029871851e-06,
+      "loss": 1.0176,
+      "step": 3728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.589621117249367e-06,
+      "loss": 0.6779,
+      "step": 3729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5886369352650353e-06,
+      "loss": 0.9877,
+      "step": 3730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.587652857283014e-06,
+      "loss": 1.0641,
+      "step": 3731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5866688835521034e-06,
+      "loss": 1.0113,
+      "step": 3732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5856850143210725e-06,
+      "loss": 0.6411,
+      "step": 3733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.584701249838668e-06,
+      "loss": 1.0512,
+      "step": 3734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5837175903536086e-06,
+      "loss": 1.0556,
+      "step": 3735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5827340361145872e-06,
+      "loss": 1.0617,
+      "step": 3736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5817505873702695e-06,
+      "loss": 1.047,
+      "step": 3737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5807672443692938e-06,
+      "loss": 1.0412,
+      "step": 3738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.579784007360273e-06,
+      "loss": 1.0413,
+      "step": 3739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.578800876591793e-06,
+      "loss": 1.0399,
+      "step": 3740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5778178523124117e-06,
+      "loss": 1.0433,
+      "step": 3741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5768349347706615e-06,
+      "loss": 0.6599,
+      "step": 3742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5758521242150477e-06,
+      "loss": 1.0347,
+      "step": 3743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5748694208940469e-06,
+      "loss": 1.0385,
+      "step": 3744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.57388682505611e-06,
+      "loss": 1.0657,
+      "step": 3745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5729043369496608e-06,
+      "loss": 1.005,
+      "step": 3746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5719219568230955e-06,
+      "loss": 1.0011,
+      "step": 3747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5709396849247834e-06,
+      "loss": 1.0583,
+      "step": 3748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5699575215030651e-06,
+      "loss": 1.0455,
+      "step": 3749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5689754668062554e-06,
+      "loss": 1.0052,
+      "step": 3750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5679935210826405e-06,
+      "loss": 1.0083,
+      "step": 3751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5670116845804802e-06,
+      "loss": 1.0601,
+      "step": 3752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5660299575480052e-06,
+      "loss": 1.0122,
+      "step": 3753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.565048340233421e-06,
+      "loss": 1.0568,
+      "step": 3754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5640668328849012e-06,
+      "loss": 1.051,
+      "step": 3755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5630854357505954e-06,
+      "loss": 1.0575,
+      "step": 3756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5621041490786237e-06,
+      "loss": 1.0481,
+      "step": 3757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.561122973117079e-06,
+      "loss": 1.0753,
+      "step": 3758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5601419081140258e-06,
+      "loss": 1.0113,
+      "step": 3759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5591609543174998e-06,
+      "loss": 1.0476,
+      "step": 3760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5581801119755096e-06,
+      "loss": 1.0331,
+      "step": 3761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5571993813360358e-06,
+      "loss": 0.9722,
+      "step": 3762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5562187626470297e-06,
+      "loss": 1.0587,
+      "step": 3763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5552382561564155e-06,
+      "loss": 1.0535,
+      "step": 3764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.554257862112089e-06,
+      "loss": 1.0316,
+      "step": 3765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5532775807619152e-06,
+      "loss": 1.0189,
+      "step": 3766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5522974123537331e-06,
+      "loss": 1.0243,
+      "step": 3767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5513173571353532e-06,
+      "loss": 1.0753,
+      "step": 3768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5503374153545556e-06,
+      "loss": 1.0317,
+      "step": 3769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.549357587259094e-06,
+      "loss": 1.0166,
+      "step": 3770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5483778730966907e-06,
+      "loss": 0.6509,
+      "step": 3771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5473982731150417e-06,
+      "loss": 1.0194,
+      "step": 3772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5464187875618124e-06,
+      "loss": 1.0678,
+      "step": 3773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5454394166846403e-06,
+      "loss": 0.6613,
+      "step": 3774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5444601607311333e-06,
+      "loss": 1.0522,
+      "step": 3775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.543481019948871e-06,
+      "loss": 1.0143,
+      "step": 3776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5425019945854026e-06,
+      "loss": 1.0316,
+      "step": 3777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5415230848882492e-06,
+      "loss": 1.0146,
+      "step": 3778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.540544291104902e-06,
+      "loss": 0.6534,
+      "step": 3779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5395656134828237e-06,
+      "loss": 1.0799,
+      "step": 3780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5385870522694471e-06,
+      "loss": 1.0216,
+      "step": 3781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5376086077121752e-06,
+      "loss": 1.0496,
+      "step": 3782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5366302800583821e-06,
+      "loss": 1.07,
+      "step": 3783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5356520695554123e-06,
+      "loss": 1.0833,
+      "step": 3784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5346739764505806e-06,
+      "loss": 0.6377,
+      "step": 3785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5336960009911716e-06,
+      "loss": 1.064,
+      "step": 3786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5327181434244417e-06,
+      "loss": 1.0221,
+      "step": 3787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.531740403997615e-06,
+      "loss": 1.0127,
+      "step": 3788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5307627829578883e-06,
+      "loss": 1.0399,
+      "step": 3789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5297852805524265e-06,
+      "loss": 0.6263,
+      "step": 3790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5288078970283657e-06,
+      "loss": 0.6841,
+      "step": 3791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5278306326328126e-06,
+      "loss": 1.0333,
+      "step": 3792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5268534876128407e-06,
+      "loss": 1.0179,
+      "step": 3793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5258764622154964e-06,
+      "loss": 1.031,
+      "step": 3794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5248995566877949e-06,
+      "loss": 1.0446,
+      "step": 3795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5239227712767208e-06,
+      "loss": 0.6482,
+      "step": 3796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.522946106229229e-06,
+      "loss": 0.9992,
+      "step": 3797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5219695617922433e-06,
+      "loss": 1.0696,
+      "step": 3798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5209931382126565e-06,
+      "loss": 1.048,
+      "step": 3799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5200168357373327e-06,
+      "loss": 1.0076,
+      "step": 3800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5190406546131037e-06,
+      "loss": 1.0238,
+      "step": 3801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5180645950867708e-06,
+      "loss": 1.055,
+      "step": 3802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5170886574051065e-06,
+      "loss": 1.0465,
+      "step": 3803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.516112841814849e-06,
+      "loss": 1.0463,
+      "step": 3804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.515137148562708e-06,
+      "loss": 1.0488,
+      "step": 3805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5141615778953624e-06,
+      "loss": 1.0669,
+      "step": 3806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5131861300594594e-06,
+      "loss": 0.6652,
+      "step": 3807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5122108053016151e-06,
+      "loss": 0.6536,
+      "step": 3808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5112356038684151e-06,
+      "loss": 1.0619,
+      "step": 3809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5102605260064128e-06,
+      "loss": 1.0834,
+      "step": 3810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.509285571962131e-06,
+      "loss": 1.0582,
+      "step": 3811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5083107419820614e-06,
+      "loss": 1.0422,
+      "step": 3812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.507336036312664e-06,
+      "loss": 1.0493,
+      "step": 3813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5063614552003681e-06,
+      "loss": 1.061,
+      "step": 3814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5053869988915694e-06,
+      "loss": 1.051,
+      "step": 3815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5044126676326343e-06,
+      "loss": 1.0103,
+      "step": 3816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5034384616698967e-06,
+      "loss": 1.0269,
+      "step": 3817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5024643812496592e-06,
+      "loss": 0.6954,
+      "step": 3818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.501490426618192e-06,
+      "loss": 1.0285,
+      "step": 3819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5005165980217342e-06,
+      "loss": 0.6322,
+      "step": 3820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4995428957064925e-06,
+      "loss": 0.6666,
+      "step": 3821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4985693199186415e-06,
+      "loss": 1.0445,
+      "step": 3822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4975958709043247e-06,
+      "loss": 1.0324,
+      "step": 3823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4966225489096529e-06,
+      "loss": 0.68,
+      "step": 3824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4956493541807058e-06,
+      "loss": 1.0682,
+      "step": 3825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4946762869635282e-06,
+      "loss": 1.0049,
+      "step": 3826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4937033475041355e-06,
+      "loss": 1.0298,
+      "step": 3827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.49273053604851e-06,
+      "loss": 1.0617,
+      "step": 3828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4917578528426007e-06,
+      "loss": 0.6521,
+      "step": 3829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4907852981323256e-06,
+      "loss": 1.0661,
+      "step": 3830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.48981287216357e-06,
+      "loss": 1.064,
+      "step": 3831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4888405751821847e-06,
+      "loss": 1.0052,
+      "step": 3832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4878684074339905e-06,
+      "loss": 1.028,
+      "step": 3833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.486896369164774e-06,
+      "loss": 1.0392,
+      "step": 3834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4859244606202897e-06,
+      "loss": 1.0505,
+      "step": 3835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4849526820462595e-06,
+      "loss": 1.0231,
+      "step": 3836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4839810336883708e-06,
+      "loss": 1.0671,
+      "step": 3837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4830095157922796e-06,
+      "loss": 1.0253,
+      "step": 3838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4820381286036092e-06,
+      "loss": 1.0543,
+      "step": 3839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4810668723679488e-06,
+      "loss": 1.0629,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4800957473308553e-06,
+      "loss": 1.0446,
+      "step": 3841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.479124753737852e-06,
+      "loss": 1.0086,
+      "step": 3842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4781538918344287e-06,
+      "loss": 1.0184,
+      "step": 3843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4771831618660425e-06,
+      "loss": 1.0257,
+      "step": 3844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4762125640781168e-06,
+      "loss": 1.0704,
+      "step": 3845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4752420987160417e-06,
+      "loss": 1.014,
+      "step": 3846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4742717660251743e-06,
+      "loss": 1.0718,
+      "step": 3847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.473301566250837e-06,
+      "loss": 1.034,
+      "step": 3848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4723314996383193e-06,
+      "loss": 1.053,
+      "step": 3849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.471361566432877e-06,
+      "loss": 1.0294,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4703917668797322e-06,
+      "loss": 1.0681,
+      "step": 3851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4694221012240733e-06,
+      "loss": 1.0194,
+      "step": 3852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4684525697110547e-06,
+      "loss": 1.0447,
+      "step": 3853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4674831725857968e-06,
+      "loss": 1.0315,
+      "step": 3854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4665139100933861e-06,
+      "loss": 1.0692,
+      "step": 3855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.465544782478875e-06,
+      "loss": 1.0112,
+      "step": 3856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4645757899872822e-06,
+      "loss": 1.0379,
+      "step": 3857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.463606932863592e-06,
+      "loss": 1.029,
+      "step": 3858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4626382113527536e-06,
+      "loss": 1.0427,
+      "step": 3859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4616696256996835e-06,
+      "loss": 1.0651,
+      "step": 3860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.460701176149263e-06,
+      "loss": 1.0089,
+      "step": 3861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4597328629463384e-06,
+      "loss": 1.0377,
+      "step": 3862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4587646863357225e-06,
+      "loss": 1.029,
+      "step": 3863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4577966465621941e-06,
+      "loss": 1.0663,
+      "step": 3864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.456828743870495e-06,
+      "loss": 1.0348,
+      "step": 3865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4558609785053348e-06,
+      "loss": 1.0249,
+      "step": 3866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4548933507113877e-06,
+      "loss": 1.0554,
+      "step": 3867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.453925860733292e-06,
+      "loss": 0.9995,
+      "step": 3868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4529585088156533e-06,
+      "loss": 1.0213,
+      "step": 3869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.45199129520304e-06,
+      "loss": 0.648,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4510242201399867e-06,
+      "loss": 0.6697,
+      "step": 3871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.450057283870993e-06,
+      "loss": 1.0315,
+      "step": 3872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4490904866405236e-06,
+      "loss": 1.0281,
+      "step": 3873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4481238286930078e-06,
+      "loss": 1.0417,
+      "step": 3874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4471573102728389e-06,
+      "loss": 1.0428,
+      "step": 3875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4461909316243756e-06,
+      "loss": 0.6325,
+      "step": 3876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4452246929919416e-06,
+      "loss": 1.0039,
+      "step": 3877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.444258594619825e-06,
+      "loss": 0.6566,
+      "step": 3878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4432926367522781e-06,
+      "loss": 1.056,
+      "step": 3879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4423268196335185e-06,
+      "loss": 1.0268,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4413611435077266e-06,
+      "loss": 1.0653,
+      "step": 3881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4403956086190485e-06,
+      "loss": 1.0887,
+      "step": 3882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4394302152115945e-06,
+      "loss": 1.0515,
+      "step": 3883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4384649635294387e-06,
+      "loss": 0.6807,
+      "step": 3884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4374998538166198e-06,
+      "loss": 1.0012,
+      "step": 3885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4365348863171408e-06,
+      "loss": 0.6663,
+      "step": 3886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.435570061274967e-06,
+      "loss": 1.038,
+      "step": 3887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4346053789340298e-06,
+      "loss": 0.6442,
+      "step": 3888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4336408395382235e-06,
+      "loss": 1.0056,
+      "step": 3889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4326764433314066e-06,
+      "loss": 1.0356,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4317121905574014e-06,
+      "loss": 1.0054,
+      "step": 3891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4307480814599935e-06,
+      "loss": 1.0473,
+      "step": 3892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4297841162829325e-06,
+      "loss": 0.6489,
+      "step": 3893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4288202952699318e-06,
+      "loss": 1.0473,
+      "step": 3894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.427856618664668e-06,
+      "loss": 0.9905,
+      "step": 3895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4268930867107813e-06,
+      "loss": 1.0481,
+      "step": 3896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4259296996518764e-06,
+      "loss": 1.0675,
+      "step": 3897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4249664577315183e-06,
+      "loss": 1.0538,
+      "step": 3898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4240033611932388e-06,
+      "loss": 0.9958,
+      "step": 3899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4230404102805308e-06,
+      "loss": 0.7159,
+      "step": 3900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4220776052368515e-06,
+      "loss": 1.0315,
+      "step": 3901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.421114946305621e-06,
+      "loss": 1.0293,
+      "step": 3902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4201524337302213e-06,
+      "loss": 1.0538,
+      "step": 3903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4191900677539993e-06,
+      "loss": 0.9724,
+      "step": 3904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4182278486202632e-06,
+      "loss": 1.1057,
+      "step": 3905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4172657765722852e-06,
+      "loss": 1.0479,
+      "step": 3906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4163038518533e-06,
+      "loss": 1.0451,
+      "step": 3907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.415342074706505e-06,
+      "loss": 1.0446,
+      "step": 3908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4143804453750593e-06,
+      "loss": 1.057,
+      "step": 3909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.413418964102086e-06,
+      "loss": 1.0329,
+      "step": 3910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4124576311306703e-06,
+      "loss": 1.0493,
+      "step": 3911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4114964467038602e-06,
+      "loss": 1.0563,
+      "step": 3912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.410535411064666e-06,
+      "loss": 0.6626,
+      "step": 3913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4095745244560592e-06,
+      "loss": 0.6453,
+      "step": 3914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4086137871209759e-06,
+      "loss": 1.0363,
+      "step": 3915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4076531993023122e-06,
+      "loss": 1.0247,
+      "step": 3916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.406692761242928e-06,
+      "loss": 1.0429,
+      "step": 3917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4057324731856445e-06,
+      "loss": 1.06,
+      "step": 3918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4047723353732463e-06,
+      "loss": 0.6419,
+      "step": 3919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4038123480484773e-06,
+      "loss": 0.6852,
+      "step": 3920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4028525114540456e-06,
+      "loss": 1.02,
+      "step": 3921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.401892825832621e-06,
+      "loss": 0.6468,
+      "step": 3922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.400933291426834e-06,
+      "loss": 1.013,
+      "step": 3923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3999739084792785e-06,
+      "loss": 1.048,
+      "step": 3924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3990146772325087e-06,
+      "loss": 1.0283,
+      "step": 3925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3980555979290407e-06,
+      "loss": 1.0187,
+      "step": 3926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3970966708113525e-06,
+      "loss": 0.6454,
+      "step": 3927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3961378961218837e-06,
+      "loss": 0.993,
+      "step": 3928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3951792741030354e-06,
+      "loss": 0.6855,
+      "step": 3929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3942208049971708e-06,
+      "loss": 1.0259,
+      "step": 3930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3932624890466116e-06,
+      "loss": 1.0434,
+      "step": 3931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3923043264936434e-06,
+      "loss": 1.0524,
+      "step": 3932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3913463175805129e-06,
+      "loss": 1.051,
+      "step": 3933
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3903884625494268e-06,
+      "loss": 0.6327,
+      "step": 3934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3894307616425544e-06,
+      "loss": 1.0268,
+      "step": 3935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3884732151020241e-06,
+      "loss": 0.7065,
+      "step": 3936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3875158231699268e-06,
+      "loss": 1.0281,
+      "step": 3937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.386558586088314e-06,
+      "loss": 1.0117,
+      "step": 3938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3856015040991974e-06,
+      "loss": 1.0544,
+      "step": 3939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3846445774445503e-06,
+      "loss": 0.6671,
+      "step": 3940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3836878063663066e-06,
+      "loss": 1.0305,
+      "step": 3941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.38273119110636e-06,
+      "loss": 0.6572,
+      "step": 3942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.381774731906566e-06,
+      "loss": 1.0606,
+      "step": 3943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3808184290087403e-06,
+      "loss": 0.9949,
+      "step": 3944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3798622826546583e-06,
+      "loss": 1.0337,
+      "step": 3945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3789062930860568e-06,
+      "loss": 1.0343,
+      "step": 3946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3779504605446319e-06,
+      "loss": 1.0724,
+      "step": 3947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.376994785272041e-06,
+      "loss": 0.6348,
+      "step": 3948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.376039267509902e-06,
+      "loss": 1.0256,
+      "step": 3949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3750839074997917e-06,
+      "loss": 1.047,
+      "step": 3950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.374128705483248e-06,
+      "loss": 1.0514,
+      "step": 3951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3731736617017692e-06,
+      "loss": 0.9926,
+      "step": 3952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3722187763968115e-06,
+      "loss": 1.0376,
+      "step": 3953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3712640498097936e-06,
+      "loss": 1.0213,
+      "step": 3954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3703094821820925e-06,
+      "loss": 1.022,
+      "step": 3955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3693550737550455e-06,
+      "loss": 1.0198,
+      "step": 3956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3684008247699505e-06,
+      "loss": 1.0393,
+      "step": 3957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3674467354680628e-06,
+      "loss": 1.0378,
+      "step": 3958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3664928060905992e-06,
+      "loss": 1.0113,
+      "step": 3959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3655390368787362e-06,
+      "loss": 1.06,
+      "step": 3960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.364585428073608e-06,
+      "loss": 1.0153,
+      "step": 3961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.363631979916311e-06,
+      "loss": 1.0543,
+      "step": 3962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3626786926478986e-06,
+      "loss": 1.0376,
+      "step": 3963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.361725566509384e-06,
+      "loss": 1.0525,
+      "step": 3964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3607726017417408e-06,
+      "loss": 1.0517,
+      "step": 3965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3598197985859009e-06,
+      "loss": 1.052,
+      "step": 3966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3588671572827548e-06,
+      "loss": 1.027,
+      "step": 3967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.357914678073154e-06,
+      "loss": 1.0371,
+      "step": 3968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3569623611979063e-06,
+      "loss": 1.0421,
+      "step": 3969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.356010206897781e-06,
+      "loss": 1.0453,
+      "step": 3970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3550582154135044e-06,
+      "loss": 0.9913,
+      "step": 3971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3541063869857631e-06,
+      "loss": 0.65,
+      "step": 3972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3531547218552017e-06,
+      "loss": 1.0411,
+      "step": 3973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3522032202624241e-06,
+      "loss": 1.014,
+      "step": 3974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.351251882447991e-06,
+      "loss": 1.0035,
+      "step": 3975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3503007086524244e-06,
+      "loss": 0.996,
+      "step": 3976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3493496991162033e-06,
+      "loss": 1.0237,
+      "step": 3977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3483988540797646e-06,
+      "loss": 1.0347,
+      "step": 3978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3474481737835065e-06,
+      "loss": 1.0188,
+      "step": 3979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3464976584677806e-06,
+      "loss": 1.0375,
+      "step": 3980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3455473083729015e-06,
+      "loss": 1.0854,
+      "step": 3981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3445971237391396e-06,
+      "loss": 1.038,
+      "step": 3982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.343647104806724e-06,
+      "loss": 1.0247,
+      "step": 3983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3426972518158424e-06,
+      "loss": 1.0313,
+      "step": 3984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3417475650066402e-06,
+      "loss": 1.0026,
+      "step": 3985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3407980446192197e-06,
+      "loss": 1.0616,
+      "step": 3986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3398486908936425e-06,
+      "loss": 1.0848,
+      "step": 3987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3388995040699284e-06,
+      "loss": 1.0272,
+      "step": 3988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3379504843880538e-06,
+      "loss": 1.0387,
+      "step": 3989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3370016320879538e-06,
+      "loss": 1.0329,
+      "step": 3990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3360529474095191e-06,
+      "loss": 0.9922,
+      "step": 3991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3351044305926013e-06,
+      "loss": 1.0486,
+      "step": 3992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.334156081877007e-06,
+      "loss": 1.0321,
+      "step": 3993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3332079015025014e-06,
+      "loss": 1.04,
+      "step": 3994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3322598897088071e-06,
+      "loss": 0.9981,
+      "step": 3995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3313120467356042e-06,
+      "loss": 1.0299,
+      "step": 3996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.330364372822529e-06,
+      "loss": 1.0608,
+      "step": 3997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3294168682091764e-06,
+      "loss": 1.0515,
+      "step": 3998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.328469533135098e-06,
+      "loss": 1.0173,
+      "step": 3999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.327522367839802e-06,
+      "loss": 0.6582,
+      "step": 4000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3265753725627555e-06,
+      "loss": 1.033,
+      "step": 4001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3256285475433794e-06,
+      "loss": 1.0233,
+      "step": 4002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3246818930210548e-06,
+      "loss": 1.0448,
+      "step": 4003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3237354092351176e-06,
+      "loss": 1.0116,
+      "step": 4004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3227890964248619e-06,
+      "loss": 1.0661,
+      "step": 4005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3218429548295376e-06,
+      "loss": 0.6806,
+      "step": 4006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.320896984688352e-06,
+      "loss": 1.0589,
+      "step": 4007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3199511862404684e-06,
+      "loss": 1.0423,
+      "step": 4008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.319005559725007e-06,
+      "loss": 1.0608,
+      "step": 4009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3180601053810447e-06,
+      "loss": 0.6402,
+      "step": 4010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3171148234476145e-06,
+      "loss": 1.0132,
+      "step": 4011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3161697141637067e-06,
+      "loss": 1.021,
+      "step": 4012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3152247777682668e-06,
+      "loss": 1.0069,
+      "step": 4013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3142800145001967e-06,
+      "loss": 0.6448,
+      "step": 4014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3133354245983552e-06,
+      "loss": 1.0327,
+      "step": 4015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.312391008301557e-06,
+      "loss": 1.0877,
+      "step": 4016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.311446765848573e-06,
+      "loss": 1.0541,
+      "step": 4017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3105026974781301e-06,
+      "loss": 1.0382,
+      "step": 4018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3095588034289107e-06,
+      "loss": 1.0247,
+      "step": 4019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3086150839395535e-06,
+      "loss": 1.0044,
+      "step": 4020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3076715392486537e-06,
+      "loss": 1.0511,
+      "step": 4021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3067281695947613e-06,
+      "loss": 1.0167,
+      "step": 4022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3057849752163827e-06,
+      "loss": 1.028,
+      "step": 4023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3048419563519792e-06,
+      "loss": 1.091,
+      "step": 4024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3038991132399689e-06,
+      "loss": 1.0255,
+      "step": 4025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3029564461187244e-06,
+      "loss": 0.6689,
+      "step": 4026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3020139552265749e-06,
+      "loss": 0.6707,
+      "step": 4027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3010716408018035e-06,
+      "loss": 1.0531,
+      "step": 4028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3001295030826503e-06,
+      "loss": 1.0162,
+      "step": 4029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2991875423073092e-06,
+      "loss": 1.0578,
+      "step": 4030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.298245758713931e-06,
+      "loss": 1.0454,
+      "step": 4031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2973041525406201e-06,
+      "loss": 1.0389,
+      "step": 4032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2963627240254371e-06,
+      "loss": 1.014,
+      "step": 4033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2954214734063981e-06,
+      "loss": 1.0333,
+      "step": 4034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2944804009214724e-06,
+      "loss": 0.687,
+      "step": 4035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.293539506808586e-06,
+      "loss": 1.0379,
+      "step": 4036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2925987913056193e-06,
+      "loss": 1.0121,
+      "step": 4037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2916582546504072e-06,
+      "loss": 1.016,
+      "step": 4038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2907178970807394e-06,
+      "loss": 1.0663,
+      "step": 4039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2897777188343615e-06,
+      "loss": 0.6592,
+      "step": 4040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2888377201489712e-06,
+      "loss": 1.0334,
+      "step": 4041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2878979012622233e-06,
+      "loss": 1.038,
+      "step": 4042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2869582624117263e-06,
+      "loss": 1.005,
+      "step": 4043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2860188038350427e-06,
+      "loss": 1.026,
+      "step": 4044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2850795257696905e-06,
+      "loss": 0.6527,
+      "step": 4045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2841404284531413e-06,
+      "loss": 1.0633,
+      "step": 4046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2832015121228204e-06,
+      "loss": 1.0126,
+      "step": 4047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2822627770161086e-06,
+      "loss": 1.0012,
+      "step": 4048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2813242233703402e-06,
+      "loss": 1.0513,
+      "step": 4049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2803858514228036e-06,
+      "loss": 1.0576,
+      "step": 4050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2794476614107428e-06,
+      "loss": 1.0283,
+      "step": 4051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2785096535713522e-06,
+      "loss": 1.054,
+      "step": 4052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2775718281417833e-06,
+      "loss": 1.0249,
+      "step": 4053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.276634185359141e-06,
+      "loss": 1.0228,
+      "step": 4054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.275696725460483e-06,
+      "loss": 1.0188,
+      "step": 4055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2747594486828216e-06,
+      "loss": 1.0594,
+      "step": 4056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2738223552631228e-06,
+      "loss": 1.0404,
+      "step": 4057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2728854454383054e-06,
+      "loss": 1.0545,
+      "step": 4058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2719487194452425e-06,
+      "loss": 1.045,
+      "step": 4059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2710121775207607e-06,
+      "loss": 1.0331,
+      "step": 4060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.27007581990164e-06,
+      "loss": 0.9785,
+      "step": 4061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2691396468246146e-06,
+      "loss": 1.0204,
+      "step": 4062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2682036585263695e-06,
+      "loss": 1.0186,
+      "step": 4063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2672678552435454e-06,
+      "loss": 0.6297,
+      "step": 4064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2663322372127355e-06,
+      "loss": 1.0111,
+      "step": 4065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2653968046704863e-06,
+      "loss": 1.0039,
+      "step": 4066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2644615578532973e-06,
+      "loss": 1.0396,
+      "step": 4067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2635264969976215e-06,
+      "loss": 0.9815,
+      "step": 4068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2625916223398632e-06,
+      "loss": 1.0058,
+      "step": 4069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2616569341163812e-06,
+      "loss": 1.0348,
+      "step": 4070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2607224325634873e-06,
+      "loss": 1.0501,
+      "step": 4071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2597881179174455e-06,
+      "loss": 1.0189,
+      "step": 4072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2588539904144728e-06,
+      "loss": 1.015,
+      "step": 4073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2579200502907379e-06,
+      "loss": 1.0577,
+      "step": 4074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2569862977823634e-06,
+      "loss": 1.0347,
+      "step": 4075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2560527331254239e-06,
+      "loss": 0.985,
+      "step": 4076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2551193565559469e-06,
+      "loss": 0.9994,
+      "step": 4077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.254186168309912e-06,
+      "loss": 1.0374,
+      "step": 4078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2532531686232512e-06,
+      "loss": 1.0154,
+      "step": 4079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.252320357731849e-06,
+      "loss": 0.9576,
+      "step": 4080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2513877358715417e-06,
+      "loss": 1.0414,
+      "step": 4081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2504553032781185e-06,
+      "loss": 1.0288,
+      "step": 4082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2495230601873201e-06,
+      "loss": 1.0354,
+      "step": 4083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.248591006834841e-06,
+      "loss": 1.0278,
+      "step": 4084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2476591434563238e-06,
+      "loss": 0.9978,
+      "step": 4085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2467274702873671e-06,
+      "loss": 1.0329,
+      "step": 4086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.24579598756352e-06,
+      "loss": 1.0321,
+      "step": 4087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2448646955202829e-06,
+      "loss": 0.6707,
+      "step": 4088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2439335943931086e-06,
+      "loss": 0.9996,
+      "step": 4089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2430026844174022e-06,
+      "loss": 0.9767,
+      "step": 4090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2420719658285185e-06,
+      "loss": 1.0432,
+      "step": 4091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2411414388617659e-06,
+      "loss": 1.0276,
+      "step": 4092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2402111037524034e-06,
+      "loss": 1.0648,
+      "step": 4093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2392809607356419e-06,
+      "loss": 1.0264,
+      "step": 4094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2383510100466436e-06,
+      "loss": 1.0614,
+      "step": 4095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.237421251920522e-06,
+      "loss": 1.024,
+      "step": 4096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2364916865923416e-06,
+      "loss": 1.0117,
+      "step": 4097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2355623142971188e-06,
+      "loss": 1.0503,
+      "step": 4098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2346331352698205e-06,
+      "loss": 0.6484,
+      "step": 4099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2337041497453658e-06,
+      "loss": 1.0388,
+      "step": 4100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2327753579586238e-06,
+      "loss": 1.073,
+      "step": 4101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2318467601444149e-06,
+      "loss": 1.0506,
+      "step": 4102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.230918356537511e-06,
+      "loss": 1.0427,
+      "step": 4103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.229990147372634e-06,
+      "loss": 1.0272,
+      "step": 4104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2290621328844575e-06,
+      "loss": 1.0119,
+      "step": 4105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.228134313307606e-06,
+      "loss": 1.0376,
+      "step": 4106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2272066888766528e-06,
+      "loss": 1.0549,
+      "step": 4107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2262792598261244e-06,
+      "loss": 1.0005,
+      "step": 4108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2253520263904968e-06,
+      "loss": 1.0354,
+      "step": 4109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2244249888041958e-06,
+      "loss": 1.0211,
+      "step": 4110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.223498147301599e-06,
+      "loss": 1.0322,
+      "step": 4111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2225715021170342e-06,
+      "loss": 1.0151,
+      "step": 4112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2216450534847785e-06,
+      "loss": 1.0211,
+      "step": 4113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2207188016390598e-06,
+      "loss": 1.0726,
+      "step": 4114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2197927468140572e-06,
+      "loss": 1.0598,
+      "step": 4115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.218866889243899e-06,
+      "loss": 1.0567,
+      "step": 4116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2179412291626646e-06,
+      "loss": 1.0488,
+      "step": 4117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2170157668043812e-06,
+      "loss": 1.0667,
+      "step": 4118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2160905024030286e-06,
+      "loss": 1.0669,
+      "step": 4119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2151654361925353e-06,
+      "loss": 1.0193,
+      "step": 4120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.21424056840678e-06,
+      "loss": 1.0065,
+      "step": 4121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.213315899279591e-06,
+      "loss": 1.0192,
+      "step": 4122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2123914290447477e-06,
+      "loss": 1.0249,
+      "step": 4123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.211467157935976e-06,
+      "loss": 1.0194,
+      "step": 4124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2105430861869545e-06,
+      "loss": 0.6076,
+      "step": 4125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2096192140313104e-06,
+      "loss": 1.0262,
+      "step": 4126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2086955417026202e-06,
+      "loss": 1.0225,
+      "step": 4127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.207772069434411e-06,
+      "loss": 1.0364,
+      "step": 4128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.206848797460157e-06,
+      "loss": 1.0539,
+      "step": 4129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.205925726013284e-06,
+      "loss": 0.9993,
+      "step": 4130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2050028553271662e-06,
+      "loss": 1.0687,
+      "step": 4131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2040801856351272e-06,
+      "loss": 1.0118,
+      "step": 4132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2031577171704398e-06,
+      "loss": 1.0554,
+      "step": 4133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2022354501663263e-06,
+      "loss": 1.0473,
+      "step": 4134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2013133848559563e-06,
+      "loss": 0.9951,
+      "step": 4135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2003915214724502e-06,
+      "loss": 1.0442,
+      "step": 4136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1994698602488774e-06,
+      "loss": 0.7076,
+      "step": 4137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.198548401418255e-06,
+      "loss": 1.0432,
+      "step": 4138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1976271452135506e-06,
+      "loss": 1.0161,
+      "step": 4139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1967060918676784e-06,
+      "loss": 1.0557,
+      "step": 4140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1957852416135028e-06,
+      "loss": 0.6343,
+      "step": 4141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1948645946838369e-06,
+      "loss": 0.6641,
+      "step": 4142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1939441513114416e-06,
+      "loss": 1.0484,
+      "step": 4143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1930239117290271e-06,
+      "loss": 1.0433,
+      "step": 4144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.192103876169252e-06,
+      "loss": 1.0507,
+      "step": 4145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1911840448647224e-06,
+      "loss": 1.0232,
+      "step": 4146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1902644180479933e-06,
+      "loss": 1.0478,
+      "step": 4147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1893449959515685e-06,
+      "loss": 1.0083,
+      "step": 4148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1884257788078992e-06,
+      "loss": 1.0463,
+      "step": 4149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1875067668493865e-06,
+      "loss": 1.023,
+      "step": 4150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.186587960308377e-06,
+      "loss": 1.0042,
+      "step": 4151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.185669359417167e-06,
+      "loss": 1.0187,
+      "step": 4152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1847509644080008e-06,
+      "loss": 0.6722,
+      "step": 4153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1838327755130699e-06,
+      "loss": 1.0089,
+      "step": 4154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1829147929645152e-06,
+      "loss": 1.0474,
+      "step": 4155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.181997016994424e-06,
+      "loss": 1.0034,
+      "step": 4156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.181079447834831e-06,
+      "loss": 0.9933,
+      "step": 4157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.18016208571772e-06,
+      "loss": 1.029,
+      "step": 4158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1792449308750217e-06,
+      "loss": 1.0073,
+      "step": 4159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1783279835386145e-06,
+      "loss": 1.0202,
+      "step": 4160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1774112439403245e-06,
+      "loss": 1.0318,
+      "step": 4161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1764947123119251e-06,
+      "loss": 1.0201,
+      "step": 4162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1755783888851371e-06,
+      "loss": 1.02,
+      "step": 4163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.174662273891629e-06,
+      "loss": 1.0371,
+      "step": 4164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1737463675630156e-06,
+      "loss": 1.0522,
+      "step": 4165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1728306701308608e-06,
+      "loss": 1.0171,
+      "step": 4166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1719151818266744e-06,
+      "loss": 1.0637,
+      "step": 4167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1709999028819124e-06,
+      "loss": 1.0488,
+      "step": 4168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1700848335279795e-06,
+      "loss": 1.0159,
+      "step": 4169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1691699739962272e-06,
+      "loss": 1.0406,
+      "step": 4170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1682553245179537e-06,
+      "loss": 1.0247,
+      "step": 4171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1673408853244035e-06,
+      "loss": 1.0222,
+      "step": 4172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1664266566467686e-06,
+      "loss": 0.6491,
+      "step": 4173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.165512638716188e-06,
+      "loss": 1.0315,
+      "step": 4174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1645988317637465e-06,
+      "loss": 1.0498,
+      "step": 4175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1636852360204764e-06,
+      "loss": 1.0323,
+      "step": 4176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1627718517173564e-06,
+      "loss": 1.0412,
+      "step": 4177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1618586790853123e-06,
+      "loss": 1.0269,
+      "step": 4178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.160945718355214e-06,
+      "loss": 1.0269,
+      "step": 4179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1600329697578808e-06,
+      "loss": 1.0632,
+      "step": 4180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1591204335240768e-06,
+      "loss": 0.6556,
+      "step": 4181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1582081098845127e-06,
+      "loss": 0.6401,
+      "step": 4182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1572959990698458e-06,
+      "loss": 1.0255,
+      "step": 4183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.156384101310679e-06,
+      "loss": 0.6477,
+      "step": 4184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1554724168375617e-06,
+      "loss": 1.0318,
+      "step": 4185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1545609458809903e-06,
+      "loss": 1.0192,
+      "step": 4186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1536496886714044e-06,
+      "loss": 1.058,
+      "step": 4187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.152738645439192e-06,
+      "loss": 1.0376,
+      "step": 4188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1518278164146866e-06,
+      "loss": 1.0416,
+      "step": 4189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1509172018281673e-06,
+      "loss": 1.0408,
+      "step": 4190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1500068019098593e-06,
+      "loss": 1.0257,
+      "step": 4191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1490966168899326e-06,
+      "loss": 1.0442,
+      "step": 4192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1481866469985035e-06,
+      "loss": 0.9979,
+      "step": 4193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1472768924656357e-06,
+      "loss": 0.6298,
+      "step": 4194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1463673535213342e-06,
+      "loss": 0.6065,
+      "step": 4195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1454580303955526e-06,
+      "loss": 1.0441,
+      "step": 4196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1445489233181898e-06,
+      "loss": 1.0386,
+      "step": 4197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1436400325190896e-06,
+      "loss": 1.0592,
+      "step": 4198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1427313582280405e-06,
+      "loss": 0.6439,
+      "step": 4199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1418229006747783e-06,
+      "loss": 1.0359,
+      "step": 4200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1409146600889808e-06,
+      "loss": 0.9815,
+      "step": 4201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1400066367002729e-06,
+      "loss": 1.027,
+      "step": 4202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1390988307382254e-06,
+      "loss": 1.0507,
+      "step": 4203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1381912424323523e-06,
+      "loss": 1.014,
+      "step": 4204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1372838720121139e-06,
+      "loss": 0.6743,
+      "step": 4205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1363767197069146e-06,
+      "loss": 0.6449,
+      "step": 4206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1354697857461038e-06,
+      "loss": 1.0391,
+      "step": 4207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1345630703589764e-06,
+      "loss": 1.0417,
+      "step": 4208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.133656573774771e-06,
+      "loss": 1.0067,
+      "step": 4209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1327502962226715e-06,
+      "loss": 0.9972,
+      "step": 4210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1318442379318074e-06,
+      "loss": 1.0433,
+      "step": 4211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1309383991312496e-06,
+      "loss": 1.014,
+      "step": 4212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1300327800500168e-06,
+      "loss": 0.6592,
+      "step": 4213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1291273809170702e-06,
+      "loss": 1.0327,
+      "step": 4214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.128222201961317e-06,
+      "loss": 1.0588,
+      "step": 4215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1273172434116082e-06,
+      "loss": 1.0493,
+      "step": 4216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1264125054967373e-06,
+      "loss": 1.0481,
+      "step": 4217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1255079884454439e-06,
+      "loss": 1.0296,
+      "step": 4218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1246036924864117e-06,
+      "loss": 1.0211,
+      "step": 4219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1236996178482678e-06,
+      "loss": 0.9935,
+      "step": 4220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.122795764759584e-06,
+      "loss": 1.0255,
+      "step": 4221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1218921334488762e-06,
+      "loss": 1.0402,
+      "step": 4222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1209887241446022e-06,
+      "loss": 1.0394,
+      "step": 4223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1200855370751662e-06,
+      "loss": 1.0353,
+      "step": 4224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.119182572468915e-06,
+      "loss": 1.0003,
+      "step": 4225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1182798305541396e-06,
+      "loss": 1.0025,
+      "step": 4226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1173773115590748e-06,
+      "loss": 1.0305,
+      "step": 4227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.116475015711898e-06,
+      "loss": 0.6496,
+      "step": 4228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1155729432407314e-06,
+      "loss": 1.0012,
+      "step": 4229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1146710943736403e-06,
+      "loss": 1.0327,
+      "step": 4230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1137694693386337e-06,
+      "loss": 1.0288,
+      "step": 4231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.112868068363663e-06,
+      "loss": 1.0325,
+      "step": 4232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1119668916766247e-06,
+      "loss": 1.0301,
+      "step": 4233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1110659395053564e-06,
+      "loss": 1.0077,
+      "step": 4234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1101652120776403e-06,
+      "loss": 1.0185,
+      "step": 4235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1092647096212022e-06,
+      "loss": 1.0391,
+      "step": 4236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.10836443236371e-06,
+      "loss": 1.0241,
+      "step": 4237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.107464380532776e-06,
+      "loss": 1.0186,
+      "step": 4238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1065645543559532e-06,
+      "loss": 1.0256,
+      "step": 4239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1056649540607394e-06,
+      "loss": 0.6709,
+      "step": 4240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.104765579874575e-06,
+      "loss": 1.02,
+      "step": 4241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.103866432024843e-06,
+      "loss": 1.0204,
+      "step": 4242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1029675107388693e-06,
+      "loss": 0.6354,
+      "step": 4243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1020688162439236e-06,
+      "loss": 1.0378,
+      "step": 4244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1011703487672149e-06,
+      "loss": 0.6546,
+      "step": 4245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.100272108535898e-06,
+      "loss": 1.0619,
+      "step": 4246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0993740957770693e-06,
+      "loss": 1.0226,
+      "step": 4247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0984763107177677e-06,
+      "loss": 1.0283,
+      "step": 4248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0975787535849747e-06,
+      "loss": 1.0355,
+      "step": 4249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0966814246056135e-06,
+      "loss": 0.9926,
+      "step": 4250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0957843240065506e-06,
+      "loss": 1.0015,
+      "step": 4251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.094887452014594e-06,
+      "loss": 0.6477,
+      "step": 4252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0939908088564942e-06,
+      "loss": 1.0277,
+      "step": 4253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0930943947589438e-06,
+      "loss": 1.0372,
+      "step": 4254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0921982099485783e-06,
+      "loss": 1.0804,
+      "step": 4255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0913022546519727e-06,
+      "loss": 1.0161,
+      "step": 4256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0904065290956462e-06,
+      "loss": 0.6619,
+      "step": 4257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.08951103350606e-06,
+      "loss": 1.0472,
+      "step": 4258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0886157681096163e-06,
+      "loss": 1.0418,
+      "step": 4259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0877207331326603e-06,
+      "loss": 1.0513,
+      "step": 4260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.086825928801476e-06,
+      "loss": 1.0384,
+      "step": 4261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0859313553422923e-06,
+      "loss": 1.0762,
+      "step": 4262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0850370129812786e-06,
+      "loss": 1.042,
+      "step": 4263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0841429019445455e-06,
+      "loss": 0.6595,
+      "step": 4264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0832490224581455e-06,
+      "loss": 1.0461,
+      "step": 4265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0823553747480725e-06,
+      "loss": 1.0405,
+      "step": 4266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.081461959040262e-06,
+      "loss": 1.0281,
+      "step": 4267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0805687755605908e-06,
+      "loss": 1.0317,
+      "step": 4268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0796758245348757e-06,
+      "loss": 1.045,
+      "step": 4269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0787831061888763e-06,
+      "loss": 1.027,
+      "step": 4270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0778906207482936e-06,
+      "loss": 1.0859,
+      "step": 4271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0769983684387682e-06,
+      "loss": 0.6432,
+      "step": 4272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.076106349485883e-06,
+      "loss": 1.0367,
+      "step": 4273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0752145641151615e-06,
+      "loss": 1.0208,
+      "step": 4274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0743230125520684e-06,
+      "loss": 1.0467,
+      "step": 4275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.073431695022008e-06,
+      "loss": 1.0352,
+      "step": 4276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0725406117503286e-06,
+      "loss": 1.0113,
+      "step": 4277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0716497629623145e-06,
+      "loss": 1.0455,
+      "step": 4278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0707591488831945e-06,
+      "loss": 0.6504,
+      "step": 4279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.069868769738137e-06,
+      "loss": 0.9705,
+      "step": 4280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.068978625752251e-06,
+      "loss": 1.0362,
+      "step": 4281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.068088717150586e-06,
+      "loss": 1.0529,
+      "step": 4282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0671990441581314e-06,
+      "loss": 1.0156,
+      "step": 4283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0663096069998173e-06,
+      "loss": 1.0394,
+      "step": 4284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0654204059005157e-06,
+      "loss": 1.0221,
+      "step": 4285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0645314410850366e-06,
+      "loss": 1.052,
+      "step": 4286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0636427127781317e-06,
+      "loss": 1.0008,
+      "step": 4287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0627542212044927e-06,
+      "loss": 0.9902,
+      "step": 4288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0618659665887507e-06,
+      "loss": 1.0167,
+      "step": 4289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0609779491554784e-06,
+      "loss": 1.0203,
+      "step": 4290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.060090169129187e-06,
+      "loss": 1.0108,
+      "step": 4291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0592026267343284e-06,
+      "loss": 1.0183,
+      "step": 4292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0583153221952949e-06,
+      "loss": 1.017,
+      "step": 4293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0574282557364167e-06,
+      "loss": 1.0573,
+      "step": 4294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0565414275819663e-06,
+      "loss": 1.0393,
+      "step": 4295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0556548379561542e-06,
+      "loss": 1.0313,
+      "step": 4296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0547684870831312e-06,
+      "loss": 1.0407,
+      "step": 4297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0538823751869883e-06,
+      "loss": 0.9991,
+      "step": 4298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0529965024917563e-06,
+      "loss": 1.0126,
+      "step": 4299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0521108692214028e-06,
+      "loss": 1.0126,
+      "step": 4300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0512254755998378e-06,
+      "loss": 1.0715,
+      "step": 4301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0503403218509098e-06,
+      "loss": 1.0009,
+      "step": 4302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0494554081984066e-06,
+      "loss": 1.0208,
+      "step": 4303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0485707348660558e-06,
+      "loss": 1.0438,
+      "step": 4304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0476863020775226e-06,
+      "loss": 1.0274,
+      "step": 4305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0468021100564131e-06,
+      "loss": 1.0359,
+      "step": 4306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.045918159026272e-06,
+      "loss": 0.9983,
+      "step": 4307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0450344492105826e-06,
+      "loss": 1.0374,
+      "step": 4308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0441509808327684e-06,
+      "loss": 1.035,
+      "step": 4309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0432677541161906e-06,
+      "loss": 0.6606,
+      "step": 4310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0423847692841502e-06,
+      "loss": 1.0347,
+      "step": 4311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.041502026559886e-06,
+      "loss": 0.997,
+      "step": 4312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0406195261665768e-06,
+      "loss": 1.0209,
+      "step": 4313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0397372683273395e-06,
+      "loss": 1.0218,
+      "step": 4314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0388552532652305e-06,
+      "loss": 1.0584,
+      "step": 4315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0379734812032418e-06,
+      "loss": 1.0318,
+      "step": 4316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.037091952364308e-06,
+      "loss": 1.0812,
+      "step": 4317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0362106669713002e-06,
+      "loss": 1.0371,
+      "step": 4318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.035329625247028e-06,
+      "loss": 1.0258,
+      "step": 4319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0344488274142395e-06,
+      "loss": 1.029,
+      "step": 4320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0335682736956221e-06,
+      "loss": 0.6622,
+      "step": 4321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.032687964313799e-06,
+      "loss": 1.0234,
+      "step": 4322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0318078994913344e-06,
+      "loss": 1.0206,
+      "step": 4323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0309280794507292e-06,
+      "loss": 1.018,
+      "step": 4324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0300485044144226e-06,
+      "loss": 1.0468,
+      "step": 4325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0291691746047928e-06,
+      "loss": 1.0248,
+      "step": 4326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0282900902441537e-06,
+      "loss": 1.0303,
+      "step": 4327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.027411251554759e-06,
+      "loss": 1.0448,
+      "step": 4328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0265326587588008e-06,
+      "loss": 0.9864,
+      "step": 4329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0256543120784074e-06,
+      "loss": 1.0515,
+      "step": 4330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0247762117356455e-06,
+      "loss": 1.01,
+      "step": 4331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0238983579525204e-06,
+      "loss": 1.0132,
+      "step": 4332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0230207509509737e-06,
+      "loss": 1.0101,
+      "step": 4333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0221433909528852e-06,
+      "loss": 1.0641,
+      "step": 4334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0212662781800726e-06,
+      "loss": 1.0199,
+      "step": 4335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0203894128542905e-06,
+      "loss": 1.0057,
+      "step": 4336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0195127951972317e-06,
+      "loss": 1.0305,
+      "step": 4337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0186364254305248e-06,
+      "loss": 0.9957,
+      "step": 4338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0177603037757377e-06,
+      "loss": 1.0219,
+      "step": 4339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0168844304543743e-06,
+      "loss": 0.9915,
+      "step": 4340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.016008805687876e-06,
+      "loss": 1.0278,
+      "step": 4341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0151334296976215e-06,
+      "loss": 1.0312,
+      "step": 4342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0142583027049278e-06,
+      "loss": 1.0377,
+      "step": 4343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0133834249310456e-06,
+      "loss": 1.0022,
+      "step": 4344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0125087965971658e-06,
+      "loss": 1.0624,
+      "step": 4345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.011634417924415e-06,
+      "loss": 1.0438,
+      "step": 4346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0107602891338568e-06,
+      "loss": 1.0236,
+      "step": 4347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0098864104464915e-06,
+      "loss": 0.6816,
+      "step": 4348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0090127820832565e-06,
+      "loss": 1.0148,
+      "step": 4349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0081394042650255e-06,
+      "loss": 1.0618,
+      "step": 4350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0072662772126102e-06,
+      "loss": 1.0278,
+      "step": 4351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0063934011467558e-06,
+      "loss": 1.0082,
+      "step": 4352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0055207762881472e-06,
+      "loss": 1.0193,
+      "step": 4353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0046484028574043e-06,
+      "loss": 1.0471,
+      "step": 4354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0037762810750838e-06,
+      "loss": 1.0217,
+      "step": 4355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0029044111616787e-06,
+      "loss": 1.0367,
+      "step": 4356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0020327933376184e-06,
+      "loss": 0.9848,
+      "step": 4357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0011614278232686e-06,
+      "loss": 1.0506,
+      "step": 4358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0002903148389315e-06,
+      "loss": 1.0346,
+      "step": 4359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.994194546048437e-07,
+      "loss": 1.0028,
+      "step": 4360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.985488473411805e-07,
+      "loss": 1.0153,
+      "step": 4361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.976784932680512e-07,
+      "loss": 1.0431,
+      "step": 4362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.968083926055024e-07,
+      "loss": 0.6192,
+      "step": 4363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.95938545573516e-07,
+      "loss": 1.0443,
+      "step": 4364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.950689523920111e-07,
+      "loss": 1.0205,
+      "step": 4365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.94199613280839e-07,
+      "loss": 1.0114,
+      "step": 4366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.933305284597905e-07,
+      "loss": 1.039,
+      "step": 4367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.924616981485908e-07,
+      "loss": 1.0406,
+      "step": 4368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.915931225669005e-07,
+      "loss": 1.053,
+      "step": 4369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.907248019343165e-07,
+      "loss": 1.0367,
+      "step": 4370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.898567364703703e-07,
+      "loss": 1.05,
+      "step": 4371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.889889263945298e-07,
+      "loss": 1.0041,
+      "step": 4372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.881213719261973e-07,
+      "loss": 1.0018,
+      "step": 4373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.87254073284712e-07,
+      "loss": 1.0023,
+      "step": 4374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863870306893465e-07,
+      "loss": 1.0322,
+      "step": 4375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.85520244359311e-07,
+      "loss": 1.0303,
+      "step": 4376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.846537145137485e-07,
+      "loss": 1.0418,
+      "step": 4377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.83787441371738e-07,
+      "loss": 1.0058,
+      "step": 4378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.829214251522945e-07,
+      "loss": 1.0502,
+      "step": 4379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.82055666074367e-07,
+      "loss": 1.0521,
+      "step": 4380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.81190164356841e-07,
+      "loss": 1.0326,
+      "step": 4381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.80324920218534e-07,
+      "loss": 1.0164,
+      "step": 4382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.79459933878201e-07,
+      "loss": 1.0397,
+      "step": 4383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.785952055545312e-07,
+      "loss": 1.0587,
+      "step": 4384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.777307354661483e-07,
+      "loss": 0.6392,
+      "step": 4385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.768665238316103e-07,
+      "loss": 0.648,
+      "step": 4386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.760025708694118e-07,
+      "loss": 1.0169,
+      "step": 4387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.751388767979787e-07,
+      "loss": 1.0204,
+      "step": 4388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.74275441835674e-07,
+      "loss": 0.6681,
+      "step": 4389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.734122662007946e-07,
+      "loss": 1.0159,
+      "step": 4390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.725493501115715e-07,
+      "loss": 0.6619,
+      "step": 4391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.716866937861707e-07,
+      "loss": 1.0494,
+      "step": 4392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.708242974426918e-07,
+      "loss": 1.0702,
+      "step": 4393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.699621612991692e-07,
+      "loss": 1.0406,
+      "step": 4394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.691002855735707e-07,
+      "loss": 1.0242,
+      "step": 4395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.682386704837998e-07,
+      "loss": 0.6357,
+      "step": 4396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.673773162476927e-07,
+      "loss": 1.0181,
+      "step": 4397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.665162230830205e-07,
+      "loss": 0.6581,
+      "step": 4398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.65655391207487e-07,
+      "loss": 1.0688,
+      "step": 4399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.647948208387313e-07,
+      "loss": 1.0324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.639345121943263e-07,
+      "loss": 1.0037,
+      "step": 4401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.63074465491778e-07,
+      "loss": 1.012,
+      "step": 4402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.622146809485272e-07,
+      "loss": 1.0521,
+      "step": 4403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.613551587819467e-07,
+      "loss": 0.6525,
+      "step": 4404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.604958992093444e-07,
+      "loss": 1.0582,
+      "step": 4405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.596369024479615e-07,
+      "loss": 1.0061,
+      "step": 4406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.58778168714973e-07,
+      "loss": 1.0292,
+      "step": 4407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.579196982274869e-07,
+      "loss": 1.0193,
+      "step": 4408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.570614912025454e-07,
+      "loss": 1.043,
+      "step": 4409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.562035478571223e-07,
+      "loss": 0.9983,
+      "step": 4410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.553458684081267e-07,
+      "loss": 1.0441,
+      "step": 4411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.544884530723998e-07,
+      "loss": 1.0423,
+      "step": 4412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.536313020667171e-07,
+      "loss": 1.0423,
+      "step": 4413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.527744156077863e-07,
+      "loss": 1.0432,
+      "step": 4414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.519177939122487e-07,
+      "loss": 0.641,
+      "step": 4415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.510614371966786e-07,
+      "loss": 1.0362,
+      "step": 4416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.502053456775829e-07,
+      "loss": 1.0358,
+      "step": 4417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.493495195714021e-07,
+      "loss": 1.0181,
+      "step": 4418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.484939590945091e-07,
+      "loss": 1.0109,
+      "step": 4419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.476386644632107e-07,
+      "loss": 1.0148,
+      "step": 4420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.467836358937438e-07,
+      "loss": 1.0798,
+      "step": 4421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.45928873602281e-07,
+      "loss": 1.0193,
+      "step": 4422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.450743778049264e-07,
+      "loss": 0.9961,
+      "step": 4423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.442201487177163e-07,
+      "loss": 0.9919,
+      "step": 4424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.433661865566211e-07,
+      "loss": 1.0415,
+      "step": 4425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.425124915375413e-07,
+      "loss": 1.0373,
+      "step": 4426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.416590638763116e-07,
+      "loss": 1.0142,
+      "step": 4427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.408059037886988e-07,
+      "loss": 1.0512,
+      "step": 4428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.399530114904022e-07,
+      "loss": 1.0556,
+      "step": 4429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.391003871970526e-07,
+      "loss": 1.0058,
+      "step": 4430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.38248031124214e-07,
+      "loss": 1.0059,
+      "step": 4431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.373959434873821e-07,
+      "loss": 1.0258,
+      "step": 4432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.365441245019847e-07,
+      "loss": 1.0131,
+      "step": 4433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.356925743833827e-07,
+      "loss": 1.0235,
+      "step": 4434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.348412933468666e-07,
+      "loss": 1.0013,
+      "step": 4435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.339902816076608e-07,
+      "loss": 1.0258,
+      "step": 4436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.331395393809216e-07,
+      "loss": 1.0387,
+      "step": 4437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.322890668817365e-07,
+      "loss": 1.0336,
+      "step": 4438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.314388643251252e-07,
+      "loss": 1.0177,
+      "step": 4439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.305889319260391e-07,
+      "loss": 1.0353,
+      "step": 4440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.29739269899361e-07,
+      "loss": 1.0371,
+      "step": 4441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.288898784599062e-07,
+      "loss": 1.0421,
+      "step": 4442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.280407578224198e-07,
+      "loss": 1.0559,
+      "step": 4443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.2719190820158e-07,
+      "loss": 1.0566,
+      "step": 4444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.263433298119963e-07,
+      "loss": 1.0509,
+      "step": 4445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.254950228682092e-07,
+      "loss": 1.0251,
+      "step": 4446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.246469875846918e-07,
+      "loss": 1.0323,
+      "step": 4447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.237992241758456e-07,
+      "loss": 0.988,
+      "step": 4448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.229517328560063e-07,
+      "loss": 1.0023,
+      "step": 4449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.221045138394394e-07,
+      "loss": 0.9766,
+      "step": 4450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.212575673403424e-07,
+      "loss": 1.0045,
+      "step": 4451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.20410893572843e-07,
+      "loss": 0.9951,
+      "step": 4452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.195644927510003e-07,
+      "loss": 0.9822,
+      "step": 4453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.187183650888046e-07,
+      "loss": 1.0188,
+      "step": 4454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.178725108001773e-07,
+      "loss": 1.0252,
+      "step": 4455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.170269300989699e-07,
+      "loss": 0.6888,
+      "step": 4456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.161816231989652e-07,
+      "loss": 1.0149,
+      "step": 4457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.153365903138779e-07,
+      "loss": 1.0212,
+      "step": 4458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.144918316573504e-07,
+      "loss": 1.0288,
+      "step": 4459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.136473474429587e-07,
+      "loss": 1.0098,
+      "step": 4460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.128031378842083e-07,
+      "loss": 1.0339,
+      "step": 4461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.119592031945353e-07,
+      "loss": 1.0321,
+      "step": 4462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.111155435873064e-07,
+      "loss": 0.6729,
+      "step": 4463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.102721592758193e-07,
+      "loss": 1.0489,
+      "step": 4464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.094290504733003e-07,
+      "loss": 1.04,
+      "step": 4465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.085862173929075e-07,
+      "loss": 1.0159,
+      "step": 4466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.077436602477295e-07,
+      "loss": 1.0561,
+      "step": 4467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.069013792507847e-07,
+      "loss": 1.0044,
+      "step": 4468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.060593746150221e-07,
+      "loss": 1.0063,
+      "step": 4469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.052176465533188e-07,
+      "loss": 1.0275,
+      "step": 4470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.043761952784848e-07,
+      "loss": 0.9591,
+      "step": 4471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.035350210032584e-07,
+      "loss": 0.6603,
+      "step": 4472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.026941239403086e-07,
+      "loss": 1.0374,
+      "step": 4473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.01853504302234e-07,
+      "loss": 1.0027,
+      "step": 4474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.010131623015631e-07,
+      "loss": 0.6598,
+      "step": 4475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.001730981507543e-07,
+      "loss": 0.6527,
+      "step": 4476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.993333120621956e-07,
+      "loss": 0.6501,
+      "step": 4477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.984938042482048e-07,
+      "loss": 1.0115,
+      "step": 4478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.976545749210294e-07,
+      "loss": 1.0009,
+      "step": 4479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.968156242928473e-07,
+      "loss": 1.0114,
+      "step": 4480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.959769525757632e-07,
+      "loss": 1.0294,
+      "step": 4481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.95138559981814e-07,
+      "loss": 0.9857,
+      "step": 4482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.943004467229654e-07,
+      "loss": 1.0313,
+      "step": 4483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.93462613011112e-07,
+      "loss": 0.9882,
+      "step": 4484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.926250590580782e-07,
+      "loss": 1.0279,
+      "step": 4485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.917877850756179e-07,
+      "loss": 0.996,
+      "step": 4486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.909507912754126e-07,
+      "loss": 1.017,
+      "step": 4487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.901140778690747e-07,
+      "loss": 1.0284,
+      "step": 4488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.892776450681452e-07,
+      "loss": 0.9619,
+      "step": 4489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.88441493084094e-07,
+      "loss": 0.6294,
+      "step": 4490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.876056221283208e-07,
+      "loss": 0.6388,
+      "step": 4491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.867700324121521e-07,
+      "loss": 1.0193,
+      "step": 4492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.859347241468458e-07,
+      "loss": 1.0605,
+      "step": 4493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.85099697543587e-07,
+      "loss": 1.0334,
+      "step": 4494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.842649528134906e-07,
+      "loss": 1.0492,
+      "step": 4495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.834304901675997e-07,
+      "loss": 1.034,
+      "step": 4496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.825963098168863e-07,
+      "loss": 1.0148,
+      "step": 4497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.817624119722507e-07,
+      "loss": 1.0276,
+      "step": 4498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.80928796844522e-07,
+      "loss": 1.0193,
+      "step": 4499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.80095464644458e-07,
+      "loss": 1.0107,
+      "step": 4500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.792624155827446e-07,
+      "loss": 1.0097,
+      "step": 4501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.784296498699975e-07,
+      "loss": 0.9863,
+      "step": 4502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.775971677167573e-07,
+      "loss": 0.9898,
+      "step": 4503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.767649693334967e-07,
+      "loss": 1.0408,
+      "step": 4504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.759330549306146e-07,
+      "loss": 0.9841,
+      "step": 4505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.751014247184387e-07,
+      "loss": 1.0206,
+      "step": 4506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.74270078907225e-07,
+      "loss": 1.0158,
+      "step": 4507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.73439017707158e-07,
+      "loss": 1.0761,
+      "step": 4508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.726082413283482e-07,
+      "loss": 1.0338,
+      "step": 4509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.717777499808363e-07,
+      "loss": 1.002,
+      "step": 4510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.709475438745899e-07,
+      "loss": 0.6641,
+      "step": 4511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.701176232195049e-07,
+      "loss": 1.022,
+      "step": 4512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.692879882254049e-07,
+      "loss": 0.9823,
+      "step": 4513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.684586391020414e-07,
+      "loss": 1.0395,
+      "step": 4514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.676295760590932e-07,
+      "loss": 0.6335,
+      "step": 4515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.668007993061672e-07,
+      "loss": 1.0055,
+      "step": 4516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.659723090527982e-07,
+      "loss": 1.023,
+      "step": 4517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.651441055084472e-07,
+      "loss": 1.0603,
+      "step": 4518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.643161888825039e-07,
+      "loss": 0.6293,
+      "step": 4519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.634885593842854e-07,
+      "loss": 1.0356,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.626612172230359e-07,
+      "loss": 1.0385,
+      "step": 4521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.618341626079273e-07,
+      "loss": 1.0393,
+      "step": 4522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.610073957480581e-07,
+      "loss": 1.0427,
+      "step": 4523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.601809168524558e-07,
+      "loss": 1.038,
+      "step": 4524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.593547261300723e-07,
+      "loss": 0.6438,
+      "step": 4525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.585288237897883e-07,
+      "loss": 0.6521,
+      "step": 4526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.57703210040412e-07,
+      "loss": 1.0052,
+      "step": 4527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.568778850906779e-07,
+      "loss": 0.6543,
+      "step": 4528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.560528491492478e-07,
+      "loss": 1.0684,
+      "step": 4529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.552281024247107e-07,
+      "loss": 1.0568,
+      "step": 4530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.544036451255809e-07,
+      "loss": 1.016,
+      "step": 4531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.535794774603014e-07,
+      "loss": 1.0415,
+      "step": 4532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.527555996372413e-07,
+      "loss": 1.0166,
+      "step": 4533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.519320118646963e-07,
+      "loss": 1.0154,
+      "step": 4534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.511087143508886e-07,
+      "loss": 1.019,
+      "step": 4535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.502857073039678e-07,
+      "loss": 1.0187,
+      "step": 4536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.494629909320093e-07,
+      "loss": 1.0363,
+      "step": 4537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.486405654430153e-07,
+      "loss": 1.0176,
+      "step": 4538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.478184310449143e-07,
+      "loss": 1.0253,
+      "step": 4539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.469965879455614e-07,
+      "loss": 1.0444,
+      "step": 4540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.461750363527389e-07,
+      "loss": 1.059,
+      "step": 4541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.453537764741525e-07,
+      "loss": 1.0287,
+      "step": 4542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.445328085174374e-07,
+      "loss": 1.0017,
+      "step": 4543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.437121326901535e-07,
+      "loss": 0.9972,
+      "step": 4544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.428917491997868e-07,
+      "loss": 1.0211,
+      "step": 4545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.420716582537502e-07,
+      "loss": 1.0183,
+      "step": 4546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.412518600593821e-07,
+      "loss": 1.0096,
+      "step": 4547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.40432354823946e-07,
+      "loss": 1.0308,
+      "step": 4548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.396131427546329e-07,
+      "loss": 0.627,
+      "step": 4549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.387942240585586e-07,
+      "loss": 0.9937,
+      "step": 4550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.379755989427655e-07,
+      "loss": 0.6681,
+      "step": 4551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.371572676142218e-07,
+      "loss": 1.0239,
+      "step": 4552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.363392302798198e-07,
+      "loss": 0.6222,
+      "step": 4553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.355214871463794e-07,
+      "loss": 1.0343,
+      "step": 4554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.347040384206452e-07,
+      "loss": 1.01,
+      "step": 4555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.338868843092879e-07,
+      "loss": 1.0252,
+      "step": 4556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.330700250189029e-07,
+      "loss": 1.0331,
+      "step": 4557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.322534607560119e-07,
+      "loss": 0.6583,
+      "step": 4558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.314371917270618e-07,
+      "loss": 1.0093,
+      "step": 4559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.306212181384243e-07,
+      "loss": 1.0678,
+      "step": 4560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.298055401963968e-07,
+      "loss": 1.0036,
+      "step": 4561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.289901581072025e-07,
+      "loss": 1.0308,
+      "step": 4562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.281750720769895e-07,
+      "loss": 0.6009,
+      "step": 4563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.273602823118297e-07,
+      "loss": 1.0332,
+      "step": 4564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.265457890177219e-07,
+      "loss": 1.0236,
+      "step": 4565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.25731592400589e-07,
+      "loss": 0.9897,
+      "step": 4566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.249176926662796e-07,
+      "loss": 0.6335,
+      "step": 4567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.241040900205663e-07,
+      "loss": 0.9858,
+      "step": 4568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.232907846691479e-07,
+      "loss": 1.0282,
+      "step": 4569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.224777768176461e-07,
+      "loss": 1.0489,
+      "step": 4570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.216650666716092e-07,
+      "loss": 0.9913,
+      "step": 4571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.20852654436509e-07,
+      "loss": 1.0342,
+      "step": 4572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.200405403177433e-07,
+      "loss": 1.0039,
+      "step": 4573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.192287245206339e-07,
+      "loss": 0.6621,
+      "step": 4574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.184172072504261e-07,
+      "loss": 0.9953,
+      "step": 4575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.176059887122908e-07,
+      "loss": 0.6416,
+      "step": 4576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.167950691113237e-07,
+      "loss": 1.0068,
+      "step": 4577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.15984448652544e-07,
+      "loss": 1.0416,
+      "step": 4578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.151741275408961e-07,
+      "loss": 1.0066,
+      "step": 4579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.143641059812483e-07,
+      "loss": 1.0107,
+      "step": 4580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.135543841783929e-07,
+      "loss": 1.0378,
+      "step": 4581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.127449623370472e-07,
+      "loss": 0.6264,
+      "step": 4582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.11935840661852e-07,
+      "loss": 1.0084,
+      "step": 4583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.11127019357372e-07,
+      "loss": 1.0261,
+      "step": 4584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.103184986280976e-07,
+      "loss": 1.036,
+      "step": 4585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.095102786784402e-07,
+      "loss": 1.0229,
+      "step": 4586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.087023597127378e-07,
+      "loss": 1.0294,
+      "step": 4587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.078947419352515e-07,
+      "loss": 1.0222,
+      "step": 4588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.070874255501659e-07,
+      "loss": 1.0213,
+      "step": 4589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.0628041076159e-07,
+      "loss": 1.0298,
+      "step": 4590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.054736977735566e-07,
+      "loss": 1.0187,
+      "step": 4591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.046672867900206e-07,
+      "loss": 1.0372,
+      "step": 4592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.038611780148623e-07,
+      "loss": 1.027,
+      "step": 4593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.030553716518853e-07,
+      "loss": 1.0175,
+      "step": 4594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.022498679048164e-07,
+      "loss": 1.0341,
+      "step": 4595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.014446669773061e-07,
+      "loss": 1.0182,
+      "step": 4596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.006397690729281e-07,
+      "loss": 0.9791,
+      "step": 4597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.998351743951795e-07,
+      "loss": 1.0482,
+      "step": 4598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.990308831474812e-07,
+      "loss": 0.6388,
+      "step": 4599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.982268955331775e-07,
+      "loss": 1.0421,
+      "step": 4600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.974232117555341e-07,
+      "loss": 1.0336,
+      "step": 4601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.966198320177419e-07,
+      "loss": 1.0012,
+      "step": 4602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.958167565229148e-07,
+      "loss": 1.0408,
+      "step": 4603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.950139854740885e-07,
+      "loss": 1.0067,
+      "step": 4604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.942115190742231e-07,
+      "loss": 1.0401,
+      "step": 4605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.934093575262009e-07,
+      "loss": 1.0401,
+      "step": 4606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.92607501032828e-07,
+      "loss": 1.0274,
+      "step": 4607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.918059497968311e-07,
+      "loss": 0.9887,
+      "step": 4608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.910047040208623e-07,
+      "loss": 1.0331,
+      "step": 4609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.902037639074952e-07,
+      "loss": 0.6961,
+      "step": 4610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.894031296592267e-07,
+      "loss": 1.0204,
+      "step": 4611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.886028014784758e-07,
+      "loss": 0.6587,
+      "step": 4612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.878027795675855e-07,
+      "loss": 1.0208,
+      "step": 4613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.870030641288183e-07,
+      "loss": 0.9854,
+      "step": 4614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.862036553643622e-07,
+      "loss": 1.0231,
+      "step": 4615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.854045534763267e-07,
+      "loss": 1.0408,
+      "step": 4616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.846057586667436e-07,
+      "loss": 1.0247,
+      "step": 4617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.83807271137567e-07,
+      "loss": 1.0302,
+      "step": 4618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.830090910906735e-07,
+      "loss": 0.6423,
+      "step": 4619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.822112187278618e-07,
+      "loss": 1.0747,
+      "step": 4620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.814136542508532e-07,
+      "loss": 1.0312,
+      "step": 4621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.806163978612903e-07,
+      "loss": 1.0246,
+      "step": 4622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.798194497607391e-07,
+      "loss": 1.0721,
+      "step": 4623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.790228101506871e-07,
+      "loss": 1.0565,
+      "step": 4624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.782264792325423e-07,
+      "loss": 0.9833,
+      "step": 4625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.774304572076369e-07,
+      "loss": 1.0627,
+      "step": 4626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.766347442772239e-07,
+      "loss": 0.9963,
+      "step": 4627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.758393406424783e-07,
+      "loss": 0.6181,
+      "step": 4628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.75044246504498e-07,
+      "loss": 0.6739,
+      "step": 4629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.742494620642999e-07,
+      "loss": 1.0414,
+      "step": 4630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.734549875228249e-07,
+      "loss": 1.0471,
+      "step": 4631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.726608230809353e-07,
+      "loss": 1.0287,
+      "step": 4632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.718669689394144e-07,
+      "loss": 1.0078,
+      "step": 4633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.710734252989672e-07,
+      "loss": 1.0386,
+      "step": 4634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.702801923602216e-07,
+      "loss": 1.0297,
+      "step": 4635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.694872703237234e-07,
+      "loss": 0.9963,
+      "step": 4636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.686946593899433e-07,
+      "loss": 1.0102,
+      "step": 4637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.679023597592719e-07,
+      "loss": 0.6432,
+      "step": 4638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.671103716320213e-07,
+      "loss": 1.0254,
+      "step": 4639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.663186952084251e-07,
+      "loss": 1.0344,
+      "step": 4640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.655273306886374e-07,
+      "loss": 1.0015,
+      "step": 4641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.64736278272734e-07,
+      "loss": 1.0577,
+      "step": 4642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.639455381607116e-07,
+      "loss": 1.0138,
+      "step": 4643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.631551105524883e-07,
+      "loss": 1.019,
+      "step": 4644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.623649956479026e-07,
+      "loss": 1.0542,
+      "step": 4645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.61575193646715e-07,
+      "loss": 1.0262,
+      "step": 4646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.607857047486049e-07,
+      "loss": 0.9944,
+      "step": 4647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.599965291531743e-07,
+      "loss": 1.0452,
+      "step": 4648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.592076670599452e-07,
+      "loss": 0.9997,
+      "step": 4649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.584191186683611e-07,
+      "loss": 1.0223,
+      "step": 4650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.57630884177786e-07,
+      "loss": 1.0487,
+      "step": 4651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.568429637875031e-07,
+      "loss": 0.996,
+      "step": 4652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.560553576967177e-07,
+      "loss": 1.0112,
+      "step": 4653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.552680661045554e-07,
+      "loss": 0.9903,
+      "step": 4654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.544810892100624e-07,
+      "loss": 1.02,
+      "step": 4655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.536944272122045e-07,
+      "loss": 0.9992,
+      "step": 4656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.529080803098698e-07,
+      "loss": 1.0074,
+      "step": 4657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.521220487018634e-07,
+      "loss": 0.9942,
+      "step": 4658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.513363325869138e-07,
+      "loss": 1.0176,
+      "step": 4659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.505509321636683e-07,
+      "loss": 0.9909,
+      "step": 4660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.497658476306949e-07,
+      "loss": 0.9923,
+      "step": 4661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.489810791864816e-07,
+      "loss": 1.0205,
+      "step": 4662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.481966270294362e-07,
+      "loss": 1.0252,
+      "step": 4663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.47412491357887e-07,
+      "loss": 0.9792,
+      "step": 4664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.466286723700818e-07,
+      "loss": 1.0415,
+      "step": 4665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.458451702641888e-07,
+      "loss": 1.0268,
+      "step": 4666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.450619852382958e-07,
+      "loss": 1.0152,
+      "step": 4667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.442791174904111e-07,
+      "loss": 1.0399,
+      "step": 4668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.434965672184606e-07,
+      "loss": 1.0048,
+      "step": 4669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.427143346202926e-07,
+      "loss": 0.6404,
+      "step": 4670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.419324198936739e-07,
+      "loss": 1.0084,
+      "step": 4671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.411508232362909e-07,
+      "loss": 1.0009,
+      "step": 4672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.403695448457505e-07,
+      "loss": 1.0037,
+      "step": 4673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.39588584919577e-07,
+      "loss": 1.0437,
+      "step": 4674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.388079436552161e-07,
+      "loss": 1.0126,
+      "step": 4675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.380276212500325e-07,
+      "loss": 1.0342,
+      "step": 4676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.372476179013103e-07,
+      "loss": 1.0335,
+      "step": 4677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.364679338062522e-07,
+      "loss": 0.6717,
+      "step": 4678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.356885691619816e-07,
+      "loss": 1.0144,
+      "step": 4679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.349095241655396e-07,
+      "loss": 0.6512,
+      "step": 4680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.341307990138876e-07,
+      "loss": 1.0678,
+      "step": 4681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.333523939039064e-07,
+      "loss": 1.0377,
+      "step": 4682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.325743090323937e-07,
+      "loss": 1.0365,
+      "step": 4683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.317965445960685e-07,
+      "loss": 1.0038,
+      "step": 4684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.310191007915681e-07,
+      "loss": 1.0124,
+      "step": 4685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.302419778154487e-07,
+      "loss": 1.0262,
+      "step": 4686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.294651758641852e-07,
+      "loss": 0.9999,
+      "step": 4687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.286886951341715e-07,
+      "loss": 1.0455,
+      "step": 4688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.279125358217204e-07,
+      "loss": 1.0041,
+      "step": 4689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.271366981230642e-07,
+      "loss": 1.026,
+      "step": 4690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.263611822343514e-07,
+      "loss": 1.0088,
+      "step": 4691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.255859883516513e-07,
+      "loss": 0.6402,
+      "step": 4692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.248111166709516e-07,
+      "loss": 1.0272,
+      "step": 4693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.240365673881581e-07,
+      "loss": 1.0245,
+      "step": 4694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.232623406990957e-07,
+      "loss": 1.0158,
+      "step": 4695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.224884367995059e-07,
+      "loss": 1.0504,
+      "step": 4696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.217148558850505e-07,
+      "loss": 1.0358,
+      "step": 4697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.209415981513094e-07,
+      "loss": 1.0143,
+      "step": 4698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.2016866379378e-07,
+      "loss": 1.0088,
+      "step": 4699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.193960530078789e-07,
+      "loss": 0.9868,
+      "step": 4700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.186237659889398e-07,
+      "loss": 1.0031,
+      "step": 4701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.178518029322156e-07,
+      "loss": 0.9971,
+      "step": 4702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.170801640328766e-07,
+      "loss": 1.0109,
+      "step": 4703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.163088494860112e-07,
+      "loss": 0.9842,
+      "step": 4704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.155378594866261e-07,
+      "loss": 0.6668,
+      "step": 4705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.147671942296468e-07,
+      "loss": 1.0019,
+      "step": 4706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.139968539099139e-07,
+      "loss": 1.0352,
+      "step": 4707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.132268387221885e-07,
+      "loss": 0.645,
+      "step": 4708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.124571488611486e-07,
+      "loss": 1.0117,
+      "step": 4709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.116877845213902e-07,
+      "loss": 1.0016,
+      "step": 4710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.109187458974266e-07,
+      "loss": 0.9929,
+      "step": 4711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.101500331836896e-07,
+      "loss": 0.6578,
+      "step": 4712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.093816465745266e-07,
+      "loss": 1.0442,
+      "step": 4713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.086135862642046e-07,
+      "loss": 1.0213,
+      "step": 4714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.078458524469077e-07,
+      "loss": 0.6333,
+      "step": 4715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.070784453167369e-07,
+      "loss": 1.037,
+      "step": 4716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.063113650677114e-07,
+      "loss": 0.9993,
+      "step": 4717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.055446118937662e-07,
+      "loss": 1.065,
+      "step": 4718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.047781859887552e-07,
+      "loss": 1.0267,
+      "step": 4719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.040120875464488e-07,
+      "loss": 1.0187,
+      "step": 4720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.032463167605349e-07,
+      "loss": 1.0357,
+      "step": 4721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.024808738246187e-07,
+      "loss": 0.981,
+      "step": 4722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.01715758932222e-07,
+      "loss": 1.0395,
+      "step": 4723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.009509722767841e-07,
+      "loss": 1.0303,
+      "step": 4724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.001865140516609e-07,
+      "loss": 1.0285,
+      "step": 4725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.994223844501259e-07,
+      "loss": 0.9872,
+      "step": 4726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.986585836653689e-07,
+      "loss": 0.9961,
+      "step": 4727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.978951118904975e-07,
+      "loss": 1.02,
+      "step": 4728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.97131969318534e-07,
+      "loss": 1.0352,
+      "step": 4729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.963691561424198e-07,
+      "loss": 0.9914,
+      "step": 4730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.95606672555012e-07,
+      "loss": 1.0381,
+      "step": 4731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.948445187490843e-07,
+      "loss": 1.0375,
+      "step": 4732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.940826949173273e-07,
+      "loss": 0.9944,
+      "step": 4733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.93321201252349e-07,
+      "loss": 0.6552,
+      "step": 4734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.925600379466714e-07,
+      "loss": 1.056,
+      "step": 4735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.917992051927353e-07,
+      "loss": 1.0142,
+      "step": 4736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.91038703182897e-07,
+      "loss": 0.6213,
+      "step": 4737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9027853210943e-07,
+      "loss": 1.0464,
+      "step": 4738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.895186921645235e-07,
+      "loss": 1.0322,
+      "step": 4739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.887591835402822e-07,
+      "loss": 1.0191,
+      "step": 4740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.880000064287281e-07,
+      "loss": 1.055,
+      "step": 4741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.872411610217996e-07,
+      "loss": 1.0439,
+      "step": 4742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.864826475113504e-07,
+      "loss": 1.0439,
+      "step": 4743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.857244660891506e-07,
+      "loss": 1.0363,
+      "step": 4744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.849666169468866e-07,
+      "loss": 1.02,
+      "step": 4745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.842091002761605e-07,
+      "loss": 1.0592,
+      "step": 4746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.834519162684907e-07,
+      "loss": 1.004,
+      "step": 4747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.826950651153108e-07,
+      "loss": 1.0712,
+      "step": 4748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.819385470079706e-07,
+      "loss": 0.9894,
+      "step": 4749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.811823621377371e-07,
+      "loss": 1.0135,
+      "step": 4750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.804265106957896e-07,
+      "loss": 0.9803,
+      "step": 4751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.796709928732265e-07,
+      "loss": 1.0032,
+      "step": 4752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.789158088610603e-07,
+      "loss": 1.0298,
+      "step": 4753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.781609588502193e-07,
+      "loss": 1.0585,
+      "step": 4754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.774064430315474e-07,
+      "loss": 1.0085,
+      "step": 4755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.76652261595805e-07,
+      "loss": 1.0118,
+      "step": 4756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.758984147336655e-07,
+      "loss": 1.0243,
+      "step": 4757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.751449026357201e-07,
+      "loss": 1.0125,
+      "step": 4758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.743917254924741e-07,
+      "loss": 1.0138,
+      "step": 4759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.736388834943485e-07,
+      "loss": 1.018,
+      "step": 4760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.728863768316799e-07,
+      "loss": 1.0259,
+      "step": 4761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.721342056947195e-07,
+      "loss": 1.0428,
+      "step": 4762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.713823702736341e-07,
+      "loss": 1.0539,
+      "step": 4763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.706308707585056e-07,
+      "loss": 1.0477,
+      "step": 4764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.698797073393311e-07,
+      "loss": 1.0661,
+      "step": 4765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.691288802060215e-07,
+      "loss": 1.0485,
+      "step": 4766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.683783895484041e-07,
+      "loss": 1.024,
+      "step": 4767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.67628235556221e-07,
+      "loss": 0.6354,
+      "step": 4768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.668784184191285e-07,
+      "loss": 1.0356,
+      "step": 4769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.661289383266984e-07,
+      "loss": 0.6361,
+      "step": 4770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.653797954684168e-07,
+      "loss": 0.9864,
+      "step": 4771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.646309900336855e-07,
+      "loss": 0.975,
+      "step": 4772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.638825222118187e-07,
+      "loss": 1.0344,
+      "step": 4773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.631343921920475e-07,
+      "loss": 1.0062,
+      "step": 4774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.623866001635173e-07,
+      "loss": 1.0424,
+      "step": 4775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.61639146315287e-07,
+      "loss": 1.0048,
+      "step": 4776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.608920308363311e-07,
+      "loss": 1.0375,
+      "step": 4777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.601452539155381e-07,
+      "loss": 1.0382,
+      "step": 4778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.593988157417103e-07,
+      "loss": 1.0366,
+      "step": 4779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.586527165035649e-07,
+      "loss": 1.0331,
+      "step": 4780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.57906956389734e-07,
+      "loss": 1.0502,
+      "step": 4781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.57161535588763e-07,
+      "loss": 1.0212,
+      "step": 4782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.56416454289112e-07,
+      "loss": 1.0384,
+      "step": 4783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.556717126791554e-07,
+      "loss": 1.0436,
+      "step": 4784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.549273109471811e-07,
+      "loss": 0.9899,
+      "step": 4785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.541832492813917e-07,
+      "loss": 0.9772,
+      "step": 4786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.534395278699037e-07,
+      "loss": 1.0493,
+      "step": 4787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.52696146900747e-07,
+      "loss": 0.6336,
+      "step": 4788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.519531065618671e-07,
+      "loss": 1.0171,
+      "step": 4789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.512104070411207e-07,
+      "loss": 1.0238,
+      "step": 4790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.504680485262801e-07,
+      "loss": 0.9915,
+      "step": 4791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.497260312050316e-07,
+      "loss": 1.0132,
+      "step": 4792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.489843552649743e-07,
+      "loss": 1.0101,
+      "step": 4793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.482430208936225e-07,
+      "loss": 1.043,
+      "step": 4794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.475020282784014e-07,
+      "loss": 1.0131,
+      "step": 4795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.467613776066521e-07,
+      "loss": 1.0559,
+      "step": 4796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.460210690656289e-07,
+      "loss": 1.0074,
+      "step": 4797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.452811028424989e-07,
+      "loss": 1.0187,
+      "step": 4798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.445414791243435e-07,
+      "loss": 1.0116,
+      "step": 4799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.438021980981574e-07,
+      "loss": 0.6643,
+      "step": 4800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.430632599508472e-07,
+      "loss": 0.6417,
+      "step": 4801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.423246648692346e-07,
+      "loss": 1.0571,
+      "step": 4802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.415864130400537e-07,
+      "loss": 0.6343,
+      "step": 4803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.408485046499524e-07,
+      "loss": 0.9881,
+      "step": 4804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.401109398854907e-07,
+      "loss": 1.0306,
+      "step": 4805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.393737189331432e-07,
+      "loss": 0.9897,
+      "step": 4806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.386368419792962e-07,
+      "loss": 0.993,
+      "step": 4807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.379003092102502e-07,
+      "loss": 0.6081,
+      "step": 4808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.371641208122175e-07,
+      "loss": 1.0154,
+      "step": 4809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.364282769713243e-07,
+      "loss": 1.0131,
+      "step": 4810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.356927778736098e-07,
+      "loss": 1.0477,
+      "step": 4811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.349576237050245e-07,
+      "loss": 1.025,
+      "step": 4812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.342228146514332e-07,
+      "loss": 0.9857,
+      "step": 4813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.334883508986131e-07,
+      "loss": 0.6562,
+      "step": 4814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.327542326322539e-07,
+      "loss": 1.0081,
+      "step": 4815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.320204600379587e-07,
+      "loss": 1.0082,
+      "step": 4816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.312870333012414e-07,
+      "loss": 1.0441,
+      "step": 4817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.305539526075303e-07,
+      "loss": 1.0539,
+      "step": 4818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.298212181421654e-07,
+      "loss": 1.0159,
+      "step": 4819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.290888300903991e-07,
+      "loss": 0.9875,
+      "step": 4820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.283567886373971e-07,
+      "loss": 1.017,
+      "step": 4821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.276250939682368e-07,
+      "loss": 0.9981,
+      "step": 4822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.268937462679067e-07,
+      "loss": 0.9811,
+      "step": 4823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.261627457213099e-07,
+      "loss": 1.0026,
+      "step": 4824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.254320925132602e-07,
+      "loss": 1.0524,
+      "step": 4825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.247017868284838e-07,
+      "loss": 1.0658,
+      "step": 4826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.239718288516198e-07,
+      "loss": 1.0236,
+      "step": 4827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.232422187672186e-07,
+      "loss": 1.0333,
+      "step": 4828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.225129567597429e-07,
+      "loss": 1.0112,
+      "step": 4829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.217840430135672e-07,
+      "loss": 1.0246,
+      "step": 4830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.210554777129784e-07,
+      "loss": 1.0298,
+      "step": 4831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.203272610421748e-07,
+      "loss": 0.9989,
+      "step": 4832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.195993931852679e-07,
+      "loss": 0.6521,
+      "step": 4833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.18871874326278e-07,
+      "loss": 0.6554,
+      "step": 4834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.181447046491401e-07,
+      "loss": 1.0165,
+      "step": 4835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.174178843376998e-07,
+      "loss": 1.002,
+      "step": 4836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.166914135757145e-07,
+      "loss": 0.9905,
+      "step": 4837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.159652925468535e-07,
+      "loss": 1.0255,
+      "step": 4838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.152395214346966e-07,
+      "loss": 1.0451,
+      "step": 4839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.145141004227364e-07,
+      "loss": 1.0187,
+      "step": 4840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.137890296943762e-07,
+      "loss": 1.017,
+      "step": 4841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.130643094329313e-07,
+      "loss": 1.0404,
+      "step": 4842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.123399398216282e-07,
+      "loss": 0.9961,
+      "step": 4843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.116159210436045e-07,
+      "loss": 1.0304,
+      "step": 4844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.108922532819092e-07,
+      "loss": 1.015,
+      "step": 4845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.101689367195027e-07,
+      "loss": 1.0139,
+      "step": 4846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.094459715392564e-07,
+      "loss": 1.0098,
+      "step": 4847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.087233579239539e-07,
+      "loss": 1.0039,
+      "step": 4848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.080010960562878e-07,
+      "loss": 1.0423,
+      "step": 4849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.072791861188631e-07,
+      "loss": 1.0144,
+      "step": 4850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.065576282941962e-07,
+      "loss": 1.0299,
+      "step": 4851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.058364227647139e-07,
+      "loss": 0.6624,
+      "step": 4852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.05115569712754e-07,
+      "loss": 1.0324,
+      "step": 4853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.043950693205648e-07,
+      "loss": 0.9746,
+      "step": 4854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.03674921770307e-07,
+      "loss": 1.0094,
+      "step": 4855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.029551272440494e-07,
+      "loss": 0.9737,
+      "step": 4856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.022356859237738e-07,
+      "loss": 1.002,
+      "step": 4857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.015165979913719e-07,
+      "loss": 0.6504,
+      "step": 4858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.007978636286464e-07,
+      "loss": 1.0207,
+      "step": 4859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.000794830173106e-07,
+      "loss": 1.0539,
+      "step": 4860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.993614563389868e-07,
+      "loss": 1.0111,
+      "step": 4861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.986437837752102e-07,
+      "loss": 0.9927,
+      "step": 4862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.979264655074252e-07,
+      "loss": 0.6137,
+      "step": 4863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.972095017169867e-07,
+      "loss": 1.0241,
+      "step": 4864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.964928925851602e-07,
+      "loss": 0.9957,
+      "step": 4865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.957766382931215e-07,
+      "loss": 1.0087,
+      "step": 4866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.950607390219564e-07,
+      "loss": 1.0037,
+      "step": 4867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.943451949526615e-07,
+      "loss": 0.9928,
+      "step": 4868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.93630006266143e-07,
+      "loss": 1.0068,
+      "step": 4869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.929151731432173e-07,
+      "loss": 1.037,
+      "step": 4870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.922006957646124e-07,
+      "loss": 1.0333,
+      "step": 4871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.914865743109633e-07,
+      "loss": 0.6612,
+      "step": 4872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.907728089628176e-07,
+      "loss": 1.0376,
+      "step": 4873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.90059399900632e-07,
+      "loss": 0.9837,
+      "step": 4874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.893463473047733e-07,
+      "loss": 1.0637,
+      "step": 4875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.886336513555179e-07,
+      "loss": 1.0262,
+      "step": 4876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.879213122330529e-07,
+      "loss": 1.0321,
+      "step": 4877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.872093301174732e-07,
+      "loss": 1.0796,
+      "step": 4878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.864977051887854e-07,
+      "loss": 0.9896,
+      "step": 4879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.857864376269051e-07,
+      "loss": 1.0341,
+      "step": 4880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.850755276116575e-07,
+      "loss": 1.0081,
+      "step": 4881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.84364975322778e-07,
+      "loss": 1.0186,
+      "step": 4882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.836547809399099e-07,
+      "loss": 1.0343,
+      "step": 4883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.829449446426078e-07,
+      "loss": 1.0454,
+      "step": 4884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.822354666103351e-07,
+      "loss": 1.006,
+      "step": 4885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.815263470224645e-07,
+      "loss": 1.0303,
+      "step": 4886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.808175860582783e-07,
+      "loss": 1.0321,
+      "step": 4887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.801091838969681e-07,
+      "loss": 1.0352,
+      "step": 4888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.794011407176345e-07,
+      "loss": 1.0418,
+      "step": 4889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.786934566992878e-07,
+      "loss": 0.9602,
+      "step": 4890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.77986132020847e-07,
+      "loss": 1.0066,
+      "step": 4891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.772791668611408e-07,
+      "loss": 0.9944,
+      "step": 4892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.76572561398907e-07,
+      "loss": 1.0348,
+      "step": 4893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.758663158127912e-07,
+      "loss": 0.9809,
+      "step": 4894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.751604302813496e-07,
+      "loss": 1.0513,
+      "step": 4895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.744549049830465e-07,
+      "loss": 0.9903,
+      "step": 4896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.737497400962557e-07,
+      "loss": 0.9616,
+      "step": 4897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.730449357992592e-07,
+      "loss": 0.9817,
+      "step": 4898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.723404922702493e-07,
+      "loss": 0.9689,
+      "step": 4899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.716364096873246e-07,
+      "loss": 1.002,
+      "step": 4900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.709326882284939e-07,
+      "loss": 1.0361,
+      "step": 4901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.702293280716752e-07,
+      "loss": 0.9959,
+      "step": 4902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.695263293946944e-07,
+      "loss": 1.0371,
+      "step": 4903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.68823692375287e-07,
+      "loss": 1.0247,
+      "step": 4904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.681214171910946e-07,
+      "loss": 1.0318,
+      "step": 4905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.674195040196701e-07,
+      "loss": 0.9657,
+      "step": 4906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.667179530384733e-07,
+      "loss": 0.9958,
+      "step": 4907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.660167644248732e-07,
+      "loss": 1.0463,
+      "step": 4908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.653159383561466e-07,
+      "loss": 0.6436,
+      "step": 4909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.646154750094791e-07,
+      "loss": 1.0138,
+      "step": 4910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.639153745619645e-07,
+      "loss": 0.6558,
+      "step": 4911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.632156371906043e-07,
+      "loss": 0.64,
+      "step": 4912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.62516263072309e-07,
+      "loss": 1.0444,
+      "step": 4913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.618172523838971e-07,
+      "loss": 0.9937,
+      "step": 4914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.611186053020953e-07,
+      "loss": 1.0044,
+      "step": 4915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.604203220035369e-07,
+      "loss": 1.036,
+      "step": 4916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.597224026647651e-07,
+      "loss": 1.0228,
+      "step": 4917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.590248474622305e-07,
+      "loss": 1.0058,
+      "step": 4918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.583276565722914e-07,
+      "loss": 1.0059,
+      "step": 4919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.576308301712145e-07,
+      "loss": 1.0217,
+      "step": 4920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.569343684351741e-07,
+      "loss": 1.0284,
+      "step": 4921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.562382715402514e-07,
+      "loss": 0.9985,
+      "step": 4922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.555425396624365e-07,
+      "loss": 1.0292,
+      "step": 4923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.54847172977627e-07,
+      "loss": 1.018,
+      "step": 4924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.541521716616278e-07,
+      "loss": 0.6503,
+      "step": 4925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.534575358901521e-07,
+      "loss": 1.0341,
+      "step": 4926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.527632658388198e-07,
+      "loss": 1.0005,
+      "step": 4927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.520693616831591e-07,
+      "loss": 1.0378,
+      "step": 4928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.513758235986053e-07,
+      "loss": 1.0692,
+      "step": 4929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.506826517605013e-07,
+      "loss": 0.9997,
+      "step": 4930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.499898463440978e-07,
+      "loss": 1.0445,
+      "step": 4931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.492974075245511e-07,
+      "loss": 1.0185,
+      "step": 4932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.486053354769271e-07,
+      "loss": 1.0255,
+      "step": 4933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.479136303761975e-07,
+      "loss": 1.0222,
+      "step": 4934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47222292397242e-07,
+      "loss": 0.64,
+      "step": 4935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465313217148473e-07,
+      "loss": 0.9867,
+      "step": 4936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.458407185037072e-07,
+      "loss": 1.0313,
+      "step": 4937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.451504829384219e-07,
+      "loss": 1.0019,
+      "step": 4938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.444606151934998e-07,
+      "loss": 0.6487,
+      "step": 4939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.437711154433555e-07,
+      "loss": 1.0417,
+      "step": 4940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.430819838623109e-07,
+      "loss": 1.0324,
+      "step": 4941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.42393220624595e-07,
+      "loss": 0.9675,
+      "step": 4942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.417048259043437e-07,
+      "loss": 1.0169,
+      "step": 4943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.410167998755985e-07,
+      "loss": 0.9884,
+      "step": 4944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.403291427123092e-07,
+      "loss": 0.9808,
+      "step": 4945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.396418545883317e-07,
+      "loss": 0.9828,
+      "step": 4946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.389549356774287e-07,
+      "loss": 0.9718,
+      "step": 4947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3826838615327e-07,
+      "loss": 1.0604,
+      "step": 4948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.375822061894309e-07,
+      "loss": 1.0281,
+      "step": 4949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.368963959593941e-07,
+      "loss": 1.0211,
+      "step": 4950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.36210955636549e-07,
+      "loss": 1.0488,
+      "step": 4951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.355258853941906e-07,
+      "loss": 0.639,
+      "step": 4952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.348411854055212e-07,
+      "loss": 0.6089,
+      "step": 4953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3415685584365e-07,
+      "loss": 1.0647,
+      "step": 4954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.334728968815898e-07,
+      "loss": 1.0456,
+      "step": 4955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.327893086922625e-07,
+      "loss": 1.0824,
+      "step": 4956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.321060914484958e-07,
+      "loss": 0.9976,
+      "step": 4957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.314232453230225e-07,
+      "loss": 1.0518,
+      "step": 4958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.307407704884832e-07,
+      "loss": 1.0448,
+      "step": 4959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.300586671174225e-07,
+      "loss": 1.0071,
+      "step": 4960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.293769353822928e-07,
+      "loss": 1.0212,
+      "step": 4961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.286955754554521e-07,
+      "loss": 1.0215,
+      "step": 4962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.280145875091639e-07,
+      "loss": 0.9916,
+      "step": 4963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.273339717155984e-07,
+      "loss": 0.9906,
+      "step": 4964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.266537282468322e-07,
+      "loss": 1.0144,
+      "step": 4965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.259738572748453e-07,
+      "loss": 1.0006,
+      "step": 4966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.25294358971526e-07,
+      "loss": 1.0594,
+      "step": 4967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.246152335086673e-07,
+      "loss": 1.0607,
+      "step": 4968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.239364810579685e-07,
+      "loss": 1.02,
+      "step": 4969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.23258101791034e-07,
+      "loss": 1.0127,
+      "step": 4970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.225800958793742e-07,
+      "loss": 1.0056,
+      "step": 4971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.219024634944052e-07,
+      "loss": 1.0353,
+      "step": 4972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.212252048074484e-07,
+      "loss": 1.052,
+      "step": 4973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.205483199897307e-07,
+      "loss": 0.6425,
+      "step": 4974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.198718092123846e-07,
+      "loss": 1.0037,
+      "step": 4975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.191956726464486e-07,
+      "loss": 0.9943,
+      "step": 4976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.185199104628651e-07,
+      "loss": 1.047,
+      "step": 4977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.178445228324828e-07,
+      "loss": 0.6674,
+      "step": 4978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.171695099260563e-07,
+      "loss": 1.0406,
+      "step": 4979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.164948719142446e-07,
+      "loss": 1.0116,
+      "step": 4980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.158206089676124e-07,
+      "loss": 0.993,
+      "step": 4981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.151467212566288e-07,
+      "loss": 1.0215,
+      "step": 4982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.144732089516683e-07,
+      "loss": 1.0107,
+      "step": 4983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.138000722230116e-07,
+      "loss": 0.9869,
+      "step": 4984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.13127311240843e-07,
+      "loss": 1.0088,
+      "step": 4985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.124549261752524e-07,
+      "loss": 1.0218,
+      "step": 4986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.117829171962356e-07,
+      "loss": 1.0322,
+      "step": 4987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.111112844736909e-07,
+      "loss": 1.0414,
+      "step": 4988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.104400281774237e-07,
+      "loss": 0.9926,
+      "step": 4989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.097691484771434e-07,
+      "loss": 0.9562,
+      "step": 4990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.090986455424639e-07,
+      "loss": 0.9807,
+      "step": 4991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.08428519542905e-07,
+      "loss": 0.9808,
+      "step": 4992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.077587706478897e-07,
+      "loss": 1.0267,
+      "step": 4993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.070893990267464e-07,
+      "loss": 1.036,
+      "step": 4994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.064204048487084e-07,
+      "loss": 0.6074,
+      "step": 4995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.057517882829133e-07,
+      "loss": 1.0287,
+      "step": 4996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.050835494984027e-07,
+      "loss": 1.0168,
+      "step": 4997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.044156886641242e-07,
+      "loss": 1.0335,
+      "step": 4998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.037482059489273e-07,
+      "loss": 1.012,
+      "step": 4999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.030811015215683e-07,
+      "loss": 0.9928,
+      "step": 5000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.024143755507068e-07,
+      "loss": 1.0078,
+      "step": 5001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.017480282049072e-07,
+      "loss": 1.0126,
+      "step": 5002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.010820596526379e-07,
+      "loss": 1.0533,
+      "step": 5003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.004164700622708e-07,
+      "loss": 1.044,
+      "step": 5004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.997512596020832e-07,
+      "loss": 1.0032,
+      "step": 5005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.990864284402559e-07,
+      "loss": 1.0263,
+      "step": 5006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.984219767448736e-07,
+      "loss": 1.0263,
+      "step": 5007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.977579046839262e-07,
+      "loss": 1.0363,
+      "step": 5008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.970942124253064e-07,
+      "loss": 1.0043,
+      "step": 5009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.96430900136811e-07,
+      "loss": 1.0046,
+      "step": 5010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.957679679861415e-07,
+      "loss": 0.983,
+      "step": 5011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.951054161409028e-07,
+      "loss": 0.9847,
+      "step": 5012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.944432447686033e-07,
+      "loss": 1.0229,
+      "step": 5013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.937814540366565e-07,
+      "loss": 1.0326,
+      "step": 5014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.931200441123775e-07,
+      "loss": 1.0113,
+      "step": 5015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.924590151629871e-07,
+      "loss": 0.6285,
+      "step": 5016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.917983673556086e-07,
+      "loss": 1.0028,
+      "step": 5017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.911381008572697e-07,
+      "loss": 1.0341,
+      "step": 5018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.904782158349015e-07,
+      "loss": 1.0573,
+      "step": 5019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.89818712455339e-07,
+      "loss": 1.0078,
+      "step": 5020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.89159590885319e-07,
+      "loss": 1.0233,
+      "step": 5021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.885008512914837e-07,
+      "loss": 1.0143,
+      "step": 5022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.878424938403778e-07,
+      "loss": 1.0441,
+      "step": 5023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.871845186984501e-07,
+      "loss": 1.0214,
+      "step": 5024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.865269260320525e-07,
+      "loss": 0.9851,
+      "step": 5025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.85869716007439e-07,
+      "loss": 0.667,
+      "step": 5026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.852128887907682e-07,
+      "loss": 1.0239,
+      "step": 5027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.845564445481018e-07,
+      "loss": 0.6291,
+      "step": 5028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.839003834454045e-07,
+      "loss": 1.0044,
+      "step": 5029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.832447056485435e-07,
+      "loss": 0.9998,
+      "step": 5030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.825894113232904e-07,
+      "loss": 1.0106,
+      "step": 5031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.819345006353186e-07,
+      "loss": 1.0259,
+      "step": 5032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.812799737502052e-07,
+      "loss": 0.9724,
+      "step": 5033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.806258308334299e-07,
+      "loss": 1.021,
+      "step": 5034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.799720720503757e-07,
+      "loss": 1.0536,
+      "step": 5035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793186975663288e-07,
+      "loss": 0.9906,
+      "step": 5036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.786657075464764e-07,
+      "loss": 0.9999,
+      "step": 5037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.780131021559106e-07,
+      "loss": 0.9861,
+      "step": 5038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.773608815596255e-07,
+      "loss": 1.0172,
+      "step": 5039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7670904592251784e-07,
+      "loss": 1.0282,
+      "step": 5040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7605759540938706e-07,
+      "loss": 1.0173,
+      "step": 5041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7540653018493573e-07,
+      "loss": 1.0552,
+      "step": 5042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.747558504137674e-07,
+      "loss": 1.0258,
+      "step": 5043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.741055562603902e-07,
+      "loss": 1.0379,
+      "step": 5044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7345564788921354e-07,
+      "loss": 1.027,
+      "step": 5045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.728061254645497e-07,
+      "loss": 1.0458,
+      "step": 5046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7215698915061384e-07,
+      "loss": 0.9888,
+      "step": 5047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7150823911152193e-07,
+      "loss": 0.9897,
+      "step": 5048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.708598755112938e-07,
+      "loss": 1.0454,
+      "step": 5049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7021189851385124e-07,
+      "loss": 1.013,
+      "step": 5050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6956430828301805e-07,
+      "loss": 0.9816,
+      "step": 5051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6891710498252023e-07,
+      "loss": 1.0287,
+      "step": 5052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.682702887759864e-07,
+      "loss": 1.0019,
+      "step": 5053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6762385982694686e-07,
+      "loss": 1.0097,
+      "step": 5054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.66977818298834e-07,
+      "loss": 0.9873,
+      "step": 5055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6633216435498244e-07,
+      "loss": 0.6133,
+      "step": 5056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6568689815862884e-07,
+      "loss": 1.0313,
+      "step": 5057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6504201987291214e-07,
+      "loss": 1.0188,
+      "step": 5058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.643975296608722e-07,
+      "loss": 1.0185,
+      "step": 5059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6375342768545135e-07,
+      "loss": 1.0251,
+      "step": 5060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6310971410949393e-07,
+      "loss": 0.6187,
+      "step": 5061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.624663890957463e-07,
+      "loss": 1.0469,
+      "step": 5062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.618234528068561e-07,
+      "loss": 0.6131,
+      "step": 5063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6118090540537303e-07,
+      "loss": 0.9928,
+      "step": 5064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.605387470537476e-07,
+      "loss": 0.99,
+      "step": 5065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.598969779143329e-07,
+      "loss": 1.0217,
+      "step": 5066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5925559814938354e-07,
+      "loss": 1.0312,
+      "step": 5067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5861460792105535e-07,
+      "loss": 0.6729,
+      "step": 5068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5797400739140577e-07,
+      "loss": 1.0471,
+      "step": 5069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.573337967223947e-07,
+      "loss": 0.6642,
+      "step": 5070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.566939760758808e-07,
+      "loss": 1.0342,
+      "step": 5071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5605454561362686e-07,
+      "loss": 1.0339,
+      "step": 5072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5541550549729615e-07,
+      "loss": 0.9835,
+      "step": 5073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.547768558884526e-07,
+      "loss": 1.0449,
+      "step": 5074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.541385969485625e-07,
+      "loss": 0.9946,
+      "step": 5075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5350072883899225e-07,
+      "loss": 1.0178,
+      "step": 5076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5286325172101047e-07,
+      "loss": 1.022,
+      "step": 5077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.522261657557862e-07,
+      "loss": 0.9582,
+      "step": 5078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5158947110438995e-07,
+      "loss": 0.6328,
+      "step": 5079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5095316792779314e-07,
+      "loss": 1.0512,
+      "step": 5080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5031725638686866e-07,
+      "loss": 1.0526,
+      "step": 5081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.496817366423893e-07,
+      "loss": 1.0342,
+      "step": 5082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4904660885502957e-07,
+      "loss": 1.0377,
+      "step": 5083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.48411873185365e-07,
+      "loss": 1.0447,
+      "step": 5084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.477775297938717e-07,
+      "loss": 0.9846,
+      "step": 5085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4714357884092724e-07,
+      "loss": 0.9968,
+      "step": 5086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4651002048680864e-07,
+      "loss": 0.6656,
+      "step": 5087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4587685489169444e-07,
+      "loss": 0.6527,
+      "step": 5088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4524408221566424e-07,
+      "loss": 1.0251,
+      "step": 5089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4461170261869795e-07,
+      "loss": 1.0408,
+      "step": 5090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.439797162606758e-07,
+      "loss": 1.0307,
+      "step": 5091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4334812330137917e-07,
+      "loss": 0.9955,
+      "step": 5092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4271692390048956e-07,
+      "loss": 1.0095,
+      "step": 5093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.420861182175892e-07,
+      "loss": 0.9539,
+      "step": 5094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4145570641216043e-07,
+      "loss": 1.0317,
+      "step": 5095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.408256886435866e-07,
+      "loss": 1.0509,
+      "step": 5096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4019606507115157e-07,
+      "loss": 1.0307,
+      "step": 5097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.395668358540379e-07,
+      "loss": 1.0559,
+      "step": 5098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3893800115132994e-07,
+      "loss": 1.006,
+      "step": 5099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3830956112201246e-07,
+      "loss": 1.0037,
+      "step": 5100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.376815159249696e-07,
+      "loss": 1.004,
+      "step": 5101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.370538657189862e-07,
+      "loss": 0.6444,
+      "step": 5102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3642661066274747e-07,
+      "loss": 1.021,
+      "step": 5103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.357997509148375e-07,
+      "loss": 0.973,
+      "step": 5104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3517328663374143e-07,
+      "loss": 1.0039,
+      "step": 5105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3454721797784443e-07,
+      "loss": 1.0042,
+      "step": 5106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.339215451054317e-07,
+      "loss": 0.6441,
+      "step": 5107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3329626817468833e-07,
+      "loss": 1.0095,
+      "step": 5108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3267138734369825e-07,
+      "loss": 1.0211,
+      "step": 5109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.320469027704463e-07,
+      "loss": 0.9905,
+      "step": 5110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3142281461281735e-07,
+      "loss": 0.6439,
+      "step": 5111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.307991230285957e-07,
+      "loss": 0.6504,
+      "step": 5112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3017582817546485e-07,
+      "loss": 1.0268,
+      "step": 5113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.29552930211009e-07,
+      "loss": 1.0022,
+      "step": 5114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2893042929271117e-07,
+      "loss": 0.9819,
+      "step": 5115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2830832557795426e-07,
+      "loss": 0.9965,
+      "step": 5116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2768661922402116e-07,
+      "loss": 1.049,
+      "step": 5117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2706531038809346e-07,
+      "loss": 0.9794,
+      "step": 5118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.264443992272537e-07,
+      "loss": 1.0395,
+      "step": 5119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.258238858984817e-07,
+      "loss": 1.0316,
+      "step": 5120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.252037705586582e-07,
+      "loss": 1.0228,
+      "step": 5121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.245840533645635e-07,
+      "loss": 0.9763,
+      "step": 5122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2396473447287627e-07,
+      "loss": 1.0341,
+      "step": 5123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.233458140401751e-07,
+      "loss": 1.0024,
+      "step": 5124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2272729222293833e-07,
+      "loss": 1.0133,
+      "step": 5125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2210916917754204e-07,
+      "loss": 1.0275,
+      "step": 5126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.214914450602625e-07,
+      "loss": 1.0288,
+      "step": 5127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.208741200272752e-07,
+      "loss": 1.026,
+      "step": 5128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2025719423465467e-07,
+      "loss": 1.0577,
+      "step": 5129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1964066783837437e-07,
+      "loss": 1.0253,
+      "step": 5130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1902454099430604e-07,
+      "loss": 0.622,
+      "step": 5131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.184088138582216e-07,
+      "loss": 0.9766,
+      "step": 5132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.177934865857915e-07,
+      "loss": 0.9689,
+      "step": 5133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1717855933258494e-07,
+      "loss": 1.0156,
+      "step": 5134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1656403225407e-07,
+      "loss": 1.0106,
+      "step": 5135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.159499055056137e-07,
+      "loss": 1.0389,
+      "step": 5136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.153361792424819e-07,
+      "loss": 1.04,
+      "step": 5137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1472285361983904e-07,
+      "loss": 1.0297,
+      "step": 5138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.141099287927483e-07,
+      "loss": 0.6351,
+      "step": 5139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.134974049161717e-07,
+      "loss": 1.0083,
+      "step": 5140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.128852821449702e-07,
+      "loss": 1.0232,
+      "step": 5141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1227356063390206e-07,
+      "loss": 0.6464,
+      "step": 5142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.116622405376251e-07,
+      "loss": 0.6716,
+      "step": 5143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.110513220106957e-07,
+      "loss": 1.0189,
+      "step": 5144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.104408052075683e-07,
+      "loss": 0.999,
+      "step": 5145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0983069028259653e-07,
+      "loss": 1.0107,
+      "step": 5146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.092209773900317e-07,
+      "loss": 0.9637,
+      "step": 5147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0861166668402316e-07,
+      "loss": 0.6454,
+      "step": 5148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0800275831861944e-07,
+      "loss": 0.6177,
+      "step": 5149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.073942524477667e-07,
+      "loss": 1.0355,
+      "step": 5150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0678614922531e-07,
+      "loss": 0.629,
+      "step": 5151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.06178448804992e-07,
+      "loss": 0.6821,
+      "step": 5152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0557115134045403e-07,
+      "loss": 0.9943,
+      "step": 5153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049642569852347e-07,
+      "loss": 1.0583,
+      "step": 5154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.043577658927717e-07,
+      "loss": 1.0226,
+      "step": 5155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.037516782164001e-07,
+      "loss": 1.0136,
+      "step": 5156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0314599410935315e-07,
+      "loss": 1.0343,
+      "step": 5157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.025407137247623e-07,
+      "loss": 0.6422,
+      "step": 5158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.019358372156565e-07,
+      "loss": 0.9921,
+      "step": 5159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013313647349632e-07,
+      "loss": 1.02,
+      "step": 5160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.007272964355073e-07,
+      "loss": 1.015,
+      "step": 5161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.00123632470011e-07,
+      "loss": 1.0051,
+      "step": 5162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.995203729910961e-07,
+      "loss": 0.9959,
+      "step": 5163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9891751815127937e-07,
+      "loss": 1.0418,
+      "step": 5164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9831506810297743e-07,
+      "loss": 0.6436,
+      "step": 5165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9771302299850376e-07,
+      "loss": 0.9852,
+      "step": 5166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9711138299006984e-07,
+      "loss": 0.6599,
+      "step": 5167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9651014822978437e-07,
+      "loss": 0.9829,
+      "step": 5168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9590931886965404e-07,
+      "loss": 1.0086,
+      "step": 5169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953088950615819e-07,
+      "loss": 1.0123,
+      "step": 5170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9470887695736986e-07,
+      "loss": 1.0054,
+      "step": 5171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9410926470871655e-07,
+      "loss": 1.022,
+      "step": 5172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9351005846721817e-07,
+      "loss": 1.0236,
+      "step": 5173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9291125838436836e-07,
+      "loss": 1.0632,
+      "step": 5174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9231286461155764e-07,
+      "loss": 1.0043,
+      "step": 5175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9171487730007446e-07,
+      "loss": 0.9952,
+      "step": 5176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.911172966011038e-07,
+      "loss": 0.9829,
+      "step": 5177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.905201226657284e-07,
+      "loss": 0.9976,
+      "step": 5178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8992335564492864e-07,
+      "loss": 1.0174,
+      "step": 5179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8932699568957996e-07,
+      "loss": 1.0008,
+      "step": 5180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8873104295045666e-07,
+      "loss": 0.9805,
+      "step": 5181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.881354975782303e-07,
+      "loss": 1.0157,
+      "step": 5182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8754035972346833e-07,
+      "loss": 1.0319,
+      "step": 5183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.869456295366358e-07,
+      "loss": 0.9905,
+      "step": 5184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8635130716809527e-07,
+      "loss": 1.0323,
+      "step": 5185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8575739276810414e-07,
+      "loss": 0.9965,
+      "step": 5186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8516388648681874e-07,
+      "loss": 0.9938,
+      "step": 5187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8457078847429124e-07,
+      "loss": 1.0117,
+      "step": 5188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.839780988804713e-07,
+      "loss": 1.0546,
+      "step": 5189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.833858178552045e-07,
+      "loss": 0.9958,
+      "step": 5190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.827939455482341e-07,
+      "loss": 1.0394,
+      "step": 5191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8220248210919824e-07,
+      "loss": 1.0525,
+      "step": 5192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8161142768763387e-07,
+      "loss": 1.031,
+      "step": 5193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.81020782432973e-07,
+      "loss": 1.0331,
+      "step": 5194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8043054649454473e-07,
+      "loss": 0.9615,
+      "step": 5195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.798407200215752e-07,
+      "loss": 1.0129,
+      "step": 5196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.792513031631859e-07,
+      "loss": 1.0285,
+      "step": 5197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7866229606839563e-07,
+      "loss": 1.0329,
+      "step": 5198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.780736988861193e-07,
+      "loss": 0.6459,
+      "step": 5199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7748551176516806e-07,
+      "loss": 0.9799,
+      "step": 5200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.768977348542499e-07,
+      "loss": 0.9834,
+      "step": 5201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.763103683019686e-07,
+      "loss": 0.9971,
+      "step": 5202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75723412256824e-07,
+      "loss": 1.0017,
+      "step": 5203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7513686686721237e-07,
+      "loss": 1.017,
+      "step": 5204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.745507322814265e-07,
+      "loss": 1.0156,
+      "step": 5205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7396500864765537e-07,
+      "loss": 0.9889,
+      "step": 5206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.733796961139839e-07,
+      "loss": 1.0153,
+      "step": 5207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7279479482839183e-07,
+      "loss": 0.9901,
+      "step": 5208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7221030493875683e-07,
+      "loss": 1.0237,
+      "step": 5209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7162622659285183e-07,
+      "loss": 0.9616,
+      "step": 5210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.710425599383453e-07,
+      "loss": 1.0297,
+      "step": 5211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.704593051228022e-07,
+      "loss": 0.9996,
+      "step": 5212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6987646229368364e-07,
+      "loss": 1.0416,
+      "step": 5213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6929403159834503e-07,
+      "loss": 1.0156,
+      "step": 5214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.687120131840391e-07,
+      "loss": 0.9904,
+      "step": 5215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.681304071979139e-07,
+      "loss": 0.6418,
+      "step": 5216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.675492137870133e-07,
+      "loss": 1.0439,
+      "step": 5217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669684330982765e-07,
+      "loss": 1.0401,
+      "step": 5218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663880652785387e-07,
+      "loss": 1.0126,
+      "step": 5219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6580811047453077e-07,
+      "loss": 1.0273,
+      "step": 5220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6522856883287865e-07,
+      "loss": 0.9966,
+      "step": 5221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6464944050010435e-07,
+      "loss": 1.0023,
+      "step": 5222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.640707256226254e-07,
+      "loss": 1.0014,
+      "step": 5223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.63492424346755e-07,
+      "loss": 1.0218,
+      "step": 5224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.629145368187001e-07,
+      "loss": 1.025,
+      "step": 5225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.62337063184565e-07,
+      "loss": 0.9977,
+      "step": 5226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6176000359034897e-07,
+      "loss": 1.0343,
+      "step": 5227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.611833581819459e-07,
+      "loss": 1.0106,
+      "step": 5228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.606071271051463e-07,
+      "loss": 0.6493,
+      "step": 5229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6003131050563363e-07,
+      "loss": 1.008,
+      "step": 5230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5945590852898875e-07,
+      "loss": 1.0017,
+      "step": 5231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5888092132068673e-07,
+      "loss": 1.0351,
+      "step": 5232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.583063490260978e-07,
+      "loss": 1.0261,
+      "step": 5233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5773219179048764e-07,
+      "loss": 0.9939,
+      "step": 5234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.571584497590174e-07,
+      "loss": 0.6388,
+      "step": 5235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5658512307674117e-07,
+      "loss": 1.0077,
+      "step": 5236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.560122118886104e-07,
+      "loss": 1.0231,
+      "step": 5237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5543971633947046e-07,
+      "loss": 1.0034,
+      "step": 5238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.548676365740619e-07,
+      "loss": 1.0062,
+      "step": 5239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.542959727370196e-07,
+      "loss": 1.0293,
+      "step": 5240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.537247249728741e-07,
+      "loss": 1.0168,
+      "step": 5241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5315389342605026e-07,
+      "loss": 0.6463,
+      "step": 5242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5258347824086786e-07,
+      "loss": 1.0497,
+      "step": 5243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5201347956154103e-07,
+      "loss": 0.9848,
+      "step": 5244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.514438975321792e-07,
+      "loss": 1.0054,
+      "step": 5245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5087473229678666e-07,
+      "loss": 1.0244,
+      "step": 5246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.503059839992606e-07,
+      "loss": 1.0235,
+      "step": 5247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4973765278339485e-07,
+      "loss": 1.0404,
+      "step": 5248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4916973879287694e-07,
+      "loss": 1.0419,
+      "step": 5249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4860224217128866e-07,
+      "loss": 1.0599,
+      "step": 5250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4803516306210744e-07,
+      "loss": 0.6235,
+      "step": 5251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.474685016087031e-07,
+      "loss": 1.0231,
+      "step": 5252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.469022579543417e-07,
+      "loss": 1.031,
+      "step": 5253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.463364322421827e-07,
+      "loss": 0.9927,
+      "step": 5254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.457710246152807e-07,
+      "loss": 1.0063,
+      "step": 5255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4520603521658375e-07,
+      "loss": 0.9888,
+      "step": 5256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.446414641889346e-07,
+      "loss": 1.0362,
+      "step": 5257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4407731167507015e-07,
+      "loss": 0.9924,
+      "step": 5258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.435135778176217e-07,
+      "loss": 0.662,
+      "step": 5259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.429502627591141e-07,
+      "loss": 1.0088,
+      "step": 5260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4238736664196724e-07,
+      "loss": 1.0109,
+      "step": 5261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.418248896084948e-07,
+      "loss": 0.9969,
+      "step": 5262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.41262831800903e-07,
+      "loss": 0.6671,
+      "step": 5263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4070119336129424e-07,
+      "loss": 0.9811,
+      "step": 5264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4013997443166396e-07,
+      "loss": 1.0252,
+      "step": 5265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3957917515390144e-07,
+      "loss": 1.0116,
+      "step": 5266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.390187956697901e-07,
+      "loss": 0.9947,
+      "step": 5267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3845883612100743e-07,
+      "loss": 1.0414,
+      "step": 5268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3789929664912366e-07,
+      "loss": 1.0061,
+      "step": 5269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.373401773956042e-07,
+      "loss": 0.9715,
+      "step": 5270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.367814785018073e-07,
+      "loss": 1.0502,
+      "step": 5271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3622320010898555e-07,
+      "loss": 0.9827,
+      "step": 5272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.356653423582851e-07,
+      "loss": 1.0098,
+      "step": 5273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3510790539074483e-07,
+      "loss": 0.6251,
+      "step": 5274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3455088934729837e-07,
+      "loss": 1.0283,
+      "step": 5275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.339942943687728e-07,
+      "loss": 1.0164,
+      "step": 5276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3343812059588806e-07,
+      "loss": 1.0356,
+      "step": 5277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3288236816925853e-07,
+      "loss": 1.0226,
+      "step": 5278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3232703722939114e-07,
+      "loss": 1.0314,
+      "step": 5279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3177212791668695e-07,
+      "loss": 1.0371,
+      "step": 5280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3121764037144e-07,
+      "loss": 0.9822,
+      "step": 5281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3066357473383776e-07,
+      "loss": 0.9848,
+      "step": 5282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3010993114396145e-07,
+      "loss": 1.0422,
+      "step": 5283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2955670974178574e-07,
+      "loss": 1.0308,
+      "step": 5284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.290039106671767e-07,
+      "loss": 0.9516,
+      "step": 5285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.284515340598959e-07,
+      "loss": 1.0255,
+      "step": 5286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2789958005959695e-07,
+      "loss": 1.0264,
+      "step": 5287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.273480488058269e-07,
+      "loss": 1.008,
+      "step": 5288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.267969404380262e-07,
+      "loss": 1.0002,
+      "step": 5289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.262462550955281e-07,
+      "loss": 0.9931,
+      "step": 5290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.256959929175582e-07,
+      "loss": 1.0347,
+      "step": 5291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.251461540432363e-07,
+      "loss": 0.6487,
+      "step": 5292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.245967386115747e-07,
+      "loss": 1.075,
+      "step": 5293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.240477467614784e-07,
+      "loss": 1.0027,
+      "step": 5294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.23499178631746e-07,
+      "loss": 1.0266,
+      "step": 5295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2295103436106776e-07,
+      "loss": 1.0136,
+      "step": 5296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2240331408802777e-07,
+      "loss": 1.01,
+      "step": 5297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2185601795110296e-07,
+      "loss": 0.6245,
+      "step": 5298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.213091460886623e-07,
+      "loss": 1.0364,
+      "step": 5299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207626986389684e-07,
+      "loss": 1.0186,
+      "step": 5300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.202166757401759e-07,
+      "loss": 0.6432,
+      "step": 5301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.196710775303322e-07,
+      "loss": 0.9657,
+      "step": 5302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.191259041473775e-07,
+      "loss": 1.0535,
+      "step": 5303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.185811557291445e-07,
+      "loss": 1.035,
+      "step": 5304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1803683241335843e-07,
+      "loss": 0.9906,
+      "step": 5305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.174929343376378e-07,
+      "loss": 0.9539,
+      "step": 5306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1694946163949187e-07,
+      "loss": 1.0309,
+      "step": 5307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.164064144563237e-07,
+      "loss": 0.9896,
+      "step": 5308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.158637929254286e-07,
+      "loss": 1.0224,
+      "step": 5309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.153215971839942e-07,
+      "loss": 1.015,
+      "step": 5310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1477982736910024e-07,
+      "loss": 0.6479,
+      "step": 5311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.142384836177194e-07,
+      "loss": 1.027,
+      "step": 5312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1369756606671535e-07,
+      "loss": 1.0493,
+      "step": 5313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1315707485284535e-07,
+      "loss": 1.0184,
+      "step": 5314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.126170101127581e-07,
+      "loss": 0.6587,
+      "step": 5315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.12077371982995e-07,
+      "loss": 0.6459,
+      "step": 5316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.115381605999894e-07,
+      "loss": 1.0203,
+      "step": 5317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1099937610006686e-07,
+      "loss": 0.6696,
+      "step": 5318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1046101861944406e-07,
+      "loss": 1.0485,
+      "step": 5319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0992308829423097e-07,
+      "loss": 0.9918,
+      "step": 5320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0938558526042924e-07,
+      "loss": 1.0206,
+      "step": 5321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0884850965393215e-07,
+      "loss": 0.9766,
+      "step": 5322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0831186161052516e-07,
+      "loss": 1.0218,
+      "step": 5323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0777564126588537e-07,
+      "loss": 1.0453,
+      "step": 5324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0723984875558253e-07,
+      "loss": 0.6328,
+      "step": 5325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.06704484215077e-07,
+      "loss": 1.0115,
+      "step": 5326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0616954777972214e-07,
+      "loss": 1.0141,
+      "step": 5327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.056350395847627e-07,
+      "loss": 0.6534,
+      "step": 5328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.051009597653342e-07,
+      "loss": 0.5943,
+      "step": 5329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0456730845646485e-07,
+      "loss": 1.0204,
+      "step": 5330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.040340857930748e-07,
+      "loss": 1.0037,
+      "step": 5331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0350129190997486e-07,
+      "loss": 0.9989,
+      "step": 5332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0296892694186807e-07,
+      "loss": 1.0186,
+      "step": 5333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.024369910233495e-07,
+      "loss": 1.012,
+      "step": 5334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0190548428890416e-07,
+      "loss": 0.9931,
+      "step": 5335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0137440687291007e-07,
+      "loss": 1.014,
+      "step": 5336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0084375890963574e-07,
+      "loss": 1.0287,
+      "step": 5337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0031354053324197e-07,
+      "loss": 1.0005,
+      "step": 5338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9978375187778014e-07,
+      "loss": 1.024,
+      "step": 5339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.992543930771938e-07,
+      "loss": 0.9928,
+      "step": 5340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9872546426531696e-07,
+      "loss": 1.0048,
+      "step": 5341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.981969655758756e-07,
+      "loss": 0.639,
+      "step": 5342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9766889714248657e-07,
+      "loss": 1.0131,
+      "step": 5343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9714125909865815e-07,
+      "loss": 1.0334,
+      "step": 5344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9661405157778996e-07,
+      "loss": 0.9908,
+      "step": 5345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9608727471317174e-07,
+      "loss": 0.9803,
+      "step": 5346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9556092863798565e-07,
+      "loss": 1.0343,
+      "step": 5347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9503501348530434e-07,
+      "loss": 1.0285,
+      "step": 5348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.945095293880917e-07,
+      "loss": 1.0014,
+      "step": 5349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.93984476479203e-07,
+      "loss": 1.0237,
+      "step": 5350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9345985489138294e-07,
+      "loss": 0.6476,
+      "step": 5351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.929356647572689e-07,
+      "loss": 1.0349,
+      "step": 5352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.924119062093884e-07,
+      "loss": 0.9741,
+      "step": 5353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9188857938016023e-07,
+      "loss": 1.0162,
+      "step": 5354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.913656844018935e-07,
+      "loss": 1.0127,
+      "step": 5355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9084322140678907e-07,
+      "loss": 0.6464,
+      "step": 5356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.903211905269367e-07,
+      "loss": 0.6739,
+      "step": 5357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8979959189431926e-07,
+      "loss": 1.0126,
+      "step": 5358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.892784256408085e-07,
+      "loss": 1.0212,
+      "step": 5359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.88757691898168e-07,
+      "loss": 0.6154,
+      "step": 5360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.882373907980515e-07,
+      "loss": 0.9897,
+      "step": 5361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8771752247200323e-07,
+      "loss": 0.9767,
+      "step": 5362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8719808705145855e-07,
+      "loss": 0.9986,
+      "step": 5363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.866790846677427e-07,
+      "loss": 0.6275,
+      "step": 5364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.861605154520719e-07,
+      "loss": 1.0255,
+      "step": 5365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.856423795355527e-07,
+      "loss": 0.997,
+      "step": 5366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.851246770491824e-07,
+      "loss": 0.9772,
+      "step": 5367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.846074081238479e-07,
+      "loss": 1.0109,
+      "step": 5368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8409057289032714e-07,
+      "loss": 1.0068,
+      "step": 5369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.835741714792883e-07,
+      "loss": 0.6236,
+      "step": 5370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.830582040212901e-07,
+      "loss": 0.9931,
+      "step": 5371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8254267064678174e-07,
+      "loss": 1.0079,
+      "step": 5372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.820275714861009e-07,
+      "loss": 1.0315,
+      "step": 5373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.815129066694779e-07,
+      "loss": 1.0022,
+      "step": 5374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.809986763270318e-07,
+      "loss": 1.0219,
+      "step": 5375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.804848805887723e-07,
+      "loss": 1.005,
+      "step": 5376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7997151958459884e-07,
+      "loss": 1.0349,
+      "step": 5377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.794585934443023e-07,
+      "loss": 1.0146,
+      "step": 5378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7894610229756097e-07,
+      "loss": 1.014,
+      "step": 5379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7843404627394586e-07,
+      "loss": 1.0388,
+      "step": 5380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7792242550291623e-07,
+      "loss": 0.9811,
+      "step": 5381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.77411240113822e-07,
+      "loss": 0.5945,
+      "step": 5382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.769004902359031e-07,
+      "loss": 1.0366,
+      "step": 5383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.763901759982894e-07,
+      "loss": 1.0249,
+      "step": 5384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7588029752999987e-07,
+      "loss": 1.0168,
+      "step": 5385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.753708549599443e-07,
+      "loss": 1.0248,
+      "step": 5386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7486184841692163e-07,
+      "loss": 1.0465,
+      "step": 5387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7435327802962074e-07,
+      "loss": 1.0103,
+      "step": 5388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7384514392662074e-07,
+      "loss": 1.0089,
+      "step": 5389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.733374462363891e-07,
+      "loss": 0.6647,
+      "step": 5390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7283018508728407e-07,
+      "loss": 1.0272,
+      "step": 5391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7232336060755344e-07,
+      "loss": 1.0281,
+      "step": 5392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7181697292533435e-07,
+      "loss": 0.6577,
+      "step": 5393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7131102216865386e-07,
+      "loss": 0.9929,
+      "step": 5394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.708055084654279e-07,
+      "loss": 1.0151,
+      "step": 5395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.703004319434623e-07,
+      "loss": 1.0147,
+      "step": 5396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.697957927304524e-07,
+      "loss": 1.016,
+      "step": 5397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6929159095398303e-07,
+      "loss": 0.989,
+      "step": 5398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6878782674152823e-07,
+      "loss": 0.9959,
+      "step": 5399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.682845002204515e-07,
+      "loss": 1.0158,
+      "step": 5400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.677816115180063e-07,
+      "loss": 1.0053,
+      "step": 5401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.672791607613338e-07,
+      "loss": 1.0225,
+      "step": 5402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6677714807746586e-07,
+      "loss": 1.0067,
+      "step": 5403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.662755735933231e-07,
+      "loss": 0.9709,
+      "step": 5404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.657744374357156e-07,
+      "loss": 1.0431,
+      "step": 5405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.65273739731342e-07,
+      "loss": 0.9802,
+      "step": 5406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6477348060679114e-07,
+      "loss": 1.0098,
+      "step": 5407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6427366018854e-07,
+      "loss": 0.9857,
+      "step": 5408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.637742786029551e-07,
+      "loss": 1.0201,
+      "step": 5409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6327533597629206e-07,
+      "loss": 1.0051,
+      "step": 5410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.627768324346944e-07,
+      "loss": 1.0381,
+      "step": 5411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6227876810419736e-07,
+      "loss": 1.0371,
+      "step": 5412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.617811431107215e-07,
+      "loss": 1.0038,
+      "step": 5413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.612839575800798e-07,
+      "loss": 1.0237,
+      "step": 5414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6078721163797213e-07,
+      "loss": 1.0363,
+      "step": 5415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.60290905409986e-07,
+      "loss": 1.0168,
+      "step": 5416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.597950390216019e-07,
+      "loss": 0.6863,
+      "step": 5417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.592996125981843e-07,
+      "loss": 1.0087,
+      "step": 5418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5880462626499057e-07,
+      "loss": 1.0526,
+      "step": 5419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5831008014716384e-07,
+      "loss": 1.0211,
+      "step": 5420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.578159743697372e-07,
+      "loss": 0.6465,
+      "step": 5421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5732230905763243e-07,
+      "loss": 1.0303,
+      "step": 5422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5682908433565996e-07,
+      "loss": 1.0427,
+      "step": 5423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.563363003285184e-07,
+      "loss": 0.9952,
+      "step": 5424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5584395716079576e-07,
+      "loss": 0.9772,
+      "step": 5425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5535205495696677e-07,
+      "loss": 0.9978,
+      "step": 5426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5486059384139745e-07,
+      "loss": 1.0179,
+      "step": 5427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.543695739383398e-07,
+      "loss": 1.0238,
+      "step": 5428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5387899537193556e-07,
+      "loss": 1.0184,
+      "step": 5429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5338885826621495e-07,
+      "loss": 1.0275,
+      "step": 5430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5289916274509494e-07,
+      "loss": 1.014,
+      "step": 5431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.524099089323841e-07,
+      "loss": 0.9413,
+      "step": 5432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5192109695177535e-07,
+      "loss": 0.9924,
+      "step": 5433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.514327269268539e-07,
+      "loss": 1.0128,
+      "step": 5434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5094479898108913e-07,
+      "loss": 0.9928,
+      "step": 5435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.504573132378429e-07,
+      "loss": 0.9753,
+      "step": 5436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.499702698203621e-07,
+      "loss": 0.9968,
+      "step": 5437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4948366885178184e-07,
+      "loss": 1.0081,
+      "step": 5438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.489975104551283e-07,
+      "loss": 1.0345,
+      "step": 5439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.485117947533122e-07,
+      "loss": 1.0118,
+      "step": 5440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4802652186913507e-07,
+      "loss": 0.6658,
+      "step": 5441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4754169192528487e-07,
+      "loss": 1.0447,
+      "step": 5442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4705730504433784e-07,
+      "loss": 1.0259,
+      "step": 5443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4657336134875886e-07,
+      "loss": 1.0341,
+      "step": 5444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4608986096090014e-07,
+      "loss": 0.6596,
+      "step": 5445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.456068040030019e-07,
+      "loss": 1.0084,
+      "step": 5446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4512419059719324e-07,
+      "loss": 1.0335,
+      "step": 5447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.446420208654887e-07,
+      "loss": 0.999,
+      "step": 5448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.441602949297936e-07,
+      "loss": 0.9803,
+      "step": 5449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.436790129118984e-07,
+      "loss": 0.987,
+      "step": 5450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4319817493348415e-07,
+      "loss": 1.0371,
+      "step": 5451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.427177811161172e-07,
+      "loss": 1.0309,
+      "step": 5452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.422378315812514e-07,
+      "loss": 1.0178,
+      "step": 5453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.417583264502312e-07,
+      "loss": 1.0354,
+      "step": 5454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4127926584428526e-07,
+      "loss": 0.6341,
+      "step": 5455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4080064988453317e-07,
+      "loss": 1.0151,
+      "step": 5456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.403224786919784e-07,
+      "loss": 0.9983,
+      "step": 5457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3984475238751576e-07,
+      "loss": 1.0376,
+      "step": 5458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.393674710919249e-07,
+      "loss": 0.666,
+      "step": 5459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3889063492587304e-07,
+      "loss": 1.0116,
+      "step": 5460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3841424400991705e-07,
+      "loss": 1.0215,
+      "step": 5461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3793829846449865e-07,
+      "loss": 0.9847,
+      "step": 5462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.374627984099493e-07,
+      "loss": 0.6467,
+      "step": 5463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.369877439664858e-07,
+      "loss": 0.6327,
+      "step": 5464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3651313525421312e-07,
+      "loss": 1.0097,
+      "step": 5465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3603897239312397e-07,
+      "loss": 1.018,
+      "step": 5466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3556525550309782e-07,
+      "loss": 1.0179,
+      "step": 5467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3509198470390145e-07,
+      "loss": 1.0311,
+      "step": 5468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3461916011518946e-07,
+      "loss": 1.0024,
+      "step": 5469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.341467818565015e-07,
+      "loss": 1.0494,
+      "step": 5470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3367485004726806e-07,
+      "loss": 1.0203,
+      "step": 5471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.332033648068028e-07,
+      "loss": 0.6355,
+      "step": 5472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3273232625430972e-07,
+      "loss": 1.0577,
+      "step": 5473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.322617345088782e-07,
+      "loss": 1.0123,
+      "step": 5474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3179158968948398e-07,
+      "loss": 1.043,
+      "step": 5475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.313218919149922e-07,
+      "loss": 1.0267,
+      "step": 5476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3085264130415204e-07,
+      "loss": 1.0341,
+      "step": 5477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3038383797560267e-07,
+      "loss": 1.0186,
+      "step": 5478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2991548204786748e-07,
+      "loss": 0.9945,
+      "step": 5479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2944757363935907e-07,
+      "loss": 1.0033,
+      "step": 5480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2898011286837483e-07,
+      "loss": 1.0309,
+      "step": 5481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2851309985309997e-07,
+      "loss": 1.0187,
+      "step": 5482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2804653471160694e-07,
+      "loss": 1.0057,
+      "step": 5483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2758041756185388e-07,
+      "loss": 0.9901,
+      "step": 5484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.271147485216871e-07,
+      "loss": 0.947,
+      "step": 5485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.266495277088376e-07,
+      "loss": 1.0209,
+      "step": 5486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2618475524092462e-07,
+      "loss": 1.0063,
+      "step": 5487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2572043123545392e-07,
+      "loss": 0.6538,
+      "step": 5488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.252565558098174e-07,
+      "loss": 1.0032,
+      "step": 5489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2479312908129365e-07,
+      "loss": 1.0325,
+      "step": 5490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2433015116704857e-07,
+      "loss": 1.0229,
+      "step": 5491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2386762218413248e-07,
+      "loss": 1.0156,
+      "step": 5492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2340554224948515e-07,
+      "loss": 1.0269,
+      "step": 5493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2294391147993007e-07,
+      "loss": 1.0096,
+      "step": 5494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2248272999217987e-07,
+      "loss": 0.9966,
+      "step": 5495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2202199790283126e-07,
+      "loss": 0.9851,
+      "step": 5496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2156171532836755e-07,
+      "loss": 1.0034,
+      "step": 5497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2110188238516048e-07,
+      "loss": 1.0257,
+      "step": 5498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2064249918946508e-07,
+      "loss": 1.047,
+      "step": 5499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2018356585742603e-07,
+      "loss": 1.0544,
+      "step": 5500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1972508250507095e-07,
+      "loss": 0.6578,
+      "step": 5501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1926704924831686e-07,
+      "loss": 0.9619,
+      "step": 5502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1880946620296403e-07,
+      "loss": 1.0381,
+      "step": 5503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1835233348470083e-07,
+      "loss": 1.0198,
+      "step": 5504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1789565120910124e-07,
+      "loss": 1.0345,
+      "step": 5505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1743941949162515e-07,
+      "loss": 0.9919,
+      "step": 5506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1698363844761892e-07,
+      "loss": 1.0571,
+      "step": 5507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1652830819231504e-07,
+      "loss": 0.9792,
+      "step": 5508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1607342884083078e-07,
+      "loss": 0.9907,
+      "step": 5509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1561900050817172e-07,
+      "loss": 1.0361,
+      "step": 5510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1516502330922726e-07,
+      "loss": 0.9864,
+      "step": 5511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.147114973587736e-07,
+      "loss": 1.0072,
+      "step": 5512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1425842277147322e-07,
+      "loss": 0.9879,
+      "step": 5513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1380579966187317e-07,
+      "loss": 0.9682,
+      "step": 5514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1335362814440881e-07,
+      "loss": 0.9669,
+      "step": 5515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1290190833339804e-07,
+      "loss": 1.0061,
+      "step": 5516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1245064034304794e-07,
+      "loss": 1.0288,
+      "step": 5517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.11999824287449e-07,
+      "loss": 1.0327,
+      "step": 5518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.115494602805774e-07,
+      "loss": 0.9987,
+      "step": 5519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.110995484362974e-07,
+      "loss": 1.0073,
+      "step": 5520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1065008886835578e-07,
+      "loss": 1.0081,
+      "step": 5521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.102010816903881e-07,
+      "loss": 1.0135,
+      "step": 5522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0975252701591217e-07,
+      "loss": 0.975,
+      "step": 5523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0930442495833534e-07,
+      "loss": 0.9782,
+      "step": 5524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0885677563094672e-07,
+      "loss": 1.0484,
+      "step": 5525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0840957914692338e-07,
+      "loss": 0.9985,
+      "step": 5526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.079628356193268e-07,
+      "loss": 1.0054,
+      "step": 5527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0751654516110472e-07,
+      "loss": 0.9974,
+      "step": 5528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0707070788508951e-07,
+      "loss": 0.6527,
+      "step": 5529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0662532390400011e-07,
+      "loss": 0.632,
+      "step": 5530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0618039333043892e-07,
+      "loss": 1.0015,
+      "step": 5531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0573591627689634e-07,
+      "loss": 0.6416,
+      "step": 5532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0529189285574523e-07,
+      "loss": 1.0472,
+      "step": 5533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0484832317924661e-07,
+      "loss": 0.9703,
+      "step": 5534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.044052073595446e-07,
+      "loss": 0.9925,
+      "step": 5535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0396254550866888e-07,
+      "loss": 1.014,
+      "step": 5536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0352033773853593e-07,
+      "loss": 0.6315,
+      "step": 5537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0307858416094525e-07,
+      "loss": 0.9862,
+      "step": 5538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0263728488758392e-07,
+      "loss": 0.9603,
+      "step": 5539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.02196440030022e-07,
+      "loss": 1.0233,
+      "step": 5540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0175604969971482e-07,
+      "loss": 1.0377,
+      "step": 5541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0131611400800485e-07,
+      "loss": 1.0099,
+      "step": 5542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.008766330661169e-07,
+      "loss": 1.0304,
+      "step": 5543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0043760698516365e-07,
+      "loss": 1.0059,
+      "step": 5544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999903587613987e-07,
+      "loss": 0.6514,
+      "step": 5545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9956091984992773e-07,
+      "loss": 1.0287,
+      "step": 5546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9912325901729288e-07,
+      "loss": 1.0208,
+      "step": 5547
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9868605348888613e-07,
+      "loss": 1.0152,
+      "step": 5548
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.982493033752435e-07,
+      "loss": 1.0219,
+      "step": 5549
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9781300878678598e-07,
+      "loss": 1.0119,
+      "step": 5550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.973771698338189e-07,
+      "loss": 1.0305,
+      "step": 5551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969417866265335e-07,
+      "loss": 1.0128,
+      "step": 5552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9650685927500321e-07,
+      "loss": 0.9982,
+      "step": 5553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9607238788918968e-07,
+      "loss": 1.0109,
+      "step": 5554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.956383725789361e-07,
+      "loss": 0.6882,
+      "step": 5555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9520481345397344e-07,
+      "loss": 1.0043,
+      "step": 5556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9477171062391462e-07,
+      "loss": 1.0234,
+      "step": 5557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9433906419825785e-07,
+      "loss": 1.0182,
+      "step": 5558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9390687428638765e-07,
+      "loss": 0.9863,
+      "step": 5559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9347514099757035e-07,
+      "loss": 0.6267,
+      "step": 5560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9304386444095998e-07,
+      "loss": 0.6344,
+      "step": 5561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.926130447255925e-07,
+      "loss": 1.0257,
+      "step": 5562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9218268196038912e-07,
+      "loss": 1.0147,
+      "step": 5563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9175277625415644e-07,
+      "loss": 0.9724,
+      "step": 5564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9132332771558412e-07,
+      "loss": 1.0338,
+      "step": 5565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9089433645324737e-07,
+      "loss": 1.0333,
+      "step": 5566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9046580257560497e-07,
+      "loss": 0.9885,
+      "step": 5567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9003772619100112e-07,
+      "loss": 0.9981,
+      "step": 5568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896101074076628e-07,
+      "loss": 0.9973,
+      "step": 5569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8918294633370247e-07,
+      "loss": 1.0275,
+      "step": 5570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8875624307711656e-07,
+      "loss": 0.992,
+      "step": 5571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8832999774578594e-07,
+      "loss": 0.6632,
+      "step": 5572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8790421044747507e-07,
+      "loss": 1.0229,
+      "step": 5573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.874788812898338e-07,
+      "loss": 0.6473,
+      "step": 5574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8705401038039414e-07,
+      "loss": 1.0033,
+      "step": 5575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8662959782657505e-07,
+      "loss": 1.03,
+      "step": 5576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8620564373567649e-07,
+      "loss": 0.9543,
+      "step": 5577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8578214821488558e-07,
+      "loss": 1.0109,
+      "step": 5578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8535911137127137e-07,
+      "loss": 1.035,
+      "step": 5579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8493653331178694e-07,
+      "loss": 1.0241,
+      "step": 5580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8451441414327107e-07,
+      "loss": 0.9911,
+      "step": 5581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8409275397244418e-07,
+      "loss": 1.012,
+      "step": 5582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8367155290591362e-07,
+      "loss": 1.0177,
+      "step": 5583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.832508110501678e-07,
+      "loss": 1.0401,
+      "step": 5584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.828305285115803e-07,
+      "loss": 1.0015,
+      "step": 5585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.824107053964088e-07,
+      "loss": 1.0161,
+      "step": 5586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8199134181079457e-07,
+      "loss": 1.0531,
+      "step": 5587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8157243786076216e-07,
+      "loss": 0.6604,
+      "step": 5588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8115399365222083e-07,
+      "loss": 1.0314,
+      "step": 5589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8073600929096311e-07,
+      "loss": 1.0028,
+      "step": 5590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8031848488266555e-07,
+      "loss": 1.0421,
+      "step": 5591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7990142053288725e-07,
+      "loss": 1.0186,
+      "step": 5592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7948481634707346e-07,
+      "loss": 0.6892,
+      "step": 5593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7906867243055013e-07,
+      "loss": 1.0062,
+      "step": 5594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7865298888852887e-07,
+      "loss": 1.0295,
+      "step": 5595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7823776582610473e-07,
+      "loss": 0.9842,
+      "step": 5596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7782300334825484e-07,
+      "loss": 0.9909,
+      "step": 5597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7740870155984222e-07,
+      "loss": 1.0462,
+      "step": 5598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.769948605656111e-07,
+      "loss": 0.9752,
+      "step": 5599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7658148047019128e-07,
+      "loss": 0.6581,
+      "step": 5600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7616856137809432e-07,
+      "loss": 0.9733,
+      "step": 5601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.757561033937156e-07,
+      "loss": 1.0135,
+      "step": 5602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7534410662133525e-07,
+      "loss": 1.0096,
+      "step": 5603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7493257116511483e-07,
+      "loss": 1.0141,
+      "step": 5604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7452149712910114e-07,
+      "loss": 0.9425,
+      "step": 5605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.741108846172228e-07,
+      "loss": 0.6375,
+      "step": 5606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7370073373329253e-07,
+      "loss": 0.995,
+      "step": 5607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7329104458100606e-07,
+      "loss": 1.0471,
+      "step": 5608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7288181726394258e-07,
+      "loss": 1.0034,
+      "step": 5609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.724730518855646e-07,
+      "loss": 0.9825,
+      "step": 5610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.720647485492175e-07,
+      "loss": 1.0057,
+      "step": 5611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7165690735813e-07,
+      "loss": 1.0271,
+      "step": 5612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.712495284154145e-07,
+      "loss": 0.6538,
+      "step": 5613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.708426118240649e-07,
+      "loss": 1.0064,
+      "step": 5614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7043615768696063e-07,
+      "loss": 1.0268,
+      "step": 5615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7003016610686172e-07,
+      "loss": 1.0262,
+      "step": 5616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6962463718641363e-07,
+      "loss": 1.0287,
+      "step": 5617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.69219571028143e-07,
+      "loss": 0.992,
+      "step": 5618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6881496773445947e-07,
+      "loss": 0.9946,
+      "step": 5619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.684108274076579e-07,
+      "loss": 1.0237,
+      "step": 5620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6800715014991296e-07,
+      "loss": 1.0068,
+      "step": 5621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6760393606328504e-07,
+      "loss": 1.0038,
+      "step": 5622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6720118524971593e-07,
+      "loss": 1.0063,
+      "step": 5623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667988978110293e-07,
+      "loss": 0.6624,
+      "step": 5624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6639707384893486e-07,
+      "loss": 1.0062,
+      "step": 5625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.659957134650216e-07,
+      "loss": 0.9829,
+      "step": 5626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6559481676076414e-07,
+      "loss": 1.027,
+      "step": 5627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6519438383751783e-07,
+      "loss": 1.0395,
+      "step": 5628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6479441479652168e-07,
+      "loss": 1.0236,
+      "step": 5629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.643949097388977e-07,
+      "loss": 1.0418,
+      "step": 5630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6399586876564997e-07,
+      "loss": 0.9899,
+      "step": 5631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6359729197766536e-07,
+      "loss": 0.9929,
+      "step": 5632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6319917947571394e-07,
+      "loss": 0.9773,
+      "step": 5633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6280153136044761e-07,
+      "loss": 0.9942,
+      "step": 5634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6240434773240153e-07,
+      "loss": 1.0036,
+      "step": 5635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6200762869199225e-07,
+      "loss": 0.9977,
+      "step": 5636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.616113743395211e-07,
+      "loss": 0.6234,
+      "step": 5637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6121558477516906e-07,
+      "loss": 0.9809,
+      "step": 5638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6082026009900274e-07,
+      "loss": 0.987,
+      "step": 5639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6042540041096843e-07,
+      "loss": 1.0174,
+      "step": 5640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.600310058108958e-07,
+      "loss": 1.0119,
+      "step": 5641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.59637076398498e-07,
+      "loss": 1.0191,
+      "step": 5642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5924361227336868e-07,
+      "loss": 0.9878,
+      "step": 5643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.588506135349863e-07,
+      "loss": 0.637,
+      "step": 5644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.584580802827089e-07,
+      "loss": 0.9781,
+      "step": 5645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5806601261577868e-07,
+      "loss": 0.9858,
+      "step": 5646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5767441063331964e-07,
+      "loss": 1.0133,
+      "step": 5647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.57283274434338e-07,
+      "loss": 1.0315,
+      "step": 5648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5689260411772275e-07,
+      "loss": 1.0309,
+      "step": 5649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5650239978224345e-07,
+      "loss": 0.9983,
+      "step": 5650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5611266152655377e-07,
+      "loss": 1.0184,
+      "step": 5651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.557233894491885e-07,
+      "loss": 1.0217,
+      "step": 5652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.553345836485649e-07,
+      "loss": 0.6242,
+      "step": 5653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5494624422298254e-07,
+      "loss": 1.0029,
+      "step": 5654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.545583712706222e-07,
+      "loss": 1.0062,
+      "step": 5655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5417096488954795e-07,
+      "loss": 0.9806,
+      "step": 5656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.537840251777054e-07,
+      "loss": 0.9853,
+      "step": 5657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5339755223292072e-07,
+      "loss": 0.9924,
+      "step": 5658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5301154615290534e-07,
+      "loss": 1.0402,
+      "step": 5659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5262600703524896e-07,
+      "loss": 1.0512,
+      "step": 5660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5224093497742651e-07,
+      "loss": 1.0202,
+      "step": 5661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5185633007679255e-07,
+      "loss": 0.9898,
+      "step": 5662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5147219243058372e-07,
+      "loss": 1.0175,
+      "step": 5663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5108852213592038e-07,
+      "loss": 1.0184,
+      "step": 5664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5070531928980223e-07,
+      "loss": 0.9797,
+      "step": 5665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5032258398911313e-07,
+      "loss": 1.0309,
+      "step": 5666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4994031633061677e-07,
+      "loss": 1.0046,
+      "step": 5667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4955851641095962e-07,
+      "loss": 0.9906,
+      "step": 5668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4917718432667003e-07,
+      "loss": 1.0307,
+      "step": 5669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4879632017415754e-07,
+      "loss": 1.0381,
+      "step": 5670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.484159240497136e-07,
+      "loss": 1.0352,
+      "step": 5671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4803599604951167e-07,
+      "loss": 0.9912,
+      "step": 5672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4765653626960584e-07,
+      "loss": 1.0078,
+      "step": 5673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4727754480593358e-07,
+      "loss": 1.0364,
+      "step": 5674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4689902175431134e-07,
+      "loss": 0.9978,
+      "step": 5675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.465209672104406e-07,
+      "loss": 1.043,
+      "step": 5676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.461433812699011e-07,
+      "loss": 1.0197,
+      "step": 5677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4576626402815584e-07,
+      "loss": 1.0266,
+      "step": 5678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.453896155805494e-07,
+      "loss": 1.0135,
+      "step": 5679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4501343602230632e-07,
+      "loss": 0.9709,
+      "step": 5680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4463772544853537e-07,
+      "loss": 1.0211,
+      "step": 5681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4426248395422325e-07,
+      "loss": 1.0167,
+      "step": 5682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.438877116342414e-07,
+      "loss": 1.0535,
+      "step": 5683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.435134085833405e-07,
+      "loss": 1.0116,
+      "step": 5684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4313957489615259e-07,
+      "loss": 1.0089,
+      "step": 5685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4276621066719297e-07,
+      "loss": 1.0026,
+      "step": 5686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.423933159908557e-07,
+      "loss": 1.0327,
+      "step": 5687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4202089096141844e-07,
+      "loss": 1.0689,
+      "step": 5688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4164893567303814e-07,
+      "loss": 0.7002,
+      "step": 5689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4127745021975425e-07,
+      "loss": 0.963,
+      "step": 5690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4090643469548736e-07,
+      "loss": 0.9847,
+      "step": 5691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4053588919403847e-07,
+      "loss": 1.015,
+      "step": 5692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4016581380909066e-07,
+      "loss": 0.9877,
+      "step": 5693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3979620863420793e-07,
+      "loss": 1.0235,
+      "step": 5694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3942707376283403e-07,
+      "loss": 1.0106,
+      "step": 5695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3905840928829672e-07,
+      "loss": 1.0036,
+      "step": 5696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.386902153038012e-07,
+      "loss": 0.6579,
+      "step": 5697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.383224919024375e-07,
+      "loss": 1.0476,
+      "step": 5698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3795523917717323e-07,
+      "loss": 0.9783,
+      "step": 5699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3758845722085965e-07,
+      "loss": 0.9981,
+      "step": 5700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3722214612622774e-07,
+      "loss": 1.0146,
+      "step": 5701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.368563059858887e-07,
+      "loss": 0.6564,
+      "step": 5702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3649093689233683e-07,
+      "loss": 0.9769,
+      "step": 5703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3612603893794482e-07,
+      "loss": 1.0028,
+      "step": 5704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3576161221496875e-07,
+      "loss": 1.014,
+      "step": 5705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.353976568155437e-07,
+      "loss": 1.018,
+      "step": 5706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3503417283168572e-07,
+      "loss": 0.9824,
+      "step": 5707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3467116035529324e-07,
+      "loss": 0.9908,
+      "step": 5708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.343086194781433e-07,
+      "loss": 0.6422,
+      "step": 5709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3394655029189572e-07,
+      "loss": 0.9872,
+      "step": 5710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.335849528880897e-07,
+      "loss": 1.0201,
+      "step": 5711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3322382735814542e-07,
+      "loss": 1.0243,
+      "step": 5712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.328631737933643e-07,
+      "loss": 0.9842,
+      "step": 5713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3250299228492812e-07,
+      "loss": 0.992,
+      "step": 5714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3214328292389908e-07,
+      "loss": 1.0095,
+      "step": 5715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.317840458012205e-07,
+      "loss": 0.9774,
+      "step": 5716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3142528100771543e-07,
+      "loss": 0.9996,
+      "step": 5717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3106698863408894e-07,
+      "loss": 1.0085,
+      "step": 5718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3070916877092474e-07,
+      "loss": 0.9915,
+      "step": 5719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.303518215086894e-07,
+      "loss": 0.997,
+      "step": 5720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.299949469377275e-07,
+      "loss": 0.9808,
+      "step": 5721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2963854514826666e-07,
+      "loss": 1.016,
+      "step": 5722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2928261623041304e-07,
+      "loss": 1.0404,
+      "step": 5723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2892716027415306e-07,
+      "loss": 1.0416,
+      "step": 5724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.28572177369356e-07,
+      "loss": 1.0262,
+      "step": 5725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2821766760576848e-07,
+      "loss": 0.6553,
+      "step": 5726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2786363107302011e-07,
+      "loss": 1.043,
+      "step": 5727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2751006786061913e-07,
+      "loss": 0.9833,
+      "step": 5728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2715697805795466e-07,
+      "loss": 0.9838,
+      "step": 5729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2680436175429598e-07,
+      "loss": 0.6355,
+      "step": 5730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2645221903879334e-07,
+      "loss": 1.0162,
+      "step": 5731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.261005500004768e-07,
+      "loss": 0.9843,
+      "step": 5732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.25749354728256e-07,
+      "loss": 0.99,
+      "step": 5733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.253986333109216e-07,
+      "loss": 0.9775,
+      "step": 5734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2504838583714428e-07,
+      "loss": 0.9996,
+      "step": 5735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2469861239547518e-07,
+      "loss": 0.9853,
+      "step": 5736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2434931307434516e-07,
+      "loss": 0.9438,
+      "step": 5737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.240004879620653e-07,
+      "loss": 1.0045,
+      "step": 5738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2365213714682643e-07,
+      "loss": 0.6719,
+      "step": 5739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2330426071670096e-07,
+      "loss": 1.0592,
+      "step": 5740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2295685875963878e-07,
+      "loss": 1.016,
+      "step": 5741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2260993136347274e-07,
+      "loss": 0.9741,
+      "step": 5742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2226347861591334e-07,
+      "loss": 0.9979,
+      "step": 5743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2191750060455275e-07,
+      "loss": 1.0277,
+      "step": 5744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.215719974168621e-07,
+      "loss": 0.6889,
+      "step": 5745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2122696914019192e-07,
+      "loss": 1.0054,
+      "step": 5746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2088241586177495e-07,
+      "loss": 1.0243,
+      "step": 5747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2053833766872123e-07,
+      "loss": 1.0343,
+      "step": 5748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.20194734648023e-07,
+      "loss": 1.0284,
+      "step": 5749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1985160688655006e-07,
+      "loss": 1.0225,
+      "step": 5750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.195089544710539e-07,
+      "loss": 1.0203,
+      "step": 5751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1916677748816505e-07,
+      "loss": 0.9657,
+      "step": 5752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1882507602439384e-07,
+      "loss": 0.9927,
+      "step": 5753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1848385016613072e-07,
+      "loss": 0.9884,
+      "step": 5754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1814309999964556e-07,
+      "loss": 0.9773,
+      "step": 5755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1780282561108746e-07,
+      "loss": 0.9954,
+      "step": 5756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1746302708648692e-07,
+      "loss": 1.0422,
+      "step": 5757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.171237045117519e-07,
+      "loss": 1.008,
+      "step": 5758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1678485797267245e-07,
+      "loss": 1.0021,
+      "step": 5759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1644648755491626e-07,
+      "loss": 0.9989,
+      "step": 5760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1610859334403067e-07,
+      "loss": 1.0175,
+      "step": 5761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1577117542544468e-07,
+      "loss": 0.9622,
+      "step": 5762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.154342338844645e-07,
+      "loss": 0.9991,
+      "step": 5763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1509776880627775e-07,
+      "loss": 0.9773,
+      "step": 5764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1476178027594974e-07,
+      "loss": 0.6342,
+      "step": 5765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1442626837842739e-07,
+      "loss": 0.6431,
+      "step": 5766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1409123319853576e-07,
+      "loss": 1.0074,
+      "step": 5767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1375667482097884e-07,
+      "loss": 0.9618,
+      "step": 5768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.134225933303421e-07,
+      "loss": 0.9853,
+      "step": 5769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1308898881108798e-07,
+      "loss": 1.0183,
+      "step": 5770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1275586134756098e-07,
+      "loss": 1.031,
+      "step": 5771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1242321102398244e-07,
+      "loss": 1.0197,
+      "step": 5772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1209103792445485e-07,
+      "loss": 1.0209,
+      "step": 5773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1175934213295924e-07,
+      "loss": 0.9722,
+      "step": 5774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1142812373335609e-07,
+      "loss": 1.0345,
+      "step": 5775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.110973828093853e-07,
+      "loss": 0.996,
+      "step": 5776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1076711944466643e-07,
+      "loss": 1.0258,
+      "step": 5777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.104373337226967e-07,
+      "loss": 1.0231,
+      "step": 5778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1010802572685518e-07,
+      "loss": 1.022,
+      "step": 5779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0977919554039749e-07,
+      "loss": 1.0046,
+      "step": 5780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.094508432464607e-07,
+      "loss": 1.0142,
+      "step": 5781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0912296892805972e-07,
+      "loss": 1.0086,
+      "step": 5782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0879557266808803e-07,
+      "loss": 1.0263,
+      "step": 5783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0846865454932075e-07,
+      "loss": 1.0304,
+      "step": 5784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0814221465440887e-07,
+      "loss": 0.9963,
+      "step": 5785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0781625306588549e-07,
+      "loss": 1.0269,
+      "step": 5786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0749076986616024e-07,
+      "loss": 0.9839,
+      "step": 5787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0716576513752395e-07,
+      "loss": 0.6196,
+      "step": 5788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0684123896214536e-07,
+      "loss": 1.0098,
+      "step": 5789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0651719142207127e-07,
+      "loss": 1.0316,
+      "step": 5790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0619362259923015e-07,
+      "loss": 0.653,
+      "step": 5791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0587053257542656e-07,
+      "loss": 1.0117,
+      "step": 5792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0554792143234625e-07,
+      "loss": 1.0008,
+      "step": 5793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0522578925155224e-07,
+      "loss": 1.0094,
+      "step": 5794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0490413611448757e-07,
+      "loss": 0.6432,
+      "step": 5795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0458296210247363e-07,
+      "loss": 0.6751,
+      "step": 5796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0426226729671061e-07,
+      "loss": 0.9995,
+      "step": 5797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0394205177827808e-07,
+      "loss": 0.9968,
+      "step": 5798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.036223156281344e-07,
+      "loss": 0.9733,
+      "step": 5799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0330305892711533e-07,
+      "loss": 1.0172,
+      "step": 5800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0298428175593766e-07,
+      "loss": 0.9921,
+      "step": 5801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0266598419519468e-07,
+      "loss": 0.6511,
+      "step": 5802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0234816632536092e-07,
+      "loss": 1.0038,
+      "step": 5803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0203082822678722e-07,
+      "loss": 1.0358,
+      "step": 5804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0171396997970405e-07,
+      "loss": 0.9823,
+      "step": 5805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0139759166422179e-07,
+      "loss": 0.6403,
+      "step": 5806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.010816933603269e-07,
+      "loss": 1.0631,
+      "step": 5807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0076627514788728e-07,
+      "loss": 1.0106,
+      "step": 5808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0045133710664688e-07,
+      "loss": 1.0245,
+      "step": 5809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0013687931623071e-07,
+      "loss": 1.055,
+      "step": 5810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.982290185614006e-08,
+      "loss": 0.987,
+      "step": 5811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.950940480575653e-08,
+      "loss": 0.9904,
+      "step": 5812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.91963882443394e-08,
+      "loss": 1.003,
+      "step": 5813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.888385225102669e-08,
+      "loss": 0.969,
+      "step": 5814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.857179690483497e-08,
+      "loss": 1.0282,
+      "step": 5815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.826022228465891e-08,
+      "loss": 0.9911,
+      "step": 5816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.794912846927217e-08,
+      "loss": 0.9872,
+      "step": 5817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.76385155373265e-08,
+      "loss": 0.624,
+      "step": 5818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.732838356735218e-08,
+      "loss": 1.0527,
+      "step": 5819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.701873263775806e-08,
+      "loss": 1.001,
+      "step": 5820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.670956282683151e-08,
+      "loss": 0.6452,
+      "step": 5821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.640087421273713e-08,
+      "loss": 0.9975,
+      "step": 5822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.609266687351958e-08,
+      "loss": 0.6307,
+      "step": 5823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.57849408871001e-08,
+      "loss": 0.9958,
+      "step": 5824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.547769633128e-08,
+      "loss": 1.0142,
+      "step": 5825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.51709332837376e-08,
+      "loss": 0.9958,
+      "step": 5826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.486465182202929e-08,
+      "loss": 0.9637,
+      "step": 5827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.455885202359138e-08,
+      "loss": 1.0046,
+      "step": 5828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.425353396573599e-08,
+      "loss": 1.0023,
+      "step": 5829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.394869772565605e-08,
+      "loss": 0.983,
+      "step": 5830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.364434338042015e-08,
+      "loss": 0.7123,
+      "step": 5831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.33404710069774e-08,
+      "loss": 1.0212,
+      "step": 5832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.30370806821532e-08,
+      "loss": 1.0068,
+      "step": 5833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.27341724826518e-08,
+      "loss": 1.0133,
+      "step": 5834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.243174648505591e-08,
+      "loss": 0.6448,
+      "step": 5835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.212980276582572e-08,
+      "loss": 1.0214,
+      "step": 5836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.182834140129991e-08,
+      "loss": 1.0191,
+      "step": 5837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.152736246769533e-08,
+      "loss": 1.0096,
+      "step": 5838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.122686604110552e-08,
+      "loss": 1.0229,
+      "step": 5839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.09268521975044e-08,
+      "loss": 0.9963,
+      "step": 5840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.062732101274129e-08,
+      "loss": 1.0066,
+      "step": 5841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.032827256254605e-08,
+      "loss": 1.0289,
+      "step": 5842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.002970692252443e-08,
+      "loss": 0.9998,
+      "step": 5843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.97316241681605e-08,
+      "loss": 1.0134,
+      "step": 5844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.943402437481773e-08,
+      "loss": 0.9704,
+      "step": 5845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.913690761773528e-08,
+      "loss": 1.0188,
+      "step": 5846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.88402739720322e-08,
+      "loss": 1.032,
+      "step": 5847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.85441235127038e-08,
+      "loss": 0.6542,
+      "step": 5848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.82484563146244e-08,
+      "loss": 1.0189,
+      "step": 5849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.79532724525458e-08,
+      "loss": 1.0206,
+      "step": 5850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.76585720010965e-08,
+      "loss": 1.0091,
+      "step": 5851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.73643550347849e-08,
+      "loss": 0.9659,
+      "step": 5852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.707062162799483e-08,
+      "loss": 0.6056,
+      "step": 5853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.677737185499023e-08,
+      "loss": 1.0194,
+      "step": 5854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.64846057899109e-08,
+      "loss": 0.638,
+      "step": 5855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.619232350677518e-08,
+      "loss": 1.0127,
+      "step": 5856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.590052507947887e-08,
+      "loss": 1.0069,
+      "step": 5857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.56092105817956e-08,
+      "loss": 0.9961,
+      "step": 5858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.531838008737669e-08,
+      "loss": 1.0217,
+      "step": 5859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.502803366975087e-08,
+      "loss": 0.9839,
+      "step": 5860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.473817140232409e-08,
+      "loss": 0.6254,
+      "step": 5861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.444879335838129e-08,
+      "loss": 1.0156,
+      "step": 5862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.415989961108305e-08,
+      "loss": 1.0311,
+      "step": 5863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.387149023346985e-08,
+      "loss": 1.0006,
+      "step": 5864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.358356529845734e-08,
+      "loss": 1.0294,
+      "step": 5865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.329612487883974e-08,
+      "loss": 0.9849,
+      "step": 5866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.30091690472896e-08,
+      "loss": 1.0049,
+      "step": 5867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.27226978763551e-08,
+      "loss": 1.0241,
+      "step": 5868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.243671143846409e-08,
+      "loss": 1.0166,
+      "step": 5869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.21512098059196e-08,
+      "loss": 1.0277,
+      "step": 5870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.186619305090414e-08,
+      "loss": 0.6392,
+      "step": 5871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.158166124547627e-08,
+      "loss": 1.0307,
+      "step": 5872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.129761446157179e-08,
+      "loss": 0.988,
+      "step": 5873
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.101405277100548e-08,
+      "loss": 0.986,
+      "step": 5874
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.073097624546732e-08,
+      "loss": 1.01,
+      "step": 5875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.044838495652717e-08,
+      "loss": 1.0142,
+      "step": 5876
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.016627897562944e-08,
+      "loss": 1.0297,
+      "step": 5877
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.988465837409797e-08,
+      "loss": 1.0114,
+      "step": 5878
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.960352322313269e-08,
+      "loss": 0.9858,
+      "step": 5879
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.932287359381162e-08,
+      "loss": 1.0456,
+      "step": 5880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.904270955708937e-08,
+      "loss": 1.0088,
+      "step": 5881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.876303118379812e-08,
+      "loss": 1.0009,
+      "step": 5882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.848383854464691e-08,
+      "loss": 1.0056,
+      "step": 5883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.820513171022281e-08,
+      "loss": 1.0203,
+      "step": 5884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.792691075098857e-08,
+      "loss": 0.6461,
+      "step": 5885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.764917573728635e-08,
+      "loss": 1.0054,
+      "step": 5886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.737192673933313e-08,
+      "loss": 1.0287,
+      "step": 5887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.709516382722392e-08,
+      "loss": 1.0172,
+      "step": 5888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.681888707093187e-08,
+      "loss": 0.974,
+      "step": 5889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.65430965403051e-08,
+      "loss": 1.0171,
+      "step": 5890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.626779230507096e-08,
+      "loss": 1.0654,
+      "step": 5891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.5992974434832e-08,
+      "loss": 1.0232,
+      "step": 5892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.571864299906971e-08,
+      "loss": 1.043,
+      "step": 5893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.544479806714088e-08,
+      "loss": 0.9934,
+      "step": 5894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.51714397082801e-08,
+      "loss": 1.0177,
+      "step": 5895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.489856799159877e-08,
+      "loss": 1.0314,
+      "step": 5896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.462618298608525e-08,
+      "loss": 0.6404,
+      "step": 5897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.435428476060557e-08,
+      "loss": 1.0719,
+      "step": 5898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.408287338390096e-08,
+      "loss": 0.9826,
+      "step": 5899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.381194892459142e-08,
+      "loss": 0.9806,
+      "step": 5900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.354151145117282e-08,
+      "loss": 0.9715,
+      "step": 5901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.327156103201804e-08,
+      "loss": 1.0201,
+      "step": 5902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.300209773537713e-08,
+      "loss": 1.0551,
+      "step": 5903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.273312162937695e-08,
+      "loss": 1.0012,
+      "step": 5904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.24646327820202e-08,
+      "loss": 0.9933,
+      "step": 5905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.219663126118813e-08,
+      "loss": 1.0384,
+      "step": 5906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.192911713463723e-08,
+      "loss": 1.0354,
+      "step": 5907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.166209047000227e-08,
+      "loss": 0.656,
+      "step": 5908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.139555133479303e-08,
+      "loss": 0.9584,
+      "step": 5909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.112949979639693e-08,
+      "loss": 1.0253,
+      "step": 5910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.086393592207862e-08,
+      "loss": 0.6244,
+      "step": 5911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.059885977897839e-08,
+      "loss": 1.0165,
+      "step": 5912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.033427143411441e-08,
+      "loss": 1.0046,
+      "step": 5913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.007017095438005e-08,
+      "loss": 1.0016,
+      "step": 5914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.980655840654704e-08,
+      "loss": 0.6217,
+      "step": 5915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.954343385726202e-08,
+      "loss": 0.6517,
+      "step": 5916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.928079737304982e-08,
+      "loss": 1.0456,
+      "step": 5917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.901864902031062e-08,
+      "loss": 1.0059,
+      "step": 5918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.875698886532189e-08,
+      "loss": 1.0342,
+      "step": 5919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.849581697423779e-08,
+      "loss": 0.987,
+      "step": 5920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.823513341308862e-08,
+      "loss": 0.9672,
+      "step": 5921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.797493824778056e-08,
+      "loss": 0.6391,
+      "step": 5922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.771523154409853e-08,
+      "loss": 1.0119,
+      "step": 5923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.745601336770113e-08,
+      "loss": 0.9885,
+      "step": 5924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.719728378412615e-08,
+      "loss": 0.9895,
+      "step": 5925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.693904285878571e-08,
+      "loss": 0.672,
+      "step": 5926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.668129065696915e-08,
+      "loss": 0.9971,
+      "step": 5927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.642402724384278e-08,
+      "loss": 0.9987,
+      "step": 5928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.616725268444834e-08,
+      "loss": 1.0073,
+      "step": 5929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.591096704370525e-08,
+      "loss": 1.0063,
+      "step": 5930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.565517038640833e-08,
+      "loss": 0.6386,
+      "step": 5931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.53998627772283e-08,
+      "loss": 1.0433,
+      "step": 5932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.514504428071377e-08,
+      "loss": 1.0326,
+      "step": 5933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.489071496128851e-08,
+      "loss": 1.0339,
+      "step": 5934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.463687488325331e-08,
+      "loss": 0.9953,
+      "step": 5935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.438352411078463e-08,
+      "loss": 1.034,
+      "step": 5936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.413066270793588e-08,
+      "loss": 0.9765,
+      "step": 5937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.387829073863594e-08,
+      "loss": 1.0314,
+      "step": 5938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.362640826669064e-08,
+      "loss": 1.0439,
+      "step": 5939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.337501535578193e-08,
+      "loss": 0.9931,
+      "step": 5940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.312411206946766e-08,
+      "loss": 0.97,
+      "step": 5941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.28736984711824e-08,
+      "loss": 1.0225,
+      "step": 5942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.262377462423662e-08,
+      "loss": 1.0195,
+      "step": 5943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.237434059181646e-08,
+      "loss": 1.0149,
+      "step": 5944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.212539643698544e-08,
+      "loss": 0.6494,
+      "step": 5945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.187694222268191e-08,
+      "loss": 1.0425,
+      "step": 5946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.16289780117214e-08,
+      "loss": 0.9847,
+      "step": 5947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.138150386679531e-08,
+      "loss": 1.0293,
+      "step": 5948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.113451985047002e-08,
+      "loss": 0.9911,
+      "step": 5949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.088802602519006e-08,
+      "loss": 0.9813,
+      "step": 5950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0642022453274e-08,
+      "loss": 1.0191,
+      "step": 5951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.039650919691807e-08,
+      "loss": 0.654,
+      "step": 5952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.015148631819334e-08,
+      "loss": 0.9891,
+      "step": 5953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.990695387904753e-08,
+      "loss": 0.9616,
+      "step": 5954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.966291194130457e-08,
+      "loss": 1.0188,
+      "step": 5955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.94193605666633e-08,
+      "loss": 1.0158,
+      "step": 5956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.9176299816699803e-08,
+      "loss": 0.9755,
+      "step": 5957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.893372975286537e-08,
+      "loss": 0.9946,
+      "step": 5958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8691650436487825e-08,
+      "loss": 1.002,
+      "step": 5959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.845006192877e-08,
+      "loss": 1.0022,
+      "step": 5960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.820896429079147e-08,
+      "loss": 0.9803,
+      "step": 5961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.796835758350749e-08,
+      "loss": 1.0267,
+      "step": 5962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.772824186774894e-08,
+      "loss": 0.9739,
+      "step": 5963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.748861720422282e-08,
+      "loss": 0.956,
+      "step": 5964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.724948365351201e-08,
+      "loss": 1.011,
+      "step": 5965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7010841276074804e-08,
+      "loss": 1.0092,
+      "step": 5966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.677269013224628e-08,
+      "loss": 0.9911,
+      "step": 5967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.653503028223583e-08,
+      "loss": 0.9956,
+      "step": 5968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.629786178613028e-08,
+      "loss": 0.6777,
+      "step": 5969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.606118470389121e-08,
+      "loss": 0.9522,
+      "step": 5970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.582499909535565e-08,
+      "loss": 0.9904,
+      "step": 5971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.558930502023762e-08,
+      "loss": 0.9863,
+      "step": 5972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5354102538125444e-08,
+      "loss": 1.0303,
+      "step": 5973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5119391708484676e-08,
+      "loss": 1.0274,
+      "step": 5974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.488517259065539e-08,
+      "loss": 0.6491,
+      "step": 5975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.465144524385357e-08,
+      "loss": 1.0039,
+      "step": 5976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.441820972717104e-08,
+      "loss": 1.0057,
+      "step": 5977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.418546609957531e-08,
+      "loss": 0.9994,
+      "step": 5978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3953214419909743e-08,
+      "loss": 1.02,
+      "step": 5979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3721454746892716e-08,
+      "loss": 1.0172,
+      "step": 5980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.34901871391189e-08,
+      "loss": 0.9946,
+      "step": 5981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.325941165505776e-08,
+      "loss": 1.0141,
+      "step": 5982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3029128353055064e-08,
+      "loss": 1.0084,
+      "step": 5983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2799337291331794e-08,
+      "loss": 1.006,
+      "step": 5984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2570038527984586e-08,
+      "loss": 1.0026,
+      "step": 5985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.234123212098573e-08,
+      "loss": 1.0504,
+      "step": 5986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.211291812818297e-08,
+      "loss": 1.0275,
+      "step": 5987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.188509660729878e-08,
+      "loss": 1.0293,
+      "step": 5988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.1657767615932654e-08,
+      "loss": 1.0447,
+      "step": 5989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.143093121155817e-08,
+      "loss": 1.03,
+      "step": 5990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.1204587451525674e-08,
+      "loss": 1.0257,
+      "step": 5991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.097873639305961e-08,
+      "loss": 0.9919,
+      "step": 5992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.07533780932603e-08,
+      "loss": 1.0006,
+      "step": 5993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.052851260910418e-08,
+      "loss": 0.6706,
+      "step": 5994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.0304139997441984e-08,
+      "loss": 0.995,
+      "step": 5995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.0080260315001236e-08,
+      "loss": 1.0223,
+      "step": 5996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.985687361838331e-08,
+      "loss": 0.9686,
+      "step": 5997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9633979964065706e-08,
+      "loss": 1.0134,
+      "step": 5998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.941157940840157e-08,
+      "loss": 0.9733,
+      "step": 5999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.918967200761881e-08,
+      "loss": 1.0097,
+      "step": 6000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.896825781782077e-08,
+      "loss": 1.0123,
+      "step": 6001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.874733689498645e-08,
+      "loss": 0.9949,
+      "step": 6002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.852690929496961e-08,
+      "loss": 0.6262,
+      "step": 6003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.83069750734999e-08,
+      "loss": 1.0069,
+      "step": 6004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.808753428618129e-08,
+      "loss": 1.0172,
+      "step": 6005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7868586988494494e-08,
+      "loss": 1.0167,
+      "step": 6006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.765013323579348e-08,
+      "loss": 1.0368,
+      "step": 6007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7432173083309604e-08,
+      "loss": 1.0061,
+      "step": 6008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.72147065861479e-08,
+      "loss": 1.0489,
+      "step": 6009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.699773379928862e-08,
+      "loss": 1.0525,
+      "step": 6010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.678125477758832e-08,
+      "loss": 0.9834,
+      "step": 6011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.656526957577722e-08,
+      "loss": 1.0156,
+      "step": 6012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6349778248462315e-08,
+      "loss": 0.6361,
+      "step": 6013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6134780850124676e-08,
+      "loss": 0.978,
+      "step": 6014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.592027743512017e-08,
+      "loss": 0.6527,
+      "step": 6015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.570626805768119e-08,
+      "loss": 1.0418,
+      "step": 6016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5492752771913555e-08,
+      "loss": 0.991,
+      "step": 6017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5279731631799655e-08,
+      "loss": 1.0238,
+      "step": 6018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5067204691195734e-08,
+      "loss": 1.0234,
+      "step": 6019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.485517200383393e-08,
+      "loss": 0.9854,
+      "step": 6020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.464363362332113e-08,
+      "loss": 1.0271,
+      "step": 6021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.443258960313922e-08,
+      "loss": 1.0442,
+      "step": 6022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4222039996644866e-08,
+      "loss": 0.6435,
+      "step": 6023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.401198485707036e-08,
+      "loss": 0.6307,
+      "step": 6024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.380242423752234e-08,
+      "loss": 1.0075,
+      "step": 6025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3593358190983086e-08,
+      "loss": 1.0182,
+      "step": 6026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.338478677030877e-08,
+      "loss": 1.0071,
+      "step": 6027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.317671002823186e-08,
+      "loss": 1.0031,
+      "step": 6028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.29691280173583e-08,
+      "loss": 1.0447,
+      "step": 6029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.276204079017076e-08,
+      "loss": 0.9944,
+      "step": 6030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.255544839902514e-08,
+      "loss": 1.0148,
+      "step": 6031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.234935089615255e-08,
+      "loss": 1.0192,
+      "step": 6032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.214374833366019e-08,
+      "loss": 1.0117,
+      "step": 6033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.193864076352849e-08,
+      "loss": 0.9723,
+      "step": 6034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.173402823761396e-08,
+      "loss": 1.0037,
+      "step": 6035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1529910807647225e-08,
+      "loss": 0.6254,
+      "step": 6036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.132628852523412e-08,
+      "loss": 1.0455,
+      "step": 6037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.112316144185524e-08,
+      "loss": 0.9735,
+      "step": 6038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.092052960886572e-08,
+      "loss": 0.9872,
+      "step": 6039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0718393077495915e-08,
+      "loss": 1.0087,
+      "step": 6040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.051675189885029e-08,
+      "loss": 1.0387,
+      "step": 6041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.031560612390916e-08,
+      "loss": 1.0094,
+      "step": 6042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0114955803526305e-08,
+      "loss": 1.0121,
+      "step": 6043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.991480098843114e-08,
+      "loss": 1.0126,
+      "step": 6044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.971514172922763e-08,
+      "loss": 1.02,
+      "step": 6045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9515978076394286e-08,
+      "loss": 1.0288,
+      "step": 6046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9317310080284384e-08,
+      "loss": 0.9873,
+      "step": 6047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.911913779112597e-08,
+      "loss": 0.6372,
+      "step": 6048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8921461259021404e-08,
+      "loss": 1.0117,
+      "step": 6049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8724280533948275e-08,
+      "loss": 0.9792,
+      "step": 6050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.852759566575825e-08,
+      "loss": 0.9974,
+      "step": 6051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.833140670417845e-08,
+      "loss": 0.9899,
+      "step": 6052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8135713698809635e-08,
+      "loss": 1.0043,
+      "step": 6053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.794051669912734e-08,
+      "loss": 0.9991,
+      "step": 6054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.774581575448277e-08,
+      "loss": 1.0206,
+      "step": 6055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.755161091410031e-08,
+      "loss": 1.041,
+      "step": 6056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7357902227080015e-08,
+      "loss": 0.6361,
+      "step": 6057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.716468974239562e-08,
+      "loss": 1.0192,
+      "step": 6058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.69719735088958e-08,
+      "loss": 1.072,
+      "step": 6059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.677975357530405e-08,
+      "loss": 1.0074,
+      "step": 6060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.658802999021793e-08,
+      "loss": 1.0141,
+      "step": 6061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.639680280210977e-08,
+      "loss": 1.0238,
+      "step": 6062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.620607205932646e-08,
+      "loss": 1.0041,
+      "step": 6063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.601583781008899e-08,
+      "loss": 1.0066,
+      "step": 6064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5826100102493096e-08,
+      "loss": 1.0092,
+      "step": 6065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.563685898450908e-08,
+      "loss": 1.0038,
+      "step": 6066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.544811450398155e-08,
+      "loss": 0.6447,
+      "step": 6067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.525986670862946e-08,
+      "loss": 0.9914,
+      "step": 6068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5072115646046506e-08,
+      "loss": 1.0063,
+      "step": 6069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4884861363700724e-08,
+      "loss": 0.9794,
+      "step": 6070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.469810390893357e-08,
+      "loss": 1.0022,
+      "step": 6071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4511843328963064e-08,
+      "loss": 0.9787,
+      "step": 6072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432607967087908e-08,
+      "loss": 0.9959,
+      "step": 6073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.414081298164784e-08,
+      "loss": 1.0058,
+      "step": 6074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.395604330810897e-08,
+      "loss": 0.9684,
+      "step": 6075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.377177069697623e-08,
+      "loss": 0.9957,
+      "step": 6076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.358799519483857e-08,
+      "loss": 1.0103,
+      "step": 6077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.340471684815838e-08,
+      "loss": 0.9867,
+      "step": 6078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.322193570327325e-08,
+      "loss": 1.0154,
+      "step": 6079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3039651806394233e-08,
+      "loss": 1.0463,
+      "step": 6080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.28578652036069e-08,
+      "loss": 0.9741,
+      "step": 6081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.267657594087159e-08,
+      "loss": 1.0124,
+      "step": 6082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.249578406402209e-08,
+      "loss": 0.6231,
+      "step": 6083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.23154896187674e-08,
+      "loss": 1.0019,
+      "step": 6084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2135692650689716e-08,
+      "loss": 1.0126,
+      "step": 6085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.195639320524601e-08,
+      "loss": 1.0038,
+      "step": 6086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1777591327767803e-08,
+      "loss": 1.0134,
+      "step": 6087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1599287063460045e-08,
+      "loss": 0.9867,
+      "step": 6088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.142148045740245e-08,
+      "loss": 1.001,
+      "step": 6089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1244171554548834e-08,
+      "loss": 0.9973,
+      "step": 6090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.10673603997269e-08,
+      "loss": 1.0424,
+      "step": 6091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.089104703763867e-08,
+      "loss": 1.0098,
+      "step": 6092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.071523151286026e-08,
+      "loss": 0.9923,
+      "step": 6093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0539913869842336e-08,
+      "loss": 1.014,
+      "step": 6094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0365094152909e-08,
+      "loss": 0.9753,
+      "step": 6095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.019077240625911e-08,
+      "loss": 0.9508,
+      "step": 6096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0016948673965203e-08,
+      "loss": 1.0107,
+      "step": 6097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9843622999973896e-08,
+      "loss": 1.0327,
+      "step": 6098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9670795428106134e-08,
+      "loss": 1.0254,
+      "step": 6099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9498466002056742e-08,
+      "loss": 1.0321,
+      "step": 6100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9326634765394874e-08,
+      "loss": 1.0409,
+      "step": 6101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9155301761563113e-08,
+      "loss": 1.0218,
+      "step": 6102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.898446703387902e-08,
+      "loss": 1.0167,
+      "step": 6103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8814130625533174e-08,
+      "loss": 1.0129,
+      "step": 6104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.864429257959111e-08,
+      "loss": 0.9849,
+      "step": 6105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.847495293899138e-08,
+      "loss": 1.0216,
+      "step": 6106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8306111746547732e-08,
+      "loss": 0.9568,
+      "step": 6107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.813776904494647e-08,
+      "loss": 1.0169,
+      "step": 6108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7969924876749097e-08,
+      "loss": 1.0023,
+      "step": 6109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7802579284390338e-08,
+      "loss": 1.0201,
+      "step": 6110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.763573231017946e-08,
+      "loss": 0.9862,
+      "step": 6111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7469383996298723e-08,
+      "loss": 0.6633,
+      "step": 6112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7303534384805593e-08,
+      "loss": 0.6272,
+      "step": 6113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.713818351763053e-08,
+      "loss": 1.0108,
+      "step": 6114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6973331436577873e-08,
+      "loss": 1.0171,
+      "step": 6115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6808978183326502e-08,
+      "loss": 0.6069,
+      "step": 6116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6645123799428294e-08,
+      "loss": 1.0138,
+      "step": 6117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6481768326310106e-08,
+      "loss": 1.0219,
+      "step": 6118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.631891180527179e-08,
+      "loss": 0.9786,
+      "step": 6119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.615655427748731e-08,
+      "loss": 1.0253,
+      "step": 6120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5994695784004707e-08,
+      "loss": 1.0186,
+      "step": 6121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.583333636574503e-08,
+      "loss": 1.0086,
+      "step": 6122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5672476063504756e-08,
+      "loss": 1.0168,
+      "step": 6123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5512114917952244e-08,
+      "loss": 0.9986,
+      "step": 6124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5352252969630837e-08,
+      "loss": 1.0046,
+      "step": 6125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5192890258957544e-08,
+      "loss": 0.9799,
+      "step": 6126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5034026826223243e-08,
+      "loss": 0.9738,
+      "step": 6127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4875662711591805e-08,
+      "loss": 1.05,
+      "step": 6128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4717797955102094e-08,
+      "loss": 0.994,
+      "step": 6129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4560432596665293e-08,
+      "loss": 1.0279,
+      "step": 6130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.440356667606802e-08,
+      "loss": 1.0103,
+      "step": 6131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4247200232968557e-08,
+      "loss": 1.0362,
+      "step": 6132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.409133330690105e-08,
+      "loss": 1.016,
+      "step": 6133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3935965937271763e-08,
+      "loss": 0.9982,
+      "step": 6134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3781098163361713e-08,
+      "loss": 1.0384,
+      "step": 6135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3626730024324692e-08,
+      "loss": 1.0001,
+      "step": 6136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3472861559188596e-08,
+      "loss": 1.054,
+      "step": 6137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.331949280685519e-08,
+      "loss": 1.0087,
+      "step": 6138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3166623806099684e-08,
+      "loss": 0.9769,
+      "step": 6139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.301425459557138e-08,
+      "loss": 1.0317,
+      "step": 6140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2862385213792135e-08,
+      "loss": 0.9968,
+      "step": 6141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.271101569915834e-08,
+      "loss": 0.958,
+      "step": 6142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2560146089940058e-08,
+      "loss": 0.9936,
+      "step": 6143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2409776424280324e-08,
+      "loss": 1.0297,
+      "step": 6144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.225990674019651e-08,
+      "loss": 1.0188,
+      "step": 6145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2110537075579195e-08,
+      "loss": 1.0075,
+      "step": 6146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.196166746819217e-08,
+      "loss": 0.9897,
+      "step": 6147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1813297955673327e-08,
+      "loss": 1.0077,
+      "step": 6148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1665428575534218e-08,
+      "loss": 0.9904,
+      "step": 6149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1518059365159603e-08,
+      "loss": 1.0222,
+      "step": 6150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.13711903618079e-08,
+      "loss": 0.9897,
+      "step": 6151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1224821602610965e-08,
+      "loss": 1.0025,
+      "step": 6152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1078953124574526e-08,
+      "loss": 1.0244,
+      "step": 6153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.093358496457709e-08,
+      "loss": 0.9938,
+      "step": 6154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0788717159371917e-08,
+      "loss": 0.9642,
+      "step": 6155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064434974558438e-08,
+      "loss": 0.9944,
+      "step": 6156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.050048275971439e-08,
+      "loss": 1.002,
+      "step": 6157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.035711623813463e-08,
+      "loss": 1.0051,
+      "step": 6158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0214250217091443e-08,
+      "loss": 1.0072,
+      "step": 6159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.007188473270527e-08,
+      "loss": 0.9489,
+      "step": 6160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.993001982096887e-08,
+      "loss": 1.0176,
+      "step": 6161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9788655517749553e-08,
+      "loss": 0.9952,
+      "step": 6162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9647791858787178e-08,
+      "loss": 0.9878,
+      "step": 6163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9507428879695697e-08,
+      "loss": 0.6265,
+      "step": 6164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.936756661596184e-08,
+      "loss": 0.6343,
+      "step": 6165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9228205102946427e-08,
+      "loss": 1.0293,
+      "step": 6166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9089344375883275e-08,
+      "loss": 1.0418,
+      "step": 6167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.895098446987964e-08,
+      "loss": 1.0121,
+      "step": 6168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8813125419915976e-08,
+      "loss": 1.0335,
+      "step": 6169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.867576726084663e-08,
+      "loss": 1.0057,
+      "step": 6170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8538910027398713e-08,
+      "loss": 1.0495,
+      "step": 6171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8402553754173433e-08,
+      "loss": 1.0092,
+      "step": 6172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8266698475644328e-08,
+      "loss": 0.97,
+      "step": 6173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8131344226158805e-08,
+      "loss": 0.9743,
+      "step": 6174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7996491039938165e-08,
+      "loss": 1.0013,
+      "step": 6175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.786213895107602e-08,
+      "loss": 1.0084,
+      "step": 6176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.772828799354009e-08,
+      "loss": 1.002,
+      "step": 6177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.759493820117064e-08,
+      "loss": 0.9843,
+      "step": 6178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.746208960768225e-08,
+      "loss": 0.9887,
+      "step": 6179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7329742246661617e-08,
+      "loss": 1.0152,
+      "step": 6180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7197896151569524e-08,
+      "loss": 1.0167,
+      "step": 6181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7066551355739755e-08,
+      "loss": 1.0619,
+      "step": 6182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6935707892379302e-08,
+      "loss": 1.0385,
+      "step": 6183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6805365794568816e-08,
+      "loss": 0.9976,
+      "step": 6184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6675525095261488e-08,
+      "loss": 1.0122,
+      "step": 6185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6546185827284176e-08,
+      "loss": 1.0186,
+      "step": 6186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.641734802333694e-08,
+      "loss": 0.9825,
+      "step": 6187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.628901171599306e-08,
+      "loss": 0.9867,
+      "step": 6188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6161176937699028e-08,
+      "loss": 1.0163,
+      "step": 6189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6033843720774543e-08,
+      "loss": 1.0085,
+      "step": 6190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907012097412298e-08,
+      "loss": 1.0134,
+      "step": 6191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5780682099678422e-08,
+      "loss": 1.0235,
+      "step": 6192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5654853759512034e-08,
+      "loss": 0.9728,
+      "step": 6193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.552952710872568e-08,
+      "loss": 1.0234,
+      "step": 6194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5404702179004914e-08,
+      "loss": 1.0078,
+      "step": 6195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5280379001908038e-08,
+      "loss": 1.0361,
+      "step": 6196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.515655760886725e-08,
+      "loss": 1.0025,
+      "step": 6197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.50332380311875e-08,
+      "loss": 1.0221,
+      "step": 6198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4910420300046966e-08,
+      "loss": 1.0227,
+      "step": 6199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4788104446496363e-08,
+      "loss": 1.027,
+      "step": 6200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4666290501460732e-08,
+      "loss": 0.9889,
+      "step": 6201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4544978495737436e-08,
+      "loss": 1.042,
+      "step": 6202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4424168459996389e-08,
+      "loss": 1.0148,
+      "step": 6203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4303860424781822e-08,
+      "loss": 0.9896,
+      "step": 6204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.418405442050985e-08,
+      "loss": 0.9898,
+      "step": 6205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4064750477471133e-08,
+      "loss": 0.9696,
+      "step": 6206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3945948625827986e-08,
+      "loss": 1.0116,
+      "step": 6207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3827648895616383e-08,
+      "loss": 1.007,
+      "step": 6208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3709851316745291e-08,
+      "loss": 0.6378,
+      "step": 6209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3592555918996661e-08,
+      "loss": 1.0298,
+      "step": 6210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3475762732025664e-08,
+      "loss": 1.0441,
+      "step": 6211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3359471785360454e-08,
+      "loss": 0.9795,
+      "step": 6212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3243683108401959e-08,
+      "loss": 1.0119,
+      "step": 6213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3128396730424318e-08,
+      "loss": 0.9953,
+      "step": 6214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.301361268057466e-08,
+      "loss": 1.0114,
+      "step": 6215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2899330987873325e-08,
+      "loss": 1.0054,
+      "step": 6216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2785551681213202e-08,
+      "loss": 0.9963,
+      "step": 6217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2672274789360172e-08,
+      "loss": 1.0168,
+      "step": 6218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2559500340953988e-08,
+      "loss": 0.6482,
+      "step": 6219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2447228364505846e-08,
+      "loss": 0.9826,
+      "step": 6220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2335458888401707e-08,
+      "loss": 1.0461,
+      "step": 6221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.222419194089852e-08,
+      "loss": 1.0017,
+      "step": 6222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2113427550128008e-08,
+      "loss": 1.009,
+      "step": 6223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2003165744093769e-08,
+      "loss": 0.9567,
+      "step": 6224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1893406550672391e-08,
+      "loss": 1.0163,
+      "step": 6225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.178414999761368e-08,
+      "loss": 1.0262,
+      "step": 6226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1675396112540426e-08,
+      "loss": 0.9727,
+      "step": 6227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.156714492294819e-08,
+      "loss": 1.0127,
+      "step": 6228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1459396456205527e-08,
+      "loss": 1.0216,
+      "step": 6229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1352150739553312e-08,
+      "loss": 1.0059,
+      "step": 6230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1245407800106078e-08,
+      "loss": 1.0028,
+      "step": 6231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1139167664851124e-08,
+      "loss": 0.9474,
+      "step": 6232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1033430360648522e-08,
+      "loss": 1.0093,
+      "step": 6233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0928195914230886e-08,
+      "loss": 0.9946,
+      "step": 6234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0823464352203826e-08,
+      "loss": 1.0072,
+      "step": 6235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.071923570104638e-08,
+      "loss": 1.0335,
+      "step": 6236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0615509987109694e-08,
+      "loss": 0.9929,
+      "step": 6237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0512287236618345e-08,
+      "loss": 1.0006,
+      "step": 6238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0409567475669456e-08,
+      "loss": 1.0117,
+      "step": 6239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.03073507302327e-08,
+      "loss": 0.9747,
+      "step": 6240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0205637026151403e-08,
+      "loss": 1.0377,
+      "step": 6241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0104426389140552e-08,
+      "loss": 0.9992,
+      "step": 6242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0003718844789232e-08,
+      "loss": 1.0014,
+      "step": 6243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.90351441855819e-09,
+      "loss": 1.0016,
+      "step": 6244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.803813135781825e-09,
+      "loss": 1.0235,
+      "step": 6245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.704615021666862e-09,
+      "loss": 1.0434,
+      "step": 6246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.605920101292797e-09,
+      "loss": 0.981,
+      "step": 6247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.507728399612114e-09,
+      "loss": 0.9907,
+      "step": 6248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.410039941450066e-09,
+      "loss": 0.9474,
+      "step": 6249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.312854751504451e-09,
+      "loss": 1.027,
+      "step": 6250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.216172854346283e-09,
+      "loss": 0.9958,
+      "step": 6251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.119994274418675e-09,
+      "loss": 1.0288,
+      "step": 6252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.024319036038398e-09,
+      "loss": 1.0233,
+      "step": 6253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.929147163393657e-09,
+      "loss": 0.9752,
+      "step": 6254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.834478680546542e-09,
+      "loss": 1.012,
+      "step": 6255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.740313611431682e-09,
+      "loss": 1.0227,
+      "step": 6256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.646651979855812e-09,
+      "loss": 0.9925,
+      "step": 6257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.553493809499323e-09,
+      "loss": 1.0013,
+      "step": 6258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.460839123914265e-09,
+      "loss": 1.0025,
+      "step": 6259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.36868794652612e-09,
+      "loss": 0.9569,
+      "step": 6260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.277040300633142e-09,
+      "loss": 1.0381,
+      "step": 6261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.185896209405685e-09,
+      "loss": 0.6429,
+      "step": 6262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.095255695887315e-09,
+      "loss": 0.9521,
+      "step": 6263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.005118782993925e-09,
+      "loss": 1.0172,
+      "step": 6264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.915485493514617e-09,
+      "loss": 0.974,
+      "step": 6265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.826355850110378e-09,
+      "loss": 1.0365,
+      "step": 6266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.737729875315625e-09,
+      "loss": 1.0167,
+      "step": 6267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.649607591536655e-09,
+      "loss": 1.0085,
+      "step": 6268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.561989021053427e-09,
+      "loss": 0.9835,
+      "step": 6269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.474874186017554e-09,
+      "loss": 1.0087,
+      "step": 6270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.3882631084540845e-09,
+      "loss": 1.0239,
+      "step": 6271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.302155810260169e-09,
+      "loss": 1.0139,
+      "step": 6272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.216552313205726e-09,
+      "loss": 0.6444,
+      "step": 6273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.1314526389332216e-09,
+      "loss": 1.0063,
+      "step": 6274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.046856808958113e-09,
+      "loss": 0.9971,
+      "step": 6275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.962764844668179e-09,
+      "loss": 1.0146,
+      "step": 6276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.879176767323747e-09,
+      "loss": 1.0138,
+      "step": 6277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.796092598058134e-09,
+      "loss": 0.9964,
+      "step": 6278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.713512357876316e-09,
+      "loss": 0.9903,
+      "step": 6279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.6314360676573696e-09,
+      "loss": 1.0455,
+      "step": 6280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.549863748151585e-09,
+      "loss": 0.9885,
+      "step": 6281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.468795419982687e-09,
+      "loss": 0.9946,
+      "step": 6282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.3882311036465016e-09,
+      "loss": 0.9966,
+      "step": 6283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.308170819511405e-09,
+      "loss": 0.6464,
+      "step": 6284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.228614587818981e-09,
+      "loss": 1.0202,
+      "step": 6285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.149562428682475e-09,
+      "loss": 1.0309,
+      "step": 6286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.0710143620883445e-09,
+      "loss": 0.9773,
+      "step": 6287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.99297040789537e-09,
+      "loss": 0.6617,
+      "step": 6288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9154305858351014e-09,
+      "loss": 1.017,
+      "step": 6289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.838394915511191e-09,
+      "loss": 0.9779,
+      "step": 6290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.761863416400059e-09,
+      "loss": 0.9993,
+      "step": 6291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.685836107850894e-09,
+      "loss": 1.0253,
+      "step": 6292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.610313009084988e-09,
+      "loss": 0.9633,
+      "step": 6293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.535294139196622e-09,
+      "loss": 0.6471,
+      "step": 6294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.460779517152181e-09,
+      "loss": 1.0043,
+      "step": 6295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3867691617903724e-09,
+      "loss": 0.9815,
+      "step": 6296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.313263091823561e-09,
+      "loss": 1.026,
+      "step": 6297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2402613258351044e-09,
+      "loss": 0.6737,
+      "step": 6298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.167763882281795e-09,
+      "loss": 0.6327,
+      "step": 6299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0957707794929695e-09,
+      "loss": 0.9735,
+      "step": 6300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.024282035669847e-09,
+      "loss": 1.0184,
+      "step": 6301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9532976688866356e-09,
+      "loss": 1.0016,
+      "step": 6302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.882817697089647e-09,
+      "loss": 0.9961,
+      "step": 6303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.812842138098183e-09,
+      "loss": 1.0343,
+      "step": 6304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.743371009603647e-09,
+      "loss": 0.9697,
+      "step": 6305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6744043291697675e-09,
+      "loss": 1.0128,
+      "step": 6306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6059421142332635e-09,
+      "loss": 0.9697,
+      "step": 6307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.537984382102733e-09,
+      "loss": 1.0001,
+      "step": 6308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.470531149959766e-09,
+      "loss": 1.003,
+      "step": 6309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.403582434857833e-09,
+      "loss": 0.9953,
+      "step": 6310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3371382537233935e-09,
+      "loss": 0.6411,
+      "step": 6311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.27119862335501e-09,
+      "loss": 0.9746,
+      "step": 6312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.205763560423791e-09,
+      "loss": 1.0572,
+      "step": 6313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.140833081473172e-09,
+      "loss": 0.6323,
+      "step": 6314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.076407202919352e-09,
+      "loss": 0.6225,
+      "step": 6315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0124859410504145e-09,
+      "loss": 0.9997,
+      "step": 6316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.949069312027431e-09,
+      "loss": 1.0157,
+      "step": 6317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8861573318833555e-09,
+      "loss": 1.0,
+      "step": 6318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.823750016524352e-09,
+      "loss": 0.652,
+      "step": 6319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7618473817278005e-09,
+      "loss": 1.0578,
+      "step": 6320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7004494431445155e-09,
+      "loss": 1.011,
+      "step": 6321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6395562162971904e-09,
+      "loss": 0.6119,
+      "step": 6322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5791677165812885e-09,
+      "loss": 1.028,
+      "step": 6323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5192839592643744e-09,
+      "loss": 0.9931,
+      "step": 6324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.459904959486337e-09,
+      "loss": 1.0534,
+      "step": 6325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4010307322596133e-09,
+      "loss": 0.9455,
+      "step": 6326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3426612924691845e-09,
+      "loss": 1.028,
+      "step": 6327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.284796654872135e-09,
+      "loss": 0.6572,
+      "step": 6328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.227436834097874e-09,
+      "loss": 0.9972,
+      "step": 6329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.170581844648579e-09,
+      "loss": 1.0074,
+      "step": 6330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1142317008983066e-09,
+      "loss": 0.9942,
+      "step": 6331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0583864170936612e-09,
+      "loss": 1.004,
+      "step": 6332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.003046007354015e-09,
+      "loss": 1.0339,
+      "step": 6333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9482104856703994e-09,
+      "loss": 0.9918,
+      "step": 6334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8938798659066123e-09,
+      "loss": 1.0128,
+      "step": 6335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8400541617987773e-09,
+      "loss": 0.9866,
+      "step": 6336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.786733386955342e-09,
+      "loss": 1.0023,
+      "step": 6337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.733917554856857e-09,
+      "loss": 0.9753,
+      "step": 6338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6816066788566407e-09,
+      "loss": 0.6238,
+      "step": 6339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.629800772179891e-09,
+      "loss": 1.0524,
+      "step": 6340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.578499847924576e-09,
+      "loss": 1.0426,
+      "step": 6341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.527703919060764e-09,
+      "loss": 1.0345,
+      "step": 6342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.47741299843085e-09,
+      "loss": 0.9633,
+      "step": 6343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4276270987493295e-09,
+      "loss": 1.0519,
+      "step": 6344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3783462326034676e-09,
+      "loss": 0.9801,
+      "step": 6345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3295704124528525e-09,
+      "loss": 1.0446,
+      "step": 6346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.281299650628732e-09,
+      "loss": 1.0486,
+      "step": 6347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.233533959335343e-09,
+      "loss": 1.0352,
+      "step": 6348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.186273350648804e-09,
+      "loss": 0.9883,
+      "step": 6349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.139517836518001e-09,
+      "loss": 1.0137,
+      "step": 6350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.093267428763701e-09,
+      "loss": 0.9798,
+      "step": 6351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0475221390789944e-09,
+      "loss": 0.987,
+      "step": 6352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022819790292965e-09,
+      "loss": 0.9917,
+      "step": 6353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9575469600527917e-09,
+      "loss": 0.9818,
+      "step": 6354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.913317093459099e-09,
+      "loss": 0.6598,
+      "step": 6355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8695923904310517e-09,
+      "loss": 0.9854,
+      "step": 6356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.826372862022696e-09,
+      "loss": 1.0064,
+      "step": 6357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7836585191615127e-09,
+      "loss": 0.9891,
+      "step": 6358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7414493726464197e-09,
+      "loss": 0.9788,
+      "step": 6359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6997454331488803e-09,
+      "loss": 1.0307,
+      "step": 6360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6585467112126828e-09,
+      "loss": 1.0445,
+      "step": 6361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6178532172537174e-09,
+      "loss": 1.0004,
+      "step": 6362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.577664961560643e-09,
+      "loss": 1.0081,
+      "step": 6363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5379819542935547e-09,
+      "loss": 1.0216,
+      "step": 6364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4988042054855377e-09,
+      "loss": 1.0007,
+      "step": 6365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4601317250413358e-09,
+      "loss": 0.9758,
+      "step": 6366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4219645227386833e-09,
+      "loss": 1.0293,
+      "step": 6367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3843026082269726e-09,
+      "loss": 1.0262,
+      "step": 6368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3471459910279204e-09,
+      "loss": 1.0182,
+      "step": 6369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3104946805357898e-09,
+      "loss": 1.0146,
+      "step": 6370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.274348686016502e-09,
+      "loss": 1.0117,
+      "step": 6371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2387080166091913e-09,
+      "loss": 1.0074,
+      "step": 6372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2035726813244274e-09,
+      "loss": 1.026,
+      "step": 6373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1689426890451049e-09,
+      "loss": 1.0026,
+      "step": 6374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1348180485264425e-09,
+      "loss": 1.0182,
+      "step": 6375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1011987683964274e-09,
+      "loss": 0.996,
+      "step": 6376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0680848571542611e-09,
+      "loss": 1.0413,
+      "step": 6377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0354763231723574e-09,
+      "loss": 1.0573,
+      "step": 6378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0033731746945662e-09,
+      "loss": 0.992,
+      "step": 6379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.717754198375062e-10,
+      "loss": 1.0357,
+      "step": 6380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.40683066589898e-10,
+      "loss": 0.9884,
+      "step": 6381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.100961228125648e-10,
+      "loss": 1.0205,
+      "step": 6382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.80014596238432e-10,
+      "loss": 1.006,
+      "step": 6383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.504384944731935e-10,
+      "loss": 0.9851,
+      "step": 6384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.213678249939793e-10,
+      "loss": 1.0251,
+      "step": 6385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.928025951509099e-10,
+      "loss": 0.6589,
+      "step": 6386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.647428121655419e-10,
+      "loss": 0.6651,
+      "step": 6387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.371884831324226e-10,
+      "loss": 0.9832,
+      "step": 6388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.101396150179795e-10,
+      "loss": 1.0092,
+      "step": 6389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.83596214660298e-10,
+      "loss": 1.0083,
+      "step": 6390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.575582887706765e-10,
+      "loss": 1.0056,
+      "step": 6391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.320258439320714e-10,
+      "loss": 1.0079,
+      "step": 6392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.069988865993191e-10,
+      "loss": 0.9903,
+      "step": 6393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.824774231002472e-10,
+      "loss": 0.6272,
+      "step": 6394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.584614596343406e-10,
+      "loss": 1.0244,
+      "step": 6395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.349510022731874e-10,
+      "loss": 0.9994,
+      "step": 6396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.119460569611434e-10,
+      "loss": 1.0133,
+      "step": 6397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.89446629514001e-10,
+      "loss": 1.053,
+      "step": 6398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.674527256203209e-10,
+      "loss": 1.0263,
+      "step": 6399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.459643508409883e-10,
+      "loss": 1.0388,
+      "step": 6400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.249815106083243e-10,
+      "loss": 1.0493,
+      "step": 6401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0450421022741877e-10,
+      "loss": 1.0543,
+      "step": 6402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8453245487568565e-10,
+      "loss": 1.0061,
+      "step": 6403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6506624960197517e-10,
+      "loss": 0.954,
+      "step": 6404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4610559932812813e-10,
+      "loss": 0.9952,
+      "step": 6405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2765050884786536e-10,
+      "loss": 1.009,
+      "step": 6406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.097009828270103e-10,
+      "loss": 0.9975,
+      "step": 6407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.922570258034884e-10,
+      "loss": 1.0189,
+      "step": 6408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7531864218777177e-10,
+      "loss": 0.6131,
+      "step": 6409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.588858362622126e-10,
+      "loss": 1.013,
+      "step": 6410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4295861218126547e-10,
+      "loss": 1.0394,
+      "step": 6411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2753697397170923e-10,
+      "loss": 0.984,
+      "step": 6412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1262092553286926e-10,
+      "loss": 1.0552,
+      "step": 6413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9821047063528495e-10,
+      "loss": 1.0582,
+      "step": 6414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8430561292293034e-10,
+      "loss": 0.9999,
+      "step": 6415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.709063559107715e-10,
+      "loss": 1.0043,
+      "step": 6416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.58012702986543e-10,
+      "loss": 0.9995,
+      "step": 6417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4562465741030371e-10,
+      "loss": 1.0544,
+      "step": 6418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3374222231377075e-10,
+      "loss": 1.0297,
+      "step": 6419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.223654007012076e-10,
+      "loss": 1.0102,
+      "step": 6420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1149419544920213e-10,
+      "loss": 1.0003,
+      "step": 6421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0112860930577837e-10,
+      "loss": 1.0049,
+      "step": 6422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.126864489195085e-11,
+      "loss": 1.0008,
+      "step": 6423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.191430470039229e-11,
+      "loss": 1.0139,
+      "step": 6424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.306559109609978e-11,
+      "loss": 1.0312,
+      "step": 6425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.472250631617271e-11,
+      "loss": 1.028,
+      "step": 6426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.688505247025688e-11,
+      "loss": 0.9879,
+      "step": 6427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9553231539434247e-11,
+      "loss": 1.041,
+      "step": 6428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2727045377777274e-11,
+      "loss": 1.0081,
+      "step": 6429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6406495710794574e-11,
+      "loss": 1.0112,
+      "step": 6430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.059158413654117e-11,
+      "loss": 1.0033,
+      "step": 6431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5282312125174398e-11,
+      "loss": 1.0127,
+      "step": 6432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0478681019175935e-11,
+      "loss": 1.0287,
+      "step": 6433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6180692032685685e-11,
+      "loss": 0.9854,
+      "step": 6434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2388346252611982e-11,
+      "loss": 1.0196,
+      "step": 6435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.101644637743433e-12,
+      "loss": 0.6394,
+      "step": 6436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.320588018748907e-12,
+      "loss": 0.9949,
+      "step": 6437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.045177099065711e-12,
+      "loss": 0.9689,
+      "step": 6438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2754124537893714e-12,
+      "loss": 1.0018,
+      "step": 6439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.011294530339768e-12,
+      "loss": 1.0222,
+      "step": 6440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5282364868317584e-13,
+      "loss": 0.665,
+      "step": 6441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7916,
+      "step": 6442
+    },
+    {
+      "epoch": 1.0,
+      "step": 6442,
+      "total_flos": 2.880889933445202e+18,
+      "train_loss": 1.0221481301776374,
+      "train_runtime": 183852.5257,
+      "train_samples_per_second": 8.97,
+      "train_steps_per_second": 0.035
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6442,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 2.880889933445202e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}