diff --git "a/exgra-med-gpt-medrag-only/trainer_state.json" "b/exgra-med-gpt-medrag-only/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/exgra-med-gpt-medrag-only/trainer_state.json"
@@ -0,0 +1,8113 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "global_step": 1348,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 5.855,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 5.8735,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 4.3488,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 3.2056,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.2717,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.8925,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 0.5534,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.6385,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 0.4929,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.3698,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 0.1405,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.268,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 0.1497,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.4364,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.3558,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.2374,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 0.1175,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.1637,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 0.0784,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.2103,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 0.1701,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.0682,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1219512195121953e-05,
+      "loss": 0.2193,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.175,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.0853,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.1163,
+      "step": 26
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3170731707317076e-05,
+      "loss": 0.1191,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.1116,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4146341463414635e-05,
+      "loss": 0.0768,
+      "step": 29
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.0739,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 0.152,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.1381,
+      "step": 32
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 0.0836,
+      "step": 33
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.0781,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 0.0648,
+      "step": 35
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.0673,
+      "step": 36
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 0.0796,
+      "step": 37
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.0609,
+      "step": 38
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 0.0598,
+      "step": 39
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.0692,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 41
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99999711119408e-05,
+      "loss": 0.0811,
+      "step": 42
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999884447930092e-05,
+      "loss": 0.0576,
+      "step": 43
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999740008468595e-05,
+      "loss": 0.0691,
+      "step": 44
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999953779439082e-05,
+      "loss": 0.0659,
+      "step": 45
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9999277806865083e-05,
+      "loss": 0.0609,
+      "step": 46
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998960047393496e-05,
+      "loss": 0.0636,
+      "step": 47
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999858451781194e-05,
+      "loss": 0.0506,
+      "step": 48
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999815122029008e-05,
+      "loss": 0.067,
+      "step": 49
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997660157331356e-05,
+      "loss": 0.0513,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999711133177292e-05,
+      "loss": 0.0887,
+      "step": 51
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996504746785685e-05,
+      "loss": 0.0576,
+      "step": 52
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995840405874257e-05,
+      "loss": 0.0475,
+      "step": 53
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995118312876946e-05,
+      "loss": 0.0496,
+      "step": 54
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9994338471965718e-05,
+      "loss": 0.0475,
+      "step": 55
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9993500887646196e-05,
+      "loss": 0.0487,
+      "step": 56
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9992605564757618e-05,
+      "loss": 0.0625,
+      "step": 57
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9991652508472807e-05,
+      "loss": 0.0596,
+      "step": 58
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9990641724298158e-05,
+      "loss": 0.0509,
+      "step": 59
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9989573218073584e-05,
+      "loss": 0.0997,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9988446995972502e-05,
+      "loss": 0.109,
+      "step": 61
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998726306450179e-05,
+      "loss": 0.0644,
+      "step": 62
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998602143050174e-05,
+      "loss": 0.1192,
+      "step": 63
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998472210114603e-05,
+      "loss": 0.0763,
+      "step": 64
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9983365083941683e-05,
+      "loss": 0.0446,
+      "step": 65
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998195038672902e-05,
+      "loss": 0.0681,
+      "step": 66
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9980478017681605e-05,
+      "loss": 0.0583,
+      "step": 67
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9978947985306222e-05,
+      "loss": 0.0551,
+      "step": 68
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9977360298442803e-05,
+      "loss": 0.0396,
+      "step": 69
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9975714966264386e-05,
+      "loss": 0.0572,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9974011998277055e-05,
+      "loss": 0.0649,
+      "step": 71
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997225140431991e-05,
+      "loss": 0.0646,
+      "step": 72
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997043319456497e-05,
+      "loss": 0.0531,
+      "step": 73
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968557379517154e-05,
+      "loss": 0.0582,
+      "step": 74
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9966623970014182e-05,
+      "loss": 0.0535,
+      "step": 75
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996463297722655e-05,
+      "loss": 0.0549,
+      "step": 76
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962584412657444e-05,
+      "loss": 0.0582,
+      "step": 77
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996047828814267e-05,
+      "loss": 0.0624,
+      "step": 78
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99583146158506e-05,
+      "loss": 0.0657,
+      "step": 79
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9956093408282093e-05,
+      "loss": 0.0544,
+      "step": 80
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995381467827042e-05,
+      "loss": 0.0541,
+      "step": 81
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9951478438981208e-05,
+      "loss": 0.062,
+      "step": 82
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9949084703912333e-05,
+      "loss": 0.0601,
+      "step": 83
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9946633486893867e-05,
+      "loss": 0.061,
+      "step": 84
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9944124802087993e-05,
+      "loss": 0.0398,
+      "step": 85
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9941558663988918e-05,
+      "loss": 0.0647,
+      "step": 86
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993893508742279e-05,
+      "loss": 0.064,
+      "step": 87
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993625408754762e-05,
+      "loss": 0.0482,
+      "step": 88
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9933515679853183e-05,
+      "loss": 0.0553,
+      "step": 89
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9930719880160932e-05,
+      "loss": 0.1394,
+      "step": 90
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9927866704623913e-05,
+      "loss": 0.0508,
+      "step": 91
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9924956169726672e-05,
+      "loss": 0.0454,
+      "step": 92
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9921988292285145e-05,
+      "loss": 0.0459,
+      "step": 93
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991896308944658e-05,
+      "loss": 0.0429,
+      "step": 94
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991588057868942e-05,
+      "loss": 0.0587,
+      "step": 95
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991274077782322e-05,
+      "loss": 0.05,
+      "step": 96
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9909543704988528e-05,
+      "loss": 0.0512,
+      "step": 97
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990628937865679e-05,
+      "loss": 0.0452,
+      "step": 98
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990297781763024e-05,
+      "loss": 0.0478,
+      "step": 99
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899609041041795e-05,
+      "loss": 0.0546,
+      "step": 100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9896183068354937e-05,
+      "loss": 0.0583,
+      "step": 101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9892699919363605e-05,
+      "loss": 0.0472,
+      "step": 102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988915961419208e-05,
+      "loss": 0.0483,
+      "step": 103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9885562173294878e-05,
+      "loss": 0.0528,
+      "step": 104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988190761745661e-05,
+      "loss": 0.0407,
+      "step": 105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9878195967791885e-05,
+      "loss": 0.0497,
+      "step": 106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.987442724574517e-05,
+      "loss": 0.0495,
+      "step": 107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.987060147309068e-05,
+      "loss": 0.0434,
+      "step": 108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9866718671932247e-05,
+      "loss": 0.0362,
+      "step": 109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9862778864703185e-05,
+      "loss": 0.0541,
+      "step": 110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9858782074166175e-05,
+      "loss": 0.0612,
+      "step": 111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9854728323413112e-05,
+      "loss": 0.0403,
+      "step": 112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9850617635865007e-05,
+      "loss": 0.044,
+      "step": 113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984645003527181e-05,
+      "loss": 0.0268,
+      "step": 114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9842225545712302e-05,
+      "loss": 0.0465,
+      "step": 115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9837944191593942e-05,
+      "loss": 0.031,
+      "step": 116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9833605997652732e-05,
+      "loss": 0.051,
+      "step": 117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9829210988953074e-05,
+      "loss": 0.0445,
+      "step": 118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.982475919088762e-05,
+      "loss": 0.048,
+      "step": 119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9820250629177134e-05,
+      "loss": 0.0499,
+      "step": 120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9815685329870337e-05,
+      "loss": 0.0386,
+      "step": 121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9811063319343753e-05,
+      "loss": 0.0552,
+      "step": 122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9806384624301565e-05,
+      "loss": 0.0427,
+      "step": 123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.980164927177546e-05,
+      "loss": 0.0423,
+      "step": 124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.979685728912446e-05,
+      "loss": 0.0465,
+      "step": 125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.979200870403479e-05,
+      "loss": 0.0454,
+      "step": 126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9787103544519684e-05,
+      "loss": 0.0576,
+      "step": 127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9782141838919253e-05,
+      "loss": 0.0564,
+      "step": 128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.977712361590031e-05,
+      "loss": 0.0376,
+      "step": 129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9772048904456192e-05,
+      "loss": 0.0444,
+      "step": 130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.976691773390662e-05,
+      "loss": 0.0457,
+      "step": 131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9761730133897497e-05,
+      "loss": 0.0432,
+      "step": 132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9756486134400772e-05,
+      "loss": 0.0419,
+      "step": 133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9751185765714234e-05,
+      "loss": 0.0394,
+      "step": 134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9745829058461353e-05,
+      "loss": 0.0489,
+      "step": 135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9740416043591106e-05,
+      "loss": 0.0409,
+      "step": 136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9734946752377794e-05,
+      "loss": 0.055,
+      "step": 137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9729421216420857e-05,
+      "loss": 0.0408,
+      "step": 138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9723839467644697e-05,
+      "loss": 0.0378,
+      "step": 139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.971820153829849e-05,
+      "loss": 0.0457,
+      "step": 140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9712507460956008e-05,
+      "loss": 0.0379,
+      "step": 141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.970675726851542e-05,
+      "loss": 0.0348,
+      "step": 142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.97009509941991e-05,
+      "loss": 0.0481,
+      "step": 143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969508867155345e-05,
+      "loss": 0.019,
+      "step": 144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9689170334448695e-05,
+      "loss": 0.0487,
+      "step": 145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9683196017078692e-05,
+      "loss": 0.0391,
+      "step": 146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.967716575396072e-05,
+      "loss": 0.0534,
+      "step": 147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.967107957993531e-05,
+      "loss": 0.0438,
+      "step": 148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9664937530166002e-05,
+      "loss": 0.044,
+      "step": 149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9658739640139185e-05,
+      "loss": 0.0447,
+      "step": 150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9652485945663853e-05,
+      "loss": 0.0419,
+      "step": 151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.964617648287143e-05,
+      "loss": 0.0391,
+      "step": 152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9639811288215545e-05,
+      "loss": 0.052,
+      "step": 153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.963339039847182e-05,
+      "loss": 0.0394,
+      "step": 154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9626913850737658e-05,
+      "loss": 0.0458,
+      "step": 155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9620381682432047e-05,
+      "loss": 0.0465,
+      "step": 156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9613793931295317e-05,
+      "loss": 0.043,
+      "step": 157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9607150635388935e-05,
+      "loss": 0.0402,
+      "step": 158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9600451833095287e-05,
+      "loss": 0.0504,
+      "step": 159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.959369756311745e-05,
+      "loss": 0.0455,
+      "step": 160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9586887864478978e-05,
+      "loss": 0.0379,
+      "step": 161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9580022776523667e-05,
+      "loss": 0.0419,
+      "step": 162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9573102338915327e-05,
+      "loss": 0.0346,
+      "step": 163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.956612659163756e-05,
+      "loss": 0.0357,
+      "step": 164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.955909557499353e-05,
+      "loss": 0.0429,
+      "step": 165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.955200932960572e-05,
+      "loss": 0.0126,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9544867896415706e-05,
+      "loss": 0.0444,
+      "step": 167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9537671316683916e-05,
+      "loss": 0.0464,
+      "step": 168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9530419631989392e-05,
+      "loss": 0.0441,
+      "step": 169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9523112884229558e-05,
+      "loss": 0.0396,
+      "step": 170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9515751115619963e-05,
+      "loss": 0.0465,
+      "step": 171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9508334368694052e-05,
+      "loss": 0.0324,
+      "step": 172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9500862686302904e-05,
+      "loss": 0.0372,
+      "step": 173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9493336111615006e-05,
+      "loss": 0.0466,
+      "step": 174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948575468811598e-05,
+      "loss": 0.0313,
+      "step": 175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947811845960835e-05,
+      "loss": 0.0492,
+      "step": 176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9470427470211282e-05,
+      "loss": 0.0323,
+      "step": 177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9462681764360326e-05,
+      "loss": 0.0164,
+      "step": 178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9454881386807164e-05,
+      "loss": 0.0406,
+      "step": 179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.944702638261935e-05,
+      "loss": 0.0572,
+      "step": 180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9439116797180046e-05,
+      "loss": 0.048,
+      "step": 181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9431152676187774e-05,
+      "loss": 0.0358,
+      "step": 182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9423134065656123e-05,
+      "loss": 0.0422,
+      "step": 183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.941506101191352e-05,
+      "loss": 0.0304,
+      "step": 184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9406933561602938e-05,
+      "loss": 0.0462,
+      "step": 185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9398751761681626e-05,
+      "loss": 0.0432,
+      "step": 186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9390515659420846e-05,
+      "loss": 0.0544,
+      "step": 187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9382225302405603e-05,
+      "loss": 0.0447,
+      "step": 188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.937388073853436e-05,
+      "loss": 0.053,
+      "step": 189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.936548201601877e-05,
+      "loss": 0.0393,
+      "step": 190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.935702918338339e-05,
+      "loss": 0.0483,
+      "step": 191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9348522289465405e-05,
+      "loss": 0.0493,
+      "step": 192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.933996138341435e-05,
+      "loss": 0.0395,
+      "step": 193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9331346514691813e-05,
+      "loss": 0.0442,
+      "step": 194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9322677733071167e-05,
+      "loss": 0.0603,
+      "step": 195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9313955088637263e-05,
+      "loss": 0.0325,
+      "step": 196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9305178631786154e-05,
+      "loss": 0.0372,
+      "step": 197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.92963484132248e-05,
+      "loss": 0.0452,
+      "step": 198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.928746448397078e-05,
+      "loss": 0.0375,
+      "step": 199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9278526895351992e-05,
+      "loss": 0.0376,
+      "step": 200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9269535699006345e-05,
+      "loss": 0.031,
+      "step": 201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9260490946881488e-05,
+      "loss": 0.0428,
+      "step": 202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9251392691234486e-05,
+      "loss": 0.0298,
+      "step": 203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.924224098463153e-05,
+      "loss": 0.0243,
+      "step": 204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9233035879947627e-05,
+      "loss": 0.0428,
+      "step": 205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9223777430366298e-05,
+      "loss": 0.0245,
+      "step": 206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.921446568937927e-05,
+      "loss": 0.0393,
+      "step": 207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.920510071078617e-05,
+      "loss": 0.0265,
+      "step": 208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.919568254869421e-05,
+      "loss": 0.0417,
+      "step": 209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9186211257517873e-05,
+      "loss": 0.032,
+      "step": 210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9176686891978603e-05,
+      "loss": 0.0451,
+      "step": 211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.916710950710449e-05,
+      "loss": 0.035,
+      "step": 212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9157479158229942e-05,
+      "loss": 0.038,
+      "step": 213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.914779590099538e-05,
+      "loss": 0.0381,
+      "step": 214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9138059791346904e-05,
+      "loss": 0.0283,
+      "step": 215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.912827088553598e-05,
+      "loss": 0.0318,
+      "step": 216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9118429240119098e-05,
+      "loss": 0.0388,
+      "step": 217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9108534911957472e-05,
+      "loss": 0.0468,
+      "step": 218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.909858795821669e-05,
+      "loss": 0.0393,
+      "step": 219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9088588436366383e-05,
+      "loss": 0.0352,
+      "step": 220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9078536404179916e-05,
+      "loss": 0.0303,
+      "step": 221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.906843191973402e-05,
+      "loss": 0.0368,
+      "step": 222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.905827504140849e-05,
+      "loss": 0.0407,
+      "step": 223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9048065827885828e-05,
+      "loss": 0.0363,
+      "step": 224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.90378043381509e-05,
+      "loss": 0.0206,
+      "step": 225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9027490631490618e-05,
+      "loss": 0.0214,
+      "step": 226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9017124767493565e-05,
+      "loss": 0.0281,
+      "step": 227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9006706806049694e-05,
+      "loss": 0.0256,
+      "step": 228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.899623680734993e-05,
+      "loss": 0.0271,
+      "step": 229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.898571483188587e-05,
+      "loss": 0.0223,
+      "step": 230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8975140940449396e-05,
+      "loss": 0.0312,
+      "step": 231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8964515194132358e-05,
+      "loss": 0.0372,
+      "step": 232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8953837654326185e-05,
+      "loss": 0.0242,
+      "step": 233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.894310838272156e-05,
+      "loss": 0.0355,
+      "step": 234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8932327441308057e-05,
+      "loss": 0.0173,
+      "step": 235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.892149489237376e-05,
+      "loss": 0.0435,
+      "step": 236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8910610798504934e-05,
+      "loss": 0.0323,
+      "step": 237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.889967522258566e-05,
+      "loss": 0.0406,
+      "step": 238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8888688227797434e-05,
+      "loss": 0.0397,
+      "step": 239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8877649877618854e-05,
+      "loss": 0.0396,
+      "step": 240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8866560235825232e-05,
+      "loss": 0.0284,
+      "step": 241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.88554193664882e-05,
+      "loss": 0.0367,
+      "step": 242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8844227333975383e-05,
+      "loss": 0.041,
+      "step": 243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.883298420295e-05,
+      "loss": 0.0425,
+      "step": 244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8821690038370497e-05,
+      "loss": 0.0238,
+      "step": 245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8810344905490173e-05,
+      "loss": 0.027,
+      "step": 246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8798948869856804e-05,
+      "loss": 0.0169,
+      "step": 247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8787501997312257e-05,
+      "loss": 0.0305,
+      "step": 248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8776004353992127e-05,
+      "loss": 0.0289,
+      "step": 249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.876445600632532e-05,
+      "loss": 0.025,
+      "step": 250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8752857021033716e-05,
+      "loss": 0.0192,
+      "step": 251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.874120746513175e-05,
+      "loss": 0.0136,
+      "step": 252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8729507405926032e-05,
+      "loss": 0.044,
+      "step": 253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.871775691101496e-05,
+      "loss": 0.0282,
+      "step": 254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8705956048288335e-05,
+      "loss": 0.0178,
+      "step": 255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.869410488592696e-05,
+      "loss": 0.0296,
+      "step": 256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.868220349240226e-05,
+      "loss": 0.0153,
+      "step": 257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.867025193647585e-05,
+      "loss": 0.0245,
+      "step": 258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8658250287199196e-05,
+      "loss": 0.0233,
+      "step": 259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.864619861391316e-05,
+      "loss": 0.0249,
+      "step": 260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8634096986247638e-05,
+      "loss": 0.0253,
+      "step": 261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8621945474121134e-05,
+      "loss": 0.0263,
+      "step": 262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8609744147740367e-05,
+      "loss": 0.0275,
+      "step": 263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.859749307759987e-05,
+      "loss": 0.0352,
+      "step": 264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8585192334481562e-05,
+      "loss": 0.0348,
+      "step": 265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8572841989454376e-05,
+      "loss": 0.0147,
+      "step": 266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.85604421138738e-05,
+      "loss": 0.0418,
+      "step": 267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8547992779381507e-05,
+      "loss": 0.0426,
+      "step": 268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8535494057904916e-05,
+      "loss": 0.0408,
+      "step": 269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.852294602165679e-05,
+      "loss": 0.0317,
+      "step": 270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.851034874313481e-05,
+      "loss": 0.0456,
+      "step": 271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849770229512117e-05,
+      "loss": 0.0306,
+      "step": 272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8485006750682126e-05,
+      "loss": 0.0305,
+      "step": 273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8472262183167614e-05,
+      "loss": 0.0372,
+      "step": 274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8459468666210797e-05,
+      "loss": 0.0322,
+      "step": 275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8446626273727643e-05,
+      "loss": 0.0369,
+      "step": 276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.843373507991652e-05,
+      "loss": 0.0494,
+      "step": 277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8420795159257737e-05,
+      "loss": 0.0326,
+      "step": 278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8407806586513134e-05,
+      "loss": 0.0353,
+      "step": 279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8394769436725645e-05,
+      "loss": 0.0362,
+      "step": 280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8381683785218855e-05,
+      "loss": 0.0353,
+      "step": 281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.836854970759659e-05,
+      "loss": 0.0325,
+      "step": 282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8355367279742436e-05,
+      "loss": 0.0387,
+      "step": 283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8342136577819358e-05,
+      "loss": 0.0479,
+      "step": 284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8328857678269214e-05,
+      "loss": 0.0407,
+      "step": 285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8315530657812326e-05,
+      "loss": 0.0294,
+      "step": 286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8302155593447047e-05,
+      "loss": 0.0277,
+      "step": 287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.828873256244931e-05,
+      "loss": 0.0309,
+      "step": 288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8275261642372177e-05,
+      "loss": 0.0371,
+      "step": 289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8261742911045392e-05,
+      "loss": 0.022,
+      "step": 290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.824817644657494e-05,
+      "loss": 0.0225,
+      "step": 291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.823456232734259e-05,
+      "loss": 0.0501,
+      "step": 292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8220900632005428e-05,
+      "loss": 0.0225,
+      "step": 293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.820719143949544e-05,
+      "loss": 0.026,
+      "step": 294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.819343482901901e-05,
+      "loss": 0.0237,
+      "step": 295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.81796308800565e-05,
+      "loss": 0.0308,
+      "step": 296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8165779672361757e-05,
+      "loss": 0.0241,
+      "step": 297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.81518812859617e-05,
+      "loss": 0.0269,
+      "step": 298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8137935801155793e-05,
+      "loss": 0.0314,
+      "step": 299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8123943298515646e-05,
+      "loss": 0.0184,
+      "step": 300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.810990385888451e-05,
+      "loss": 0.0281,
+      "step": 301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8095817563376806e-05,
+      "loss": 0.0246,
+      "step": 302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.808168449337769e-05,
+      "loss": 0.0316,
+      "step": 303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.806750473054255e-05,
+      "loss": 0.0252,
+      "step": 304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.805327835679656e-05,
+      "loss": 0.0411,
+      "step": 305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8039005454334177e-05,
+      "loss": 0.0234,
+      "step": 306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8024686105618695e-05,
+      "loss": 0.038,
+      "step": 307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.801032039338175e-05,
+      "loss": 0.0202,
+      "step": 308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.799590840062285e-05,
+      "loss": 0.0332,
+      "step": 309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7981450210608898e-05,
+      "loss": 0.0274,
+      "step": 310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7966945906873706e-05,
+      "loss": 0.047,
+      "step": 311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.795239557321751e-05,
+      "loss": 0.0236,
+      "step": 312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.793779929370649e-05,
+      "loss": 0.0215,
+      "step": 313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.792315715267228e-05,
+      "loss": 0.0301,
+      "step": 314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7908469234711486e-05,
+      "loss": 0.0239,
+      "step": 315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7893735624685203e-05,
+      "loss": 0.0242,
+      "step": 316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7878956407718505e-05,
+      "loss": 0.0274,
+      "step": 317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7864131669199975e-05,
+      "loss": 0.0209,
+      "step": 318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.78492614947812e-05,
+      "loss": 0.0218,
+      "step": 319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7834345970376272e-05,
+      "loss": 0.0181,
+      "step": 320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7819385182161296e-05,
+      "loss": 0.0133,
+      "step": 321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.780437921657391e-05,
+      "loss": 0.0218,
+      "step": 322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.778932816031275e-05,
+      "loss": 0.0108,
+      "step": 323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7774232100336983e-05,
+      "loss": 0.0127,
+      "step": 324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7759091123865778e-05,
+      "loss": 0.0214,
+      "step": 325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7743905318377826e-05,
+      "loss": 0.019,
+      "step": 326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7728674771610814e-05,
+      "loss": 0.0146,
+      "step": 327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.771339957156093e-05,
+      "loss": 0.0179,
+      "step": 328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7698079806482344e-05,
+      "loss": 0.0201,
+      "step": 329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.768271556488672e-05,
+      "loss": 0.0174,
+      "step": 330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7667306935542682e-05,
+      "loss": 0.0254,
+      "step": 331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7651854007475307e-05,
+      "loss": 0.016,
+      "step": 332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7636356869965615e-05,
+      "loss": 0.0086,
+      "step": 333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.762081561255005e-05,
+      "loss": 0.0297,
+      "step": 334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.760523032501997e-05,
+      "loss": 0.0202,
+      "step": 335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.758960109742111e-05,
+      "loss": 0.0224,
+      "step": 336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7573928020053083e-05,
+      "loss": 0.0299,
+      "step": 337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7558211183468848e-05,
+      "loss": 0.0375,
+      "step": 338
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7542450678474187e-05,
+      "loss": 0.0262,
+      "step": 339
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7526646596127177e-05,
+      "loss": 0.0196,
+      "step": 340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7510799027737672e-05,
+      "loss": 0.0262,
+      "step": 341
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7494908064866775e-05,
+      "loss": 0.012,
+      "step": 342
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7478973799326296e-05,
+      "loss": 0.0174,
+      "step": 343
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7462996323178236e-05,
+      "loss": 0.0209,
+      "step": 344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.744697572873425e-05,
+      "loss": 0.0192,
+      "step": 345
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7430912108555125e-05,
+      "loss": 0.0135,
+      "step": 346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.741480555545021e-05,
+      "loss": 0.0294,
+      "step": 347
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.739865616247692e-05,
+      "loss": 0.0149,
+      "step": 348
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7382464022940183e-05,
+      "loss": 0.0169,
+      "step": 349
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7366229230391894e-05,
+      "loss": 0.0159,
+      "step": 350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.734995187863038e-05,
+      "loss": 0.022,
+      "step": 351
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7333632061699866e-05,
+      "loss": 0.0123,
+      "step": 352
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7317269873889915e-05,
+      "loss": 0.0229,
+      "step": 353
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7300865409734902e-05,
+      "loss": 0.0134,
+      "step": 354
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.728441876401345e-05,
+      "loss": 0.0251,
+      "step": 355
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7267930031747896e-05,
+      "loss": 0.014,
+      "step": 356
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.725139930820373e-05,
+      "loss": 0.0395,
+      "step": 357
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7234826688889064e-05,
+      "loss": 0.0201,
+      "step": 358
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.721821226955405e-05,
+      "loss": 0.0118,
+      "step": 359
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7201556146190363e-05,
+      "loss": 0.0436,
+      "step": 360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7184858415030613e-05,
+      "loss": 0.0398,
+      "step": 361
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.716811917254781e-05,
+      "loss": 0.0361,
+      "step": 362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7151338515454802e-05,
+      "loss": 0.0169,
+      "step": 363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7134516540703713e-05,
+      "loss": 0.0324,
+      "step": 364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7117653345485378e-05,
+      "loss": 0.0401,
+      "step": 365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7100749027228797e-05,
+      "loss": 0.024,
+      "step": 366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7083803683600555e-05,
+      "loss": 0.0159,
+      "step": 367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.706681741250428e-05,
+      "loss": 0.0307,
+      "step": 368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.704979031208004e-05,
+      "loss": 0.0772,
+      "step": 369
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7032722480703826e-05,
+      "loss": 0.0316,
+      "step": 370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.701561401698693e-05,
+      "loss": 0.0166,
+      "step": 371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.699846501977542e-05,
+      "loss": 0.0264,
+      "step": 372
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.698127558814955e-05,
+      "loss": 0.0327,
+      "step": 373
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.696404582142318e-05,
+      "loss": 0.0218,
+      "step": 374
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6946775819143205e-05,
+      "loss": 0.0191,
+      "step": 375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.692946568108901e-05,
+      "loss": 0.0291,
+      "step": 376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6912115507271845e-05,
+      "loss": 0.0295,
+      "step": 377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.689472539793428e-05,
+      "loss": 0.0522,
+      "step": 378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6877295453549617e-05,
+      "loss": 0.0283,
+      "step": 379
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.685982577482131e-05,
+      "loss": 0.0378,
+      "step": 380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6842316462682383e-05,
+      "loss": 0.0249,
+      "step": 381
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6824767618294842e-05,
+      "loss": 0.0225,
+      "step": 382
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.68071793430491e-05,
+      "loss": 0.0314,
+      "step": 383
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6789551738563384e-05,
+      "loss": 0.0305,
+      "step": 384
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6771884906683156e-05,
+      "loss": 0.0179,
+      "step": 385
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.67541789494805e-05,
+      "loss": 0.0177,
+      "step": 386
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6736433969253576e-05,
+      "loss": 0.0156,
+      "step": 387
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6718650068525985e-05,
+      "loss": 0.0159,
+      "step": 388
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6700827350046206e-05,
+      "loss": 0.0258,
+      "step": 389
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6682965916786988e-05,
+      "loss": 0.0464,
+      "step": 390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.666506587194476e-05,
+      "loss": 0.0121,
+      "step": 391
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6647127318939027e-05,
+      "loss": 0.0266,
+      "step": 392
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.662915036141179e-05,
+      "loss": 0.0182,
+      "step": 393
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6611135103226936e-05,
+      "loss": 0.0356,
+      "step": 394
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6593081648469627e-05,
+      "loss": 0.0268,
+      "step": 395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6574990101445716e-05,
+      "loss": 0.0156,
+      "step": 396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6556860566681147e-05,
+      "loss": 0.011,
+      "step": 397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.653869314892133e-05,
+      "loss": 0.0598,
+      "step": 398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6520487953130553e-05,
+      "loss": 0.0309,
+      "step": 399
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.650224508449137e-05,
+      "loss": 0.035,
+      "step": 400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6483964648403997e-05,
+      "loss": 0.0201,
+      "step": 401
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6465646750485695e-05,
+      "loss": 0.0193,
+      "step": 402
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.644729149657017e-05,
+      "loss": 0.0247,
+      "step": 403
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6428898992706957e-05,
+      "loss": 0.0233,
+      "step": 404
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.64104693451608e-05,
+      "loss": 0.0301,
+      "step": 405
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.639200266041105e-05,
+      "loss": 0.0247,
+      "step": 406
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6373499045151046e-05,
+      "loss": 0.0171,
+      "step": 407
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6354958606287488e-05,
+      "loss": 0.0213,
+      "step": 408
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6336381450939843e-05,
+      "loss": 0.0246,
+      "step": 409
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.63177676864397e-05,
+      "loss": 0.0166,
+      "step": 410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6299117420330164e-05,
+      "loss": 0.0279,
+      "step": 411
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6280430760365234e-05,
+      "loss": 0.0397,
+      "step": 412
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.626170781450918e-05,
+      "loss": 0.0135,
+      "step": 413
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.624294869093591e-05,
+      "loss": 0.0129,
+      "step": 414
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6224153498028367e-05,
+      "loss": 0.0202,
+      "step": 415
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6205322344377873e-05,
+      "loss": 0.026,
+      "step": 416
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.618645533878353e-05,
+      "loss": 0.016,
+      "step": 417
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6167552590251568e-05,
+      "loss": 0.0187,
+      "step": 418
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6148614207994735e-05,
+      "loss": 0.0196,
+      "step": 419
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6129640301431648e-05,
+      "loss": 0.0155,
+      "step": 420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6110630980186174e-05,
+      "loss": 0.0224,
+      "step": 421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6091586354086798e-05,
+      "loss": 0.0141,
+      "step": 422
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.607250653316598e-05,
+      "loss": 0.0233,
+      "step": 423
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6053391627659505e-05,
+      "loss": 0.0107,
+      "step": 424
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6034241748005887e-05,
+      "loss": 0.0357,
+      "step": 425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6015057004845697e-05,
+      "loss": 0.0192,
+      "step": 426
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5995837509020934e-05,
+      "loss": 0.0316,
+      "step": 427
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.597658337157439e-05,
+      "loss": 0.0286,
+      "step": 428
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5957294703748983e-05,
+      "loss": 0.009,
+      "step": 429
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.593797161698716e-05,
+      "loss": 0.0153,
+      "step": 430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5918614222930214e-05,
+      "loss": 0.0292,
+      "step": 431
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5899222633417654e-05,
+      "loss": 0.0189,
+      "step": 432
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5879796960486555e-05,
+      "loss": 0.0124,
+      "step": 433
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5860337316370918e-05,
+      "loss": 0.0082,
+      "step": 434
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5840843813501014e-05,
+      "loss": 0.0246,
+      "step": 435
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.582131656450273e-05,
+      "loss": 0.0211,
+      "step": 436
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5801755682196933e-05,
+      "loss": 0.0133,
+      "step": 437
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5782161279598813e-05,
+      "loss": 0.0086,
+      "step": 438
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5762533469917217e-05,
+      "loss": 0.0058,
+      "step": 439
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5742872366554018e-05,
+      "loss": 0.0259,
+      "step": 440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5723178083103428e-05,
+      "loss": 0.009,
+      "step": 441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.570345073335138e-05,
+      "loss": 0.026,
+      "step": 442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5683690431274844e-05,
+      "loss": 0.0196,
+      "step": 443
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5663897291041177e-05,
+      "loss": 0.0117,
+      "step": 444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5644071427007454e-05,
+      "loss": 0.0107,
+      "step": 445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5624212953719825e-05,
+      "loss": 0.0099,
+      "step": 446
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5604321985912842e-05,
+      "loss": 0.0095,
+      "step": 447
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5584398638508788e-05,
+      "loss": 0.0189,
+      "step": 448
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5564443026617042e-05,
+      "loss": 0.0162,
+      "step": 449
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5544455265533377e-05,
+      "loss": 0.0156,
+      "step": 450
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5524435470739322e-05,
+      "loss": 0.019,
+      "step": 451
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.550438375790148e-05,
+      "loss": 0.016,
+      "step": 452
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.548430024287086e-05,
+      "loss": 0.0045,
+      "step": 453
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.546418504168222e-05,
+      "loss": 0.0089,
+      "step": 454
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.544403827055338e-05,
+      "loss": 0.0134,
+      "step": 455
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5423860045884575e-05,
+      "loss": 0.007,
+      "step": 456
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.540365048425774e-05,
+      "loss": 0.0226,
+      "step": 457
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5383409702435885e-05,
+      "loss": 0.0288,
+      "step": 458
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5363137817362393e-05,
+      "loss": 0.0224,
+      "step": 459
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.534283494616034e-05,
+      "loss": 0.0184,
+      "step": 460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5322501206131845e-05,
+      "loss": 0.0204,
+      "step": 461
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5302136714757354e-05,
+      "loss": 0.033,
+      "step": 462
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5281741589695002e-05,
+      "loss": 0.0159,
+      "step": 463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5261315948779904e-05,
+      "loss": 0.0193,
+      "step": 464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5240859910023479e-05,
+      "loss": 0.0082,
+      "step": 465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5220373591612785e-05,
+      "loss": 0.018,
+      "step": 466
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5199857111909812e-05,
+      "loss": 0.0134,
+      "step": 467
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5179310589450824e-05,
+      "loss": 0.0217,
+      "step": 468
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5158734142945645e-05,
+      "loss": 0.021,
+      "step": 469
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5138127891277e-05,
+      "loss": 0.0096,
+      "step": 470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5117491953499813e-05,
+      "loss": 0.0178,
+      "step": 471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5096826448840518e-05,
+      "loss": 0.0081,
+      "step": 472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5076131496696388e-05,
+      "loss": 0.0076,
+      "step": 473
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.505540721663481e-05,
+      "loss": 0.0299,
+      "step": 474
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.503465372839264e-05,
+      "loss": 0.0175,
+      "step": 475
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5013871151875478e-05,
+      "loss": 0.0115,
+      "step": 476
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.499305960715698e-05,
+      "loss": 0.0138,
+      "step": 477
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4972219214478177e-05,
+      "loss": 0.0122,
+      "step": 478
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4951350094246761e-05,
+      "loss": 0.0128,
+      "step": 479
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.493045236703642e-05,
+      "loss": 0.0059,
+      "step": 480
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.49095261535861e-05,
+      "loss": 0.0129,
+      "step": 481
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.488857157479934e-05,
+      "loss": 0.0166,
+      "step": 482
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4867588751743569e-05,
+      "loss": 0.0168,
+      "step": 483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4846577805649389e-05,
+      "loss": 0.009,
+      "step": 484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4825538857909893e-05,
+      "loss": 0.0035,
+      "step": 485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4804472030079953e-05,
+      "loss": 0.0249,
+      "step": 486
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4783377443875526e-05,
+      "loss": 0.0079,
+      "step": 487
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4762255221172941e-05,
+      "loss": 0.0096,
+      "step": 488
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.47411054840082e-05,
+      "loss": 0.0169,
+      "step": 489
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4719928354576278e-05,
+      "loss": 0.0266,
+      "step": 490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4698723955230407e-05,
+      "loss": 0.0099,
+      "step": 491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4677492408481376e-05,
+      "loss": 0.0212,
+      "step": 492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4656233836996822e-05,
+      "loss": 0.0138,
+      "step": 493
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4634948363600518e-05,
+      "loss": 0.0091,
+      "step": 494
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4613636111271668e-05,
+      "loss": 0.0155,
+      "step": 495
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.459229720314419e-05,
+      "loss": 0.0233,
+      "step": 496
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.457093176250602e-05,
+      "loss": 0.0199,
+      "step": 497
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4549539912798376e-05,
+      "loss": 0.0187,
+      "step": 498
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4528121777615058e-05,
+      "loss": 0.0071,
+      "step": 499
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4506677480701743e-05,
+      "loss": 0.0138,
+      "step": 500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4485207145955253e-05,
+      "loss": 0.0171,
+      "step": 501
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4463710897422853e-05,
+      "loss": 0.0131,
+      "step": 502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4442188859301516e-05,
+      "loss": 0.0078,
+      "step": 503
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4420641155937225e-05,
+      "loss": 0.0133,
+      "step": 504
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4399067911824251e-05,
+      "loss": 0.0078,
+      "step": 505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4377469251604421e-05,
+      "loss": 0.0094,
+      "step": 506
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4355845300066413e-05,
+      "loss": 0.0144,
+      "step": 507
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.433419618214502e-05,
+      "loss": 0.0105,
+      "step": 508
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4312522022920445e-05,
+      "loss": 0.0052,
+      "step": 509
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4290822947617572e-05,
+      "loss": 0.0205,
+      "step": 510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4269099081605227e-05,
+      "loss": 0.0152,
+      "step": 511
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4247350550395479e-05,
+      "loss": 0.0218,
+      "step": 512
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4225577479642898e-05,
+      "loss": 0.0131,
+      "step": 513
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4203779995143841e-05,
+      "loss": 0.0275,
+      "step": 514
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4181958222835706e-05,
+      "loss": 0.0219,
+      "step": 515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4160112288796227e-05,
+      "loss": 0.0311,
+      "step": 516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4138242319242728e-05,
+      "loss": 0.0232,
+      "step": 517
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4116348440531406e-05,
+      "loss": 0.0103,
+      "step": 518
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4094430779156596e-05,
+      "loss": 0.0164,
+      "step": 519
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4072489461750034e-05,
+      "loss": 0.0514,
+      "step": 520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4050524615080137e-05,
+      "loss": 0.0137,
+      "step": 521
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4028536366051265e-05,
+      "loss": 0.0131,
+      "step": 522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.400652484170298e-05,
+      "loss": 0.0204,
+      "step": 523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3984490169209333e-05,
+      "loss": 0.0144,
+      "step": 524
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3962432475878103e-05,
+      "loss": 0.0151,
+      "step": 525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3940351889150084e-05,
+      "loss": 0.0126,
+      "step": 526
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3918248536598333e-05,
+      "loss": 0.0108,
+      "step": 527
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3896122545927442e-05,
+      "loss": 0.0112,
+      "step": 528
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.38739740449728e-05,
+      "loss": 0.0068,
+      "step": 529
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.385180316169984e-05,
+      "loss": 0.0069,
+      "step": 530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.382961002420333e-05,
+      "loss": 0.0105,
+      "step": 531
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3807394760706596e-05,
+      "loss": 0.0041,
+      "step": 532
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3785157499560814e-05,
+      "loss": 0.0036,
+      "step": 533
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.376289836924424e-05,
+      "loss": 0.0103,
+      "step": 534
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3740617498361494e-05,
+      "loss": 0.015,
+      "step": 535
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3718315015642801e-05,
+      "loss": 0.0176,
+      "step": 536
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.369599104994324e-05,
+      "loss": 0.0105,
+      "step": 537
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.367364573024203e-05,
+      "loss": 0.0141,
+      "step": 538
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3651279185641753e-05,
+      "loss": 0.017,
+      "step": 539
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3628891545367616e-05,
+      "loss": 0.0109,
+      "step": 540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3606482938766719e-05,
+      "loss": 0.0161,
+      "step": 541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3584053495307287e-05,
+      "loss": 0.0293,
+      "step": 542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3561603344577949e-05,
+      "loss": 0.0411,
+      "step": 543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3539132616286956e-05,
+      "loss": 0.0181,
+      "step": 544
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.351664144026145e-05,
+      "loss": 0.0076,
+      "step": 545
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3494129946446722e-05,
+      "loss": 0.0056,
+      "step": 546
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.347159826490544e-05,
+      "loss": 0.0163,
+      "step": 547
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.344904652581692e-05,
+      "loss": 0.0136,
+      "step": 548
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.342647485947635e-05,
+      "loss": 0.0146,
+      "step": 549
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3403883396294062e-05,
+      "loss": 0.0116,
+      "step": 550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.338127226679476e-05,
+      "loss": 0.0082,
+      "step": 551
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3358641601616774e-05,
+      "loss": 0.0187,
+      "step": 552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.33359915315113e-05,
+      "loss": 0.01,
+      "step": 553
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3313322187341653e-05,
+      "loss": 0.0091,
+      "step": 554
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3290633700082502e-05,
+      "loss": 0.0069,
+      "step": 555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3267926200819128e-05,
+      "loss": 0.0094,
+      "step": 556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3245199820746637e-05,
+      "loss": 0.0257,
+      "step": 557
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3222454691169239e-05,
+      "loss": 0.0048,
+      "step": 558
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3199690943499457e-05,
+      "loss": 0.0112,
+      "step": 559
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3176908709257398e-05,
+      "loss": 0.0161,
+      "step": 560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3154108120069963e-05,
+      "loss": 0.0067,
+      "step": 561
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3131289307670107e-05,
+      "loss": 0.0123,
+      "step": 562
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.310845240389607e-05,
+      "loss": 0.0155,
+      "step": 563
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3085597540690618e-05,
+      "loss": 0.0076,
+      "step": 564
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3062724850100279e-05,
+      "loss": 0.0149,
+      "step": 565
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.303983446427458e-05,
+      "loss": 0.0129,
+      "step": 566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3016926515465288e-05,
+      "loss": 0.0168,
+      "step": 567
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.299400113602564e-05,
+      "loss": 0.0123,
+      "step": 568
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2971058458409576e-05,
+      "loss": 0.0084,
+      "step": 569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2948098615170983e-05,
+      "loss": 0.0036,
+      "step": 570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2925121738962922e-05,
+      "loss": 0.0137,
+      "step": 571
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2902127962536867e-05,
+      "loss": 0.0194,
+      "step": 572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2879117418741935e-05,
+      "loss": 0.0084,
+      "step": 573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2856090240524112e-05,
+      "loss": 0.0055,
+      "step": 574
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2833046560925499e-05,
+      "loss": 0.0077,
+      "step": 575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2809986513083527e-05,
+      "loss": 0.0144,
+      "step": 576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.278691023023021e-05,
+      "loss": 0.0084,
+      "step": 577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2763817845691345e-05,
+      "loss": 0.0142,
+      "step": 578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.274070949288577e-05,
+      "loss": 0.0051,
+      "step": 579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2717585305324575e-05,
+      "loss": 0.0249,
+      "step": 580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2694445416610344e-05,
+      "loss": 0.0107,
+      "step": 581
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2671289960436371e-05,
+      "loss": 0.0097,
+      "step": 582
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2648119070585888e-05,
+      "loss": 0.0172,
+      "step": 583
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2624932880931312e-05,
+      "loss": 0.0139,
+      "step": 584
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2601731525433443e-05,
+      "loss": 0.0141,
+      "step": 585
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2578515138140702e-05,
+      "loss": 0.0042,
+      "step": 586
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2555283853188371e-05,
+      "loss": 0.0134,
+      "step": 587
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2532037804797791e-05,
+      "loss": 0.0218,
+      "step": 588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2508777127275611e-05,
+      "loss": 0.019,
+      "step": 589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2485501955012996e-05,
+      "loss": 0.0209,
+      "step": 590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2462212422484852e-05,
+      "loss": 0.0063,
+      "step": 591
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2438908664249064e-05,
+      "loss": 0.0135,
+      "step": 592
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2415590814945697e-05,
+      "loss": 0.0069,
+      "step": 593
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2392259009296239e-05,
+      "loss": 0.0064,
+      "step": 594
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.23689133821028e-05,
+      "loss": 0.0034,
+      "step": 595
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2345554068247359e-05,
+      "loss": 0.0166,
+      "step": 596
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2322181202690954e-05,
+      "loss": 0.0054,
+      "step": 597
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.229879492047294e-05,
+      "loss": 0.0112,
+      "step": 598
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2275395356710176e-05,
+      "loss": 0.0033,
+      "step": 599
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2251982646596254e-05,
+      "loss": 0.0063,
+      "step": 600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.222855692540073e-05,
+      "loss": 0.0066,
+      "step": 601
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2205118328468327e-05,
+      "loss": 0.0048,
+      "step": 602
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2181666991218156e-05,
+      "loss": 0.0048,
+      "step": 603
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2158203049142947e-05,
+      "loss": 0.0094,
+      "step": 604
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.213472663780824e-05,
+      "loss": 0.0077,
+      "step": 605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.211123789285164e-05,
+      "loss": 0.0046,
+      "step": 606
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2087736949981994e-05,
+      "loss": 0.0104,
+      "step": 607
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2064223944978623e-05,
+      "loss": 0.0192,
+      "step": 608
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2040699013690545e-05,
+      "loss": 0.0246,
+      "step": 609
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2017162292035683e-05,
+      "loss": 0.0175,
+      "step": 610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1993613916000078e-05,
+      "loss": 0.0091,
+      "step": 611
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1970054021637106e-05,
+      "loss": 0.0078,
+      "step": 612
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.194648274506669e-05,
+      "loss": 0.0084,
+      "step": 613
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1922900222474522e-05,
+      "loss": 0.0088,
+      "step": 614
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1899306590111257e-05,
+      "loss": 0.0124,
+      "step": 615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1875701984291751e-05,
+      "loss": 0.0096,
+      "step": 616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1852086541394247e-05,
+      "loss": 0.0053,
+      "step": 617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1828460397859616e-05,
+      "loss": 0.0206,
+      "step": 618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1804823690190539e-05,
+      "loss": 0.0036,
+      "step": 619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1781176554950738e-05,
+      "loss": 0.0156,
+      "step": 620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1757519128764183e-05,
+      "loss": 0.0191,
+      "step": 621
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1733851548314297e-05,
+      "loss": 0.0054,
+      "step": 622
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.171017395034318e-05,
+      "loss": 0.0099,
+      "step": 623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1686486471650797e-05,
+      "loss": 0.0048,
+      "step": 624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1662789249094203e-05,
+      "loss": 0.0063,
+      "step": 625
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1639082419586758e-05,
+      "loss": 0.0069,
+      "step": 626
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1615366120097318e-05,
+      "loss": 0.022,
+      "step": 627
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1591640487649457e-05,
+      "loss": 0.0102,
+      "step": 628
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1567905659320665e-05,
+      "loss": 0.0147,
+      "step": 629
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1544161772241568e-05,
+      "loss": 0.0049,
+      "step": 630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1520408963595137e-05,
+      "loss": 0.0067,
+      "step": 631
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1496647370615874e-05,
+      "loss": 0.0175,
+      "step": 632
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1472877130589042e-05,
+      "loss": 0.0042,
+      "step": 633
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1449098380849858e-05,
+      "loss": 0.019,
+      "step": 634
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1425311258782711e-05,
+      "loss": 0.003,
+      "step": 635
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1401515901820363e-05,
+      "loss": 0.0101,
+      "step": 636
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.137771244744314e-05,
+      "loss": 0.0076,
+      "step": 637
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1353901033178171e-05,
+      "loss": 0.0148,
+      "step": 638
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.133008179659856e-05,
+      "loss": 0.0029,
+      "step": 639
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1306254875322612e-05,
+      "loss": 0.0072,
+      "step": 640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1282420407013033e-05,
+      "loss": 0.0108,
+      "step": 641
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1258578529376123e-05,
+      "loss": 0.0094,
+      "step": 642
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1234729380160998e-05,
+      "loss": 0.0024,
+      "step": 643
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1210873097158785e-05,
+      "loss": 0.018,
+      "step": 644
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1187009818201832e-05,
+      "loss": 0.0162,
+      "step": 645
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1163139681162898e-05,
+      "loss": 0.0025,
+      "step": 646
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1139262823954367e-05,
+      "loss": 0.0049,
+      "step": 647
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1115379384527459e-05,
+      "loss": 0.0028,
+      "step": 648
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1091489500871409e-05,
+      "loss": 0.0059,
+      "step": 649
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1067593311012697e-05,
+      "loss": 0.011,
+      "step": 650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1043690953014226e-05,
+      "loss": 0.0035,
+      "step": 651
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.101978256497455e-05,
+      "loss": 0.004,
+      "step": 652
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0995868285027051e-05,
+      "loss": 0.0033,
+      "step": 653
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0971948251339157e-05,
+      "loss": 0.0078,
+      "step": 654
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0948022602111537e-05,
+      "loss": 0.0074,
+      "step": 655
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0924091475577305e-05,
+      "loss": 0.0016,
+      "step": 656
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0900155010001226e-05,
+      "loss": 0.0019,
+      "step": 657
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0876213343678899e-05,
+      "loss": 0.0084,
+      "step": 658
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0852266614935982e-05,
+      "loss": 0.0236,
+      "step": 659
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0828314962127382e-05,
+      "loss": 0.0034,
+      "step": 660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0804358523636447e-05,
+      "loss": 0.0143,
+      "step": 661
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0780397437874183e-05,
+      "loss": 0.0182,
+      "step": 662
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.075643184327844e-05,
+      "loss": 0.014,
+      "step": 663
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0732461878313125e-05,
+      "loss": 0.0096,
+      "step": 664
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0708487681467388e-05,
+      "loss": 0.0065,
+      "step": 665
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0684509391254834e-05,
+      "loss": 0.0299,
+      "step": 666
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0660527146212716e-05,
+      "loss": 0.0154,
+      "step": 667
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0636541084901132e-05,
+      "loss": 0.0171,
+      "step": 668
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0612551345902245e-05,
+      "loss": 0.0161,
+      "step": 669
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0588558067819447e-05,
+      "loss": 0.0058,
+      "step": 670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0564561389276588e-05,
+      "loss": 0.0099,
+      "step": 671
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0540561448917159e-05,
+      "loss": 0.0134,
+      "step": 672
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0516558385403503e-05,
+      "loss": 0.0119,
+      "step": 673
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0492552337416007e-05,
+      "loss": 0.0087,
+      "step": 674
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.046854344365229e-05,
+      "loss": 0.0054,
+      "step": 675
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0444531842826427e-05,
+      "loss": 0.0037,
+      "step": 676
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0420517673668125e-05,
+      "loss": 0.0033,
+      "step": 677
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0396501074921932e-05,
+      "loss": 0.0113,
+      "step": 678
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0372482185346435e-05,
+      "loss": 0.003,
+      "step": 679
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0348461143713454e-05,
+      "loss": 0.0076,
+      "step": 680
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0324438088807238e-05,
+      "loss": 0.0111,
+      "step": 681
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0300413159423684e-05,
+      "loss": 0.0088,
+      "step": 682
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0276386494369501e-05,
+      "loss": 0.0053,
+      "step": 683
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0252358232461437e-05,
+      "loss": 0.0074,
+      "step": 684
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0228328512525457e-05,
+      "loss": 0.0072,
+      "step": 685
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0204297473395964e-05,
+      "loss": 0.0041,
+      "step": 686
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0180265253914967e-05,
+      "loss": 0.0037,
+      "step": 687
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0156231992931305e-05,
+      "loss": 0.0099,
+      "step": 688
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.013219782929983e-05,
+      "loss": 0.0025,
+      "step": 689
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0108162901880611e-05,
+      "loss": 0.0021,
+      "step": 690
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0084127349538133e-05,
+      "loss": 0.0017,
+      "step": 691
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0060091311140481e-05,
+      "loss": 0.0054,
+      "step": 692
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0036054925558557e-05,
+      "loss": 0.0008,
+      "step": 693
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0012018331665272e-05,
+      "loss": 0.0041,
+      "step": 694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.98798166833473e-06,
+      "loss": 0.0033,
+      "step": 695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.963945074441444e-06,
+      "loss": 0.0047,
+      "step": 696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.939908688859522e-06,
+      "loss": 0.0053,
+      "step": 697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.915872650461872e-06,
+      "loss": 0.0082,
+      "step": 698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.891837098119389e-06,
+      "loss": 0.001,
+      "step": 699
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.86780217070017e-06,
+      "loss": 0.0031,
+      "step": 700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.843768007068696e-06,
+      "loss": 0.0014,
+      "step": 701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.819734746085034e-06,
+      "loss": 0.0027,
+      "step": 702
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.795702526604041e-06,
+      "loss": 0.0022,
+      "step": 703
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.771671487474546e-06,
+      "loss": 0.0026,
+      "step": 704
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.747641767538568e-06,
+      "loss": 0.0075,
+      "step": 705
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.723613505630504e-06,
+      "loss": 0.002,
+      "step": 706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.699586840576321e-06,
+      "loss": 0.0087,
+      "step": 707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.675561911192762e-06,
+      "loss": 0.0025,
+      "step": 708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.651538856286551e-06,
+      "loss": 0.0103,
+      "step": 709
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.627517814653568e-06,
+      "loss": 0.0087,
+      "step": 710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.60349892507807e-06,
+      "loss": 0.0025,
+      "step": 711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.579482326331877e-06,
+      "loss": 0.0091,
+      "step": 712
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.555468157173576e-06,
+      "loss": 0.001,
+      "step": 713
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.531456556347713e-06,
+      "loss": 0.0025,
+      "step": 714
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.507447662583998e-06,
+      "loss": 0.0041,
+      "step": 715
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.483441614596495e-06,
+      "loss": 0.0099,
+      "step": 716
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.459438551082841e-06,
+      "loss": 0.0037,
+      "step": 717
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.435438610723415e-06,
+      "loss": 0.0023,
+      "step": 718
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.411441932180555e-06,
+      "loss": 0.007,
+      "step": 719
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.387448654097757e-06,
+      "loss": 0.0134,
+      "step": 720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.36345891509887e-06,
+      "loss": 0.0087,
+      "step": 721
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.339472853787289e-06,
+      "loss": 0.0009,
+      "step": 722
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.315490608745173e-06,
+      "loss": 0.0028,
+      "step": 723
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.291512318532615e-06,
+      "loss": 0.0173,
+      "step": 724
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.267538121686877e-06,
+      "loss": 0.0031,
+      "step": 725
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.243568156721561e-06,
+      "loss": 0.0155,
+      "step": 726
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.219602562125819e-06,
+      "loss": 0.004,
+      "step": 727
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.195641476363556e-06,
+      "loss": 0.0014,
+      "step": 728
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.171685037872621e-06,
+      "loss": 0.0014,
+      "step": 729
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.147733385064021e-06,
+      "loss": 0.005,
+      "step": 730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.123786656321105e-06,
+      "loss": 0.0038,
+      "step": 731
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.099844989998779e-06,
+      "loss": 0.0119,
+      "step": 732
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.075908524422695e-06,
+      "loss": 0.0072,
+      "step": 733
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.051977397888465e-06,
+      "loss": 0.0049,
+      "step": 734
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.028051748660845e-06,
+      "loss": 0.0058,
+      "step": 735
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.00413171497295e-06,
+      "loss": 0.0064,
+      "step": 736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.980217435025453e-06,
+      "loss": 0.004,
+      "step": 737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.956309046985775e-06,
+      "loss": 0.0125,
+      "step": 738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.93240668898731e-06,
+      "loss": 0.0074,
+      "step": 739
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.908510499128598e-06,
+      "loss": 0.0028,
+      "step": 740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.884620615472545e-06,
+      "loss": 0.0024,
+      "step": 741
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.860737176045633e-06,
+      "loss": 0.0084,
+      "step": 742
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.836860318837107e-06,
+      "loss": 0.0038,
+      "step": 743
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.81299018179817e-06,
+      "loss": 0.0053,
+      "step": 744
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.789126902841217e-06,
+      "loss": 0.0012,
+      "step": 745
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.765270619839004e-06,
+      "loss": 0.0083,
+      "step": 746
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.741421470623884e-06,
+      "loss": 0.006,
+      "step": 747
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.71757959298697e-06,
+      "loss": 0.0157,
+      "step": 748
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.693745124677386e-06,
+      "loss": 0.005,
+      "step": 749
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.669918203401441e-06,
+      "loss": 0.0039,
+      "step": 750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.64609896682183e-06,
+      "loss": 0.0026,
+      "step": 751
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.622287552556863e-06,
+      "loss": 0.0097,
+      "step": 752
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.59848409817964e-06,
+      "loss": 0.0209,
+      "step": 753
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.57468874121729e-06,
+      "loss": 0.0093,
+      "step": 754
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.550901619150143e-06,
+      "loss": 0.0174,
+      "step": 755
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.527122869410962e-06,
+      "loss": 0.0081,
+      "step": 756
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.503352629384131e-06,
+      "loss": 0.0084,
+      "step": 757
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.479591036404862e-06,
+      "loss": 0.0034,
+      "step": 758
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.455838227758432e-06,
+      "loss": 0.0034,
+      "step": 759
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.43209434067934e-06,
+      "loss": 0.0102,
+      "step": 760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.408359512350548e-06,
+      "loss": 0.0016,
+      "step": 761
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.384633879902685e-06,
+      "loss": 0.003,
+      "step": 762
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.360917580413245e-06,
+      "loss": 0.0063,
+      "step": 763
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.3372107509058e-06,
+      "loss": 0.0009,
+      "step": 764
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.313513528349208e-06,
+      "loss": 0.0062,
+      "step": 765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.289826049656821e-06,
+      "loss": 0.0014,
+      "step": 766
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.266148451685703e-06,
+      "loss": 0.0013,
+      "step": 767
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.242480871235819e-06,
+      "loss": 0.0011,
+      "step": 768
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.218823445049265e-06,
+      "loss": 0.0048,
+      "step": 769
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.195176309809465e-06,
+      "loss": 0.0016,
+      "step": 770
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.171539602140389e-06,
+      "loss": 0.002,
+      "step": 771
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.147913458605755e-06,
+      "loss": 0.0025,
+      "step": 772
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.124298015708254e-06,
+      "loss": 0.0036,
+      "step": 773
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.100693409888748e-06,
+      "loss": 0.0088,
+      "step": 774
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.07709977752548e-06,
+      "loss": 0.0009,
+      "step": 775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.05351725493331e-06,
+      "loss": 0.0033,
+      "step": 776
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.029945978362899e-06,
+      "loss": 0.0006,
+      "step": 777
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.006386083999925e-06,
+      "loss": 0.0013,
+      "step": 778
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.982837707964322e-06,
+      "loss": 0.0047,
+      "step": 779
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.959300986309459e-06,
+      "loss": 0.001,
+      "step": 780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.935776055021382e-06,
+      "loss": 0.001,
+      "step": 781
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.91226305001801e-06,
+      "loss": 0.0014,
+      "step": 782
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.888762107148357e-06,
+      "loss": 0.0021,
+      "step": 783
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.86527336219176e-06,
+      "loss": 0.0033,
+      "step": 784
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.841796950857056e-06,
+      "loss": 0.0009,
+      "step": 785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.818333008781847e-06,
+      "loss": 0.0141,
+      "step": 786
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.794881671531678e-06,
+      "loss": 0.001,
+      "step": 787
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.771443074599275e-06,
+      "loss": 0.0012,
+      "step": 788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.74801735340375e-06,
+      "loss": 0.0064,
+      "step": 789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.724604643289829e-06,
+      "loss": 0.0012,
+      "step": 790
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.70120507952706e-06,
+      "loss": 0.0032,
+      "step": 791
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.677818797309044e-06,
+      "loss": 0.0139,
+      "step": 792
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.654445931752646e-06,
+      "loss": 0.0037,
+      "step": 793
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.631086617897203e-06,
+      "loss": 0.0035,
+      "step": 794
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.607740990703764e-06,
+      "loss": 0.0027,
+      "step": 795
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.584409185054305e-06,
+      "loss": 0.0155,
+      "step": 796
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.5610913357509395e-06,
+      "loss": 0.0165,
+      "step": 797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.537787577515152e-06,
+      "loss": 0.0016,
+      "step": 798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.51449804498701e-06,
+      "loss": 0.0064,
+      "step": 799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.4912228727243905e-06,
+      "loss": 0.0049,
+      "step": 800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.467962195202211e-06,
+      "loss": 0.0059,
+      "step": 801
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.444716146811633e-06,
+      "loss": 0.0158,
+      "step": 802
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4214848618593006e-06,
+      "loss": 0.0017,
+      "step": 803
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.3982684745665614e-06,
+      "loss": 0.0017,
+      "step": 804
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.375067119068692e-06,
+      "loss": 0.0108,
+      "step": 805
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.351880929414113e-06,
+      "loss": 0.0085,
+      "step": 806
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.3287100395636355e-06,
+      "loss": 0.0017,
+      "step": 807
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.305554583389658e-06,
+      "loss": 0.0075,
+      "step": 808
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.282414694675426e-06,
+      "loss": 0.0066,
+      "step": 809
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.259290507114234e-06,
+      "loss": 0.0007,
+      "step": 810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.236182154308658e-06,
+      "loss": 0.008,
+      "step": 811
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.213089769769795e-06,
+      "loss": 0.0106,
+      "step": 812
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.190013486916476e-06,
+      "loss": 0.003,
+      "step": 813
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.1669534390745045e-06,
+      "loss": 0.0033,
+      "step": 814
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.143909759475892e-06,
+      "loss": 0.0017,
+      "step": 815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.1208825812580665e-06,
+      "loss": 0.0059,
+      "step": 816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.0978720374631326e-06,
+      "loss": 0.0012,
+      "step": 817
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.07487826103708e-06,
+      "loss": 0.0164,
+      "step": 818
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.051901384829021e-06,
+      "loss": 0.01,
+      "step": 819
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.028941541590428e-06,
+      "loss": 0.0066,
+      "step": 820
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.005998863974363e-06,
+      "loss": 0.0043,
+      "step": 821
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.983073484534715e-06,
+      "loss": 0.0013,
+      "step": 822
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.960165535725423e-06,
+      "loss": 0.003,
+      "step": 823
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.937275149899726e-06,
+      "loss": 0.0013,
+      "step": 824
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.914402459309384e-06,
+      "loss": 0.0095,
+      "step": 825
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.891547596103931e-06,
+      "loss": 0.0008,
+      "step": 826
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.868710692329895e-06,
+      "loss": 0.0017,
+      "step": 827
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.845891879930038e-06,
+      "loss": 0.0015,
+      "step": 828
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.823091290742603e-06,
+      "loss": 0.0041,
+      "step": 829
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.8003090565005445e-06,
+      "loss": 0.0081,
+      "step": 830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.777545308830765e-06,
+      "loss": 0.0009,
+      "step": 831
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.754800179253367e-06,
+      "loss": 0.001,
+      "step": 832
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.732073799180876e-06,
+      "loss": 0.0013,
+      "step": 833
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.709366299917497e-06,
+      "loss": 0.003,
+      "step": 834
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.686677812658351e-06,
+      "loss": 0.0016,
+      "step": 835
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.664008468488703e-06,
+      "loss": 0.0059,
+      "step": 836
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.6413583983832304e-06,
+      "loss": 0.0002,
+      "step": 837
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.6187277332052415e-06,
+      "loss": 0.002,
+      "step": 838
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.59611660370594e-06,
+      "loss": 0.0021,
+      "step": 839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.573525140523651e-06,
+      "loss": 0.0107,
+      "step": 840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.5509534741830845e-06,
+      "loss": 0.0079,
+      "step": 841
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.52840173509456e-06,
+      "loss": 0.0004,
+      "step": 842
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.505870053553279e-06,
+      "loss": 0.0038,
+      "step": 843
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.483358559738551e-06,
+      "loss": 0.0011,
+      "step": 844
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.460867383713046e-06,
+      "loss": 0.0067,
+      "step": 845
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.438396655422052e-06,
+      "loss": 0.0011,
+      "step": 846
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.415946504692714e-06,
+      "loss": 0.0011,
+      "step": 847
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.393517061233286e-06,
+      "loss": 0.0028,
+      "step": 848
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.371108454632391e-06,
+      "loss": 0.0016,
+      "step": 849
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.348720814358251e-06,
+      "loss": 0.0033,
+      "step": 850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.32635426975797e-06,
+      "loss": 0.0011,
+      "step": 851
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.304008950056761e-06,
+      "loss": 0.0053,
+      "step": 852
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.281684984357204e-06,
+      "loss": 0.0016,
+      "step": 853
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.259382501638509e-06,
+      "loss": 0.0008,
+      "step": 854
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.237101630755762e-06,
+      "loss": 0.0057,
+      "step": 855
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.214842500439191e-06,
+      "loss": 0.0013,
+      "step": 856
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.192605239293407e-06,
+      "loss": 0.0008,
+      "step": 857
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.170389975796671e-06,
+      "loss": 0.0034,
+      "step": 858
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.14819683830016e-06,
+      "loss": 0.0053,
+      "step": 859
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.126025955027204e-06,
+      "loss": 0.0016,
+      "step": 860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.10387745407256e-06,
+      "loss": 0.0004,
+      "step": 861
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.0817514634016695e-06,
+      "loss": 0.0008,
+      "step": 862
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.05964811084992e-06,
+      "loss": 0.0091,
+      "step": 863
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.0375675241219e-06,
+      "loss": 0.0051,
+      "step": 864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.015509830790672e-06,
+      "loss": 0.0004,
+      "step": 865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.993475158297027e-06,
+      "loss": 0.0012,
+      "step": 866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.97146363394874e-06,
+      "loss": 0.0102,
+      "step": 867
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.949475384919865e-06,
+      "loss": 0.0004,
+      "step": 868
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.92751053824997e-06,
+      "loss": 0.0017,
+      "step": 869
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.905569220843406e-06,
+      "loss": 0.0144,
+      "step": 870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.8836515594685975e-06,
+      "loss": 0.0007,
+      "step": 871
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.861757680757275e-06,
+      "loss": 0.0033,
+      "step": 872
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.839887711203778e-06,
+      "loss": 0.0009,
+      "step": 873
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8180417771643e-06,
+      "loss": 0.0012,
+      "step": 874
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.796220004856161e-06,
+      "loss": 0.0003,
+      "step": 875
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.774422520357101e-06,
+      "loss": 0.0096,
+      "step": 876
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.752649449604526e-06,
+      "loss": 0.0007,
+      "step": 877
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.730900918394777e-06,
+      "loss": 0.0013,
+      "step": 878
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.709177052382432e-06,
+      "loss": 0.0021,
+      "step": 879
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.687477977079555e-06,
+      "loss": 0.0016,
+      "step": 880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.665803817854985e-06,
+      "loss": 0.0019,
+      "step": 881
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.644154699933593e-06,
+      "loss": 0.0019,
+      "step": 882
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.622530748395581e-06,
+      "loss": 0.0022,
+      "step": 883
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.60093208817575e-06,
+      "loss": 0.005,
+      "step": 884
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.579358844062774e-06,
+      "loss": 0.0013,
+      "step": 885
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.557811140698486e-06,
+      "loss": 0.0012,
+      "step": 886
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.536289102577152e-06,
+      "loss": 0.0005,
+      "step": 887
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.5147928540447486e-06,
+      "loss": 0.0099,
+      "step": 888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4933225192982586e-06,
+      "loss": 0.0017,
+      "step": 889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.471878222384944e-06,
+      "loss": 0.0004,
+      "step": 890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.450460087201632e-06,
+      "loss": 0.0002,
+      "step": 891
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.429068237493982e-06,
+      "loss": 0.0044,
+      "step": 892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.407702796855809e-06,
+      "loss": 0.0015,
+      "step": 893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.386363888728337e-06,
+      "loss": 0.0119,
+      "step": 894
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.365051636399485e-06,
+      "loss": 0.0002,
+      "step": 895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.343766163003181e-06,
+      "loss": 0.0006,
+      "step": 896
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.322507591518627e-06,
+      "loss": 0.0014,
+      "step": 897
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.301276044769599e-06,
+      "loss": 0.0009,
+      "step": 898
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.2800716454237266e-06,
+      "loss": 0.0007,
+      "step": 899
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.2588945159918e-06,
+      "loss": 0.004,
+      "step": 900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.237744778827063e-06,
+      "loss": 0.0083,
+      "step": 901
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.216622556124475e-06,
+      "loss": 0.0066,
+      "step": 902
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.195527969920047e-06,
+      "loss": 0.0017,
+      "step": 903
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.174461142090112e-06,
+      "loss": 0.001,
+      "step": 904
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.153422194350614e-06,
+      "loss": 0.0012,
+      "step": 905
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1324112482564345e-06,
+      "loss": 0.0031,
+      "step": 906
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.111428425200662e-06,
+      "loss": 0.0006,
+      "step": 907
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.090473846413908e-06,
+      "loss": 0.0004,
+      "step": 908
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0695476329635825e-06,
+      "loss": 0.0022,
+      "step": 909
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.048649905753238e-06,
+      "loss": 0.0002,
+      "step": 910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.027780785521828e-06,
+      "loss": 0.0002,
+      "step": 911
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.006940392843022e-06,
+      "loss": 0.0008,
+      "step": 912
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.986128848124523e-06,
+      "loss": 0.0206,
+      "step": 913
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.965346271607359e-06,
+      "loss": 0.0045,
+      "step": 914
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.944592783365194e-06,
+      "loss": 0.0027,
+      "step": 915
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.923868503303619e-06,
+      "loss": 0.0007,
+      "step": 916
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.903173551159481e-06,
+      "loss": 0.003,
+      "step": 917
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.882508046500191e-06,
+      "loss": 0.0019,
+      "step": 918
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.861872108723001e-06,
+      "loss": 0.0072,
+      "step": 919
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.841265857054356e-06,
+      "loss": 0.0023,
+      "step": 920
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.820689410549181e-06,
+      "loss": 0.0047,
+      "step": 921
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.80014288809019e-06,
+      "loss": 0.0046,
+      "step": 922
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7796264083872194e-06,
+      "loss": 0.0024,
+      "step": 923
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.759140089976524e-06,
+      "loss": 0.0018,
+      "step": 924
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7386840512201e-06,
+      "loss": 0.0067,
+      "step": 925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.718258410304999e-06,
+      "loss": 0.012,
+      "step": 926
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.697863285242645e-06,
+      "loss": 0.003,
+      "step": 927
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.67749879386816e-06,
+      "loss": 0.0001,
+      "step": 928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6571650538396615e-06,
+      "loss": 0.0044,
+      "step": 929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6368621826376115e-06,
+      "loss": 0.0038,
+      "step": 930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.616590297564116e-06,
+      "loss": 0.0051,
+      "step": 931
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.596349515742266e-06,
+      "loss": 0.0013,
+      "step": 932
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.576139954115432e-06,
+      "loss": 0.0007,
+      "step": 933
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.555961729446617e-06,
+      "loss": 0.0026,
+      "step": 934
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.535814958317783e-06,
+      "loss": 0.0019,
+      "step": 935
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5156997571291425e-06,
+      "loss": 0.0008,
+      "step": 936
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.495616242098523e-06,
+      "loss": 0.0004,
+      "step": 937
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.475564529260681e-06,
+      "loss": 0.0001,
+      "step": 938
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.455544734466624e-06,
+      "loss": 0.0006,
+      "step": 939
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.43555697338296e-06,
+      "loss": 0.0004,
+      "step": 940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.415601361491213e-06,
+      "loss": 0.0101,
+      "step": 941
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.395678014087162e-06,
+      "loss": 0.0009,
+      "step": 942
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.375787046280177e-06,
+      "loss": 0.0018,
+      "step": 943
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.355928572992547e-06,
+      "loss": 0.0018,
+      "step": 944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.336102708958827e-06,
+      "loss": 0.0029,
+      "step": 945
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.316309568725156e-06,
+      "loss": 0.0014,
+      "step": 946
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.296549266648621e-06,
+      "loss": 0.003,
+      "step": 947
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.276821916896574e-06,
+      "loss": 0.0038,
+      "step": 948
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.25712763344599e-06,
+      "loss": 0.0051,
+      "step": 949
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2374665300827865e-06,
+      "loss": 0.0064,
+      "step": 950
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.217838720401188e-06,
+      "loss": 0.0024,
+      "step": 951
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.198244317803071e-06,
+      "loss": 0.0007,
+      "step": 952
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.178683435497275e-06,
+      "loss": 0.0045,
+      "step": 953
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1591561864989905e-06,
+      "loss": 0.0076,
+      "step": 954
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.139662683629087e-06,
+      "loss": 0.0025,
+      "step": 955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.120203039513448e-06,
+      "loss": 0.0006,
+      "step": 956
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.100777366582349e-06,
+      "loss": 0.0001,
+      "step": 957
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.081385777069789e-06,
+      "loss": 0.0008,
+      "step": 958
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.062028383012841e-06,
+      "loss": 0.0004,
+      "step": 959
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.042705296251018e-06,
+      "loss": 0.0001,
+      "step": 960
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.023416628425614e-06,
+      "loss": 0.0063,
+      "step": 961
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.004162490979067e-06,
+      "loss": 0.0008,
+      "step": 962
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.984942995154305e-06,
+      "loss": 0.0056,
+      "step": 963
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.965758251994115e-06,
+      "loss": 0.0115,
+      "step": 964
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.946608372340498e-06,
+      "loss": 0.0006,
+      "step": 965
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.927493466834028e-06,
+      "loss": 0.0015,
+      "step": 966
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.908413645913201e-06,
+      "loss": 0.0149,
+      "step": 967
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.889369019813823e-06,
+      "loss": 0.0024,
+      "step": 968
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.870359698568355e-06,
+      "loss": 0.0037,
+      "step": 969
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.851385792005269e-06,
+      "loss": 0.0009,
+      "step": 970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8324474097484335e-06,
+      "loss": 0.0009,
+      "step": 971
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.813544661216475e-06,
+      "loss": 0.0003,
+      "step": 972
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.79467765562213e-06,
+      "loss": 0.0052,
+      "step": 973
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.775846501971636e-06,
+      "loss": 0.0029,
+      "step": 974
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7570513090640915e-06,
+      "loss": 0.0025,
+      "step": 975
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.738292185490825e-06,
+      "loss": 0.0018,
+      "step": 976
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7195692396347693e-06,
+      "loss": 0.0001,
+      "step": 977
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7008825796698388e-06,
+      "loss": 0.0004,
+      "step": 978
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6822323135603054e-06,
+      "loss": 0.0002,
+      "step": 979
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6636185490601607e-06,
+      "loss": 0.0001,
+      "step": 980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.645041393712514e-06,
+      "loss": 0.0007,
+      "step": 981
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.626500954848957e-06,
+      "loss": 0.0004,
+      "step": 982
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6079973395889545e-06,
+      "loss": 0.0024,
+      "step": 983
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5895306548392006e-06,
+      "loss": 0.0177,
+      "step": 984
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.571101007293043e-06,
+      "loss": 0.0017,
+      "step": 985
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5527085034298314e-06,
+      "loss": 0.0011,
+      "step": 986
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5343532495143075e-06,
+      "loss": 0.0002,
+      "step": 987
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5160353515960053e-06,
+      "loss": 0.005,
+      "step": 988
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.497754915508632e-06,
+      "loss": 0.002,
+      "step": 989
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.479512046869452e-06,
+      "loss": 0.0014,
+      "step": 990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.461306851078674e-06,
+      "loss": 0.0122,
+      "step": 991
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.443139433318855e-06,
+      "loss": 0.0039,
+      "step": 992
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4250098985542847e-06,
+      "loss": 0.0077,
+      "step": 993
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.406918351530376e-06,
+      "loss": 0.0011,
+      "step": 994
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3888648967730654e-06,
+      "loss": 0.0015,
+      "step": 995
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3708496385882117e-06,
+      "loss": 0.0003,
+      "step": 996
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.352872681060976e-06,
+      "loss": 0.0008,
+      "step": 997
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.334934128055245e-06,
+      "loss": 0.0006,
+      "step": 998
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3170340832130134e-06,
+      "loss": 0.0013,
+      "step": 999
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2991726499537967e-06,
+      "loss": 0.0017,
+      "step": 1000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.281349931474015e-06,
+      "loss": 0.0128,
+      "step": 1001
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2635660307464247e-06,
+      "loss": 0.0013,
+      "step": 1002
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.245821050519501e-06,
+      "loss": 0.0003,
+      "step": 1003
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.228115093316848e-06,
+      "loss": 0.0062,
+      "step": 1004
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.210448261436615e-06,
+      "loss": 0.0008,
+      "step": 1005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1928206569508992e-06,
+      "loss": 0.002,
+      "step": 1006
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.175232381705161e-06,
+      "loss": 0.0017,
+      "step": 1007
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1576835373176206e-06,
+      "loss": 0.0032,
+      "step": 1008
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1401742251786926e-06,
+      "loss": 0.0043,
+      "step": 1009
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1227045464503856e-06,
+      "loss": 0.0013,
+      "step": 1010
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.105274602065723e-06,
+      "loss": 0.0119,
+      "step": 1011
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.087884492728158e-06,
+      "loss": 0.0046,
+      "step": 1012
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0705343189109948e-06,
+      "loss": 0.0002,
+      "step": 1013
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0532241808567966e-06,
+      "loss": 0.0004,
+      "step": 1014
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0359541785768255e-06,
+      "loss": 0.0009,
+      "step": 1015
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.018724411850451e-06,
+      "loss": 0.0022,
+      "step": 1016
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0015349802245817e-06,
+      "loss": 0.0002,
+      "step": 1017
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9843859830130696e-06,
+      "loss": 0.0001,
+      "step": 1018
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9672775192961755e-06,
+      "loss": 0.0022,
+      "step": 1019
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9502096879199606e-06,
+      "loss": 0.0003,
+      "step": 1020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.933182587495724e-06,
+      "loss": 0.0003,
+      "step": 1021
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9161963163994454e-06,
+      "loss": 0.0004,
+      "step": 1022
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.899250972771207e-06,
+      "loss": 0.0002,
+      "step": 1023
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.882346654514627e-06,
+      "loss": 0.0002,
+      "step": 1024
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8654834592962923e-06,
+      "loss": 0.0005,
+      "step": 1025
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8486614845451965e-06,
+      "loss": 0.0002,
+      "step": 1026
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8318808274521915e-06,
+      "loss": 0.0001,
+      "step": 1027
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8151415849693886e-06,
+      "loss": 0.0004,
+      "step": 1028
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.7984438538096393e-06,
+      "loss": 0.0006,
+      "step": 1029
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.781787730445953e-06,
+      "loss": 0.0029,
+      "step": 1030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7651733111109415e-06,
+      "loss": 0.0011,
+      "step": 1031
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7486006917962727e-06,
+      "loss": 0.0003,
+      "step": 1032
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.732069968252108e-06,
+      "loss": 0.0002,
+      "step": 1033
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.715581235986552e-06,
+      "loss": 0.0019,
+      "step": 1034
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.699134590265099e-06,
+      "loss": 0.0011,
+      "step": 1035
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6827301261100846e-06,
+      "loss": 0.0018,
+      "step": 1036
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.666367938300137e-06,
+      "loss": 0.0029,
+      "step": 1037
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6500481213696227e-06,
+      "loss": 0.0002,
+      "step": 1038
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6337707696081093e-06,
+      "loss": 0.0001,
+      "step": 1039
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6175359770598195e-06,
+      "loss": 0.0006,
+      "step": 1040
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6013438375230836e-06,
+      "loss": 0.0009,
+      "step": 1041
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5851944445497955e-06,
+      "loss": 0.0004,
+      "step": 1042
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.569087891444877e-06,
+      "loss": 0.0002,
+      "step": 1043
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5530242712657494e-06,
+      "loss": 0.0005,
+      "step": 1044
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5370036768217666e-06,
+      "loss": 0.0006,
+      "step": 1045
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5210262006737074e-06,
+      "loss": 0.0009,
+      "step": 1046
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5050919351332303e-06,
+      "loss": 0.0002,
+      "step": 1047
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.48920097226233e-06,
+      "loss": 0.0008,
+      "step": 1048
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.473353403872826e-06,
+      "loss": 0.0011,
+      "step": 1049
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.457549321525815e-06,
+      "loss": 0.0003,
+      "step": 1050
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4417888165311533e-06,
+      "loss": 0.0008,
+      "step": 1051
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.426071979946918e-06,
+      "loss": 0.0001,
+      "step": 1052
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4103989025788923e-06,
+      "loss": 0.0011,
+      "step": 1053
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.394769674980035e-06,
+      "loss": 0.0006,
+      "step": 1054
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.379184387449952e-06,
+      "loss": 0.0001,
+      "step": 1055
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3636431300343875e-06,
+      "loss": 0.0004,
+      "step": 1056
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.348145992524694e-06,
+      "loss": 0.0008,
+      "step": 1057
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.332693064457321e-06,
+      "loss": 0.0002,
+      "step": 1058
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3172844351132786e-06,
+      "loss": 0.0007,
+      "step": 1059
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.301920193517655e-06,
+      "loss": 0.0007,
+      "step": 1060
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.286600428439074e-06,
+      "loss": 0.0004,
+      "step": 1061
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2713252283891873e-06,
+      "loss": 0.0002,
+      "step": 1062
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.256094681622174e-06,
+      "loss": 0.0005,
+      "step": 1063
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2409088761342234e-06,
+      "loss": 0.0002,
+      "step": 1064
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2257678996630193e-06,
+      "loss": 0.0063,
+      "step": 1065
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.210671839687252e-06,
+      "loss": 0.0002,
+      "step": 1066
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.195620783426092e-06,
+      "loss": 0.0002,
+      "step": 1067
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1806148178387043e-06,
+      "loss": 0.0003,
+      "step": 1068
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1656540296237316e-06,
+      "loss": 0.0001,
+      "step": 1069
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.150738505218801e-06,
+      "loss": 0.001,
+      "step": 1070
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1358683308000263e-06,
+      "loss": 0.0009,
+      "step": 1071
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1210435922814964e-06,
+      "loss": 0.0003,
+      "step": 1072
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1062643753148003e-06,
+      "loss": 0.0002,
+      "step": 1073
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0915307652885163e-06,
+      "loss": 0.0009,
+      "step": 1074
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.076842847327727e-06,
+      "loss": 0.0007,
+      "step": 1075
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0622007062935123e-06,
+      "loss": 0.0005,
+      "step": 1076
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.047604426782489e-06,
+      "loss": 0.0001,
+      "step": 1077
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.033054093126294e-06,
+      "loss": 0.0007,
+      "step": 1078
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.018549789391102e-06,
+      "loss": 0.0004,
+      "step": 1079
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0040915993771513e-06,
+      "loss": 0.0002,
+      "step": 1080
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9896796066182567e-06,
+      "loss": 0.0003,
+      "step": 1081
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.975313894381311e-06,
+      "loss": 0.0013,
+      "step": 1082
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9609945456658276e-06,
+      "loss": 0.0003,
+      "step": 1083
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.946721643203443e-06,
+      "loss": 0.0004,
+      "step": 1084
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9324952694574493e-06,
+      "loss": 0.0003,
+      "step": 1085
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9183155066223113e-06,
+      "loss": 0.0001,
+      "step": 1086
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.9041824366231953e-06,
+      "loss": 0.0003,
+      "step": 1087
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.890096141115495e-06,
+      "loss": 0.0008,
+      "step": 1088
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8760567014843545e-06,
+      "loss": 0.0001,
+      "step": 1089
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.862064198844208e-06,
+      "loss": 0.0001,
+      "step": 1090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.848118714038305e-06,
+      "loss": 0.001,
+      "step": 1091
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8342203276382454e-06,
+      "loss": 0.0008,
+      "step": 1092
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8203691199435037e-06,
+      "loss": 0.0003,
+      "step": 1093
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8065651709809906e-06,
+      "loss": 0.0003,
+      "step": 1094
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.792808560504563e-06,
+      "loss": 0.0001,
+      "step": 1095
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7790993679945723e-06,
+      "loss": 0.0009,
+      "step": 1096
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.765437672657414e-06,
+      "loss": 0.0001,
+      "step": 1097
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7518235534250639e-06,
+      "loss": 0.0004,
+      "step": 1098
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7382570889546124e-06,
+      "loss": 0.0004,
+      "step": 1099
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7247383576278277e-06,
+      "loss": 0.0004,
+      "step": 1100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7112674375506931e-06,
+      "loss": 0.0011,
+      "step": 1101
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6978444065529554e-06,
+      "loss": 0.0001,
+      "step": 1102
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6844693421876778e-06,
+      "loss": 0.0003,
+      "step": 1103
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6711423217307888e-06,
+      "loss": 0.0002,
+      "step": 1104
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6578634221806445e-06,
+      "loss": 0.0001,
+      "step": 1105
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6446327202575662e-06,
+      "loss": 0.0001,
+      "step": 1106
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6314502924034158e-06,
+      "loss": 0.0001,
+      "step": 1107
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6183162147811448e-06,
+      "loss": 0.0001,
+      "step": 1108
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6052305632743592e-06,
+      "loss": 0.0002,
+      "step": 1109
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5921934134868655e-06,
+      "loss": 0.0001,
+      "step": 1110
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5792048407422633e-06,
+      "loss": 0.0034,
+      "step": 1111
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5662649200834822e-06,
+      "loss": 0.0,
+      "step": 1112
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.553373726272358e-06,
+      "loss": 0.0002,
+      "step": 1113
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.540531333789207e-06,
+      "loss": 0.0005,
+      "step": 1114
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5277378168323886e-06,
+      "loss": 0.0,
+      "step": 1115
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5149932493178754e-06,
+      "loss": 0.0001,
+      "step": 1116
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5022977048788335e-06,
+      "loss": 0.0002,
+      "step": 1117
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4896512568651888e-06,
+      "loss": 0.0001,
+      "step": 1118
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4770539783432113e-06,
+      "loss": 0.0004,
+      "step": 1119
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.464505942095087e-06,
+      "loss": 0.001,
+      "step": 1120
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4520072206184954e-06,
+      "loss": 0.0002,
+      "step": 1121
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4395578861262037e-06,
+      "loss": 0.0002,
+      "step": 1122
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4271580105456273e-06,
+      "loss": 0.0002,
+      "step": 1123
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4148076655184373e-06,
+      "loss": 0.0001,
+      "step": 1124
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4025069224001342e-06,
+      "loss": 0.0003,
+      "step": 1125
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3902558522596355e-06,
+      "loss": 0.0002,
+      "step": 1126
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3780545258788703e-06,
+      "loss": 0.0003,
+      "step": 1127
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3659030137523655e-06,
+      "loss": 0.0004,
+      "step": 1128
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3538013860868438e-06,
+      "loss": 0.0002,
+      "step": 1129
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3417497128008084e-06,
+      "loss": 0.0001,
+      "step": 1130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3297480635241522e-06,
+      "loss": 0.0001,
+      "step": 1131
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.317796507597745e-06,
+      "loss": 0.0001,
+      "step": 1132
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3058951140730415e-06,
+      "loss": 0.0006,
+      "step": 1133
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2940439517116677e-06,
+      "loss": 0.0002,
+      "step": 1134
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2822430889850434e-06,
+      "loss": 0.0002,
+      "step": 1135
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.270492594073972e-06,
+      "loss": 0.0003,
+      "step": 1136
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2587925348682517e-06,
+      "loss": 0.0001,
+      "step": 1137
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2471429789662847e-06,
+      "loss": 0.0001,
+      "step": 1138
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2355439936746826e-06,
+      "loss": 0.0,
+      "step": 1139
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2239956460078795e-06,
+      "loss": 0.0,
+      "step": 1140
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2124980026877442e-06,
+      "loss": 0.0001,
+      "step": 1141
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2010511301431981e-06,
+      "loss": 0.0004,
+      "step": 1142
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1896550945098295e-06,
+      "loss": 0.0014,
+      "step": 1143
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1783099616295058e-06,
+      "loss": 0.0002,
+      "step": 1144
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.167015797050003e-06,
+      "loss": 0.0001,
+      "step": 1145
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1557726660246205e-06,
+      "loss": 0.0013,
+      "step": 1146
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1445806335118037e-06,
+      "loss": 0.0001,
+      "step": 1147
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1334397641747718e-06,
+      "loss": 0.0001,
+      "step": 1148
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.122350122381145e-06,
+      "loss": 0.0002,
+      "step": 1149
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.111311772202569e-06,
+      "loss": 0.0002,
+      "step": 1150
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1003247774143455e-06,
+      "loss": 0.0013,
+      "step": 1151
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0893892014950637e-06,
+      "loss": 0.0001,
+      "step": 1152
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.078505107626242e-06,
+      "loss": 0.0001,
+      "step": 1153
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0676725586919456e-06,
+      "loss": 0.0025,
+      "step": 1154
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0568916172784383e-06,
+      "loss": 0.0003,
+      "step": 1155
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0461623456738167e-06,
+      "loss": 0.0003,
+      "step": 1156
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0354848058676448e-06,
+      "loss": 0.0001,
+      "step": 1157
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0248590595506046e-06,
+      "loss": 0.0018,
+      "step": 1158
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.014285168114133e-06,
+      "loss": 0.0002,
+      "step": 1159
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0037631926500712e-06,
+      "loss": 0.0001,
+      "step": 1160
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.932931939503088e-07,
+      "loss": 0.0003,
+      "step": 1161
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.828752325064339e-07,
+      "loss": 0.0,
+      "step": 1162
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.725093685093867e-07,
+      "loss": 0.0001,
+      "step": 1163
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.621956618491024e-07,
+      "loss": 0.0001,
+      "step": 1164
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.519341721141751e-07,
+      "loss": 0.0004,
+      "step": 1165
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.41724958591509e-07,
+      "loss": 0.0001,
+      "step": 1166
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.31568080265981e-07,
+      "loss": 0.0003,
+      "step": 1167
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.214635958200846e-07,
+      "loss": 0.0013,
+      "step": 1168
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.114115636336151e-07,
+      "loss": 0.0001,
+      "step": 1169
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.014120417833116e-07,
+      "loss": 0.0,
+      "step": 1170
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.91465088042528e-07,
+      "loss": 0.0002,
+      "step": 1171
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.815707598809021e-07,
+      "loss": 0.0002,
+      "step": 1172
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.717291144640238e-07,
+      "loss": 0.0001,
+      "step": 1173
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.61940208653097e-07,
+      "loss": 0.0001,
+      "step": 1174
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.52204099004621e-07,
+      "loss": 0.0001,
+      "step": 1175
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.425208417700581e-07,
+      "loss": 0.0002,
+      "step": 1176
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.328904928955117e-07,
+      "loss": 0.0001,
+      "step": 1177
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.233131080213975e-07,
+      "loss": 0.0005,
+      "step": 1178
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.137887424821278e-07,
+      "loss": 0.0003,
+      "step": 1179
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.043174513057927e-07,
+      "loss": 0.0003,
+      "step": 1180
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.948992892138318e-07,
+      "loss": 0.0004,
+      "step": 1181
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.855343106207325e-07,
+      "loss": 0.001,
+      "step": 1182
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.762225696337056e-07,
+      "loss": 0.0013,
+      "step": 1183
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.66964120052377e-07,
+      "loss": 0.0055,
+      "step": 1184
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.57759015368471e-07,
+      "loss": 0.0001,
+      "step": 1185
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.486073087655133e-07,
+      "loss": 0.0001,
+      "step": 1186
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.395090531185123e-07,
+      "loss": 0.0002,
+      "step": 1187
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.304643009936552e-07,
+      "loss": 0.0001,
+      "step": 1188
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.214731046480095e-07,
+      "loss": 0.0004,
+      "step": 1189
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.125355160292202e-07,
+      "loss": 0.0001,
+      "step": 1190
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.036515867752014e-07,
+      "loss": 0.0001,
+      "step": 1191
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.948213682138505e-07,
+      "loss": 0.0008,
+      "step": 1192
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.86044911362741e-07,
+      "loss": 0.0,
+      "step": 1193
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.773222669288359e-07,
+      "loss": 0.0,
+      "step": 1194
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.686534853081872e-07,
+      "loss": 0.0001,
+      "step": 1195
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.600386165856521e-07,
+      "loss": 0.0001,
+      "step": 1196
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.514777105345981e-07,
+      "loss": 0.0006,
+      "step": 1197
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.429708166166148e-07,
+      "loss": 0.0005,
+      "step": 1198
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.345179839812343e-07,
+      "loss": 0.0,
+      "step": 1199
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.261192614656442e-07,
+      "loss": 0.0002,
+      "step": 1200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.177746975944021e-07,
+      "loss": 0.0013,
+      "step": 1201
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.094843405791573e-07,
+      "loss": 0.0002,
+      "step": 1202
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.012482383183771e-07,
+      "loss": 0.0,
+      "step": 1203
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.930664383970642e-07,
+      "loss": 0.0002,
+      "step": 1204
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.849389880864787e-07,
+      "loss": 0.0001,
+      "step": 1205
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.768659343438764e-07,
+      "loss": 0.0,
+      "step": 1206
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.688473238122305e-07,
+      "loss": 0.0001,
+      "step": 1207
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.608832028199552e-07,
+      "loss": 0.0,
+      "step": 1208
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.529736173806533e-07,
+      "loss": 0.0008,
+      "step": 1209
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.451186131928387e-07,
+      "loss": 0.0,
+      "step": 1210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.373182356396756e-07,
+      "loss": 0.0,
+      "step": 1211
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.2957252978872e-07,
+      "loss": 0.0001,
+      "step": 1212
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.218815403916511e-07,
+      "loss": 0.0021,
+      "step": 1213
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.142453118840241e-07,
+      "loss": 0.0002,
+      "step": 1214
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.066638883849983e-07,
+      "loss": 0.0007,
+      "step": 1215
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.99137313697099e-07,
+      "loss": 0.0014,
+      "step": 1216
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.91665631305952e-07,
+      "loss": 0.0,
+      "step": 1217
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.842488843800409e-07,
+      "loss": 0.0001,
+      "step": 1218
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.768871157704436e-07,
+      "loss": 0.0001,
+      "step": 1219
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6958036801060926e-07,
+      "loss": 0.0,
+      "step": 1220
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6232868331608784e-07,
+      "loss": 0.0,
+      "step": 1221
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5513210358429707e-07,
+      "loss": 0.0,
+      "step": 1222
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.479906703942816e-07,
+      "loss": 0.0008,
+      "step": 1223
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.40904425006472e-07,
+      "loss": 0.0,
+      "step": 1224
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3387340836244183e-07,
+      "loss": 0.0001,
+      "step": 1225
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2689766108467643e-07,
+      "loss": 0.0005,
+      "step": 1226
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.199772234763366e-07,
+      "loss": 0.0003,
+      "step": 1227
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1311213552102324e-07,
+      "loss": 0.0001,
+      "step": 1228
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0630243688255187e-07,
+      "loss": 0.0,
+      "step": 1229
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9954816690471633e-07,
+      "loss": 0.0007,
+      "step": 1230
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9284936461106873e-07,
+      "loss": 0.0004,
+      "step": 1231
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.8620606870468535e-07,
+      "loss": 0.0005,
+      "step": 1232
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7961831756795353e-07,
+      "loss": 0.0002,
+      "step": 1233
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7308614926234164e-07,
+      "loss": 0.0001,
+      "step": 1234
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.666096015281828e-07,
+      "loss": 0.0001,
+      "step": 1235
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6018871178445604e-07,
+      "loss": 0.0001,
+      "step": 1236
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.538235171285698e-07,
+      "loss": 0.0066,
+      "step": 1237
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4751405433614884e-07,
+      "loss": 0.0,
+      "step": 1238
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4126035986081884e-07,
+      "loss": 0.0001,
+      "step": 1239
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.350624698339988e-07,
+      "loss": 0.0006,
+      "step": 1240
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2892042006469447e-07,
+      "loss": 0.0001,
+      "step": 1241
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.228342460392808e-07,
+      "loss": 0.0004,
+      "step": 1242
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.168039829213121e-07,
+      "loss": 0.0001,
+      "step": 1243
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.108296655513066e-07,
+      "loss": 0.0,
+      "step": 1244
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.049113284465521e-07,
+      "loss": 0.0001,
+      "step": 1245
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.990490058009021e-07,
+      "loss": 0.0,
+      "step": 1246
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9324273148458205e-07,
+      "loss": 0.0003,
+      "step": 1247
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8749253904399197e-07,
+      "loss": 0.0,
+      "step": 1248
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8179846170150906e-07,
+      "loss": 0.0,
+      "step": 1249
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7616053235530515e-07,
+      "loss": 0.0001,
+      "step": 1250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7057878357914513e-07,
+      "loss": 0.0017,
+      "step": 1251
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.6505324762220783e-07,
+      "loss": 0.001,
+      "step": 1252
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.595839564088953e-07,
+      "loss": 0.0001,
+      "step": 1253
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.541709415386495e-07,
+      "loss": 0.0004,
+      "step": 1254
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4881423428577025e-07,
+      "loss": 0.0,
+      "step": 1255
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.4351386559922973e-07,
+      "loss": 0.0015,
+      "step": 1256
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3826986610250292e-07,
+      "loss": 0.0001,
+      "step": 1257
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3308226609338401e-07,
+      "loss": 0.0009,
+      "step": 1258
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2795109554381024e-07,
+      "loss": 0.0002,
+      "step": 1259
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2287638409969393e-07,
+      "loss": 0.0,
+      "step": 1260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.178581610807484e-07,
+      "loss": 0.0,
+      "step": 1261
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1289645548031923e-07,
+      "loss": 0.0002,
+      "step": 1262
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0799129596521417e-07,
+      "loss": 0.0002,
+      "step": 1263
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0314271087554126e-07,
+      "loss": 0.0,
+      "step": 1264
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9835072822454448e-07,
+      "loss": 0.0001,
+      "step": 1265
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9361537569843712e-07,
+      "loss": 0.0001,
+      "step": 1266
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.889366806562487e-07,
+      "loss": 0.0002,
+      "step": 1267
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8431467012966387e-07,
+      "loss": 0.0001,
+      "step": 1268
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7974937082286593e-07,
+      "loss": 0.0002,
+      "step": 1269
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7524080911238028e-07,
+      "loss": 0.0001,
+      "step": 1270
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7078901104692792e-07,
+      "loss": 0.0,
+      "step": 1271
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.66394002347271e-07,
+      "loss": 0.0003,
+      "step": 1272
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6205580840606084e-07,
+      "loss": 0.0001,
+      "step": 1273
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5777445428770022e-07,
+      "loss": 0.0008,
+      "step": 1274
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5354996472819128e-07,
+      "loss": 0.0001,
+      "step": 1275
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4938236413499453e-07,
+      "loss": 0.0003,
+      "step": 1276
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.452716765868878e-07,
+      "loss": 0.0001,
+      "step": 1277
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4121792583382975e-07,
+      "loss": 0.0008,
+      "step": 1278
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.372211352968167e-07,
+      "loss": 0.0,
+      "step": 1279
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3328132806775473e-07,
+      "loss": 0.0003,
+      "step": 1280
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.293985269093212e-07,
+      "loss": 0.0,
+      "step": 1281
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.255727542548324e-07,
+      "loss": 0.0,
+      "step": 1282
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.218040322081182e-07,
+      "loss": 0.0002,
+      "step": 1283
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1809238254339106e-07,
+      "loss": 0.0,
+      "step": 1284
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1443782670512383e-07,
+      "loss": 0.0001,
+      "step": 1285
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1084038580791989e-07,
+      "loss": 0.0002,
+      "step": 1286
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0730008063639774e-07,
+      "loss": 0.0001,
+      "step": 1287
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0381693164506546e-07,
+      "loss": 0.0,
+      "step": 1288
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0039095895820639e-07,
+      "loss": 0.0006,
+      "step": 1289
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.702218236976147e-08,
+      "loss": 0.0005,
+      "step": 1290
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.371062134321263e-08,
+      "loss": 0.0,
+      "step": 1291
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.045629501147401e-08,
+      "loss": 0.0,
+      "step": 1292
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.725922217678206e-08,
+      "loss": 0.0001,
+      "step": 1293
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.411942131058115e-08,
+      "loss": 0.0005,
+      "step": 1294
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.103691055342145e-08,
+      "loss": 0.0,
+      "step": 1295
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.801170771485567e-08,
+      "loss": 0.0004,
+      "step": 1296
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.504383027333029e-08,
+      "loss": 0.0003,
+      "step": 1297
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.213329537608893e-08,
+      "loss": 0.0002,
+      "step": 1298
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.928011983907246e-08,
+      "loss": 0.0001,
+      "step": 1299
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.648432014682127e-08,
+      "loss": 0.0002,
+      "step": 1300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.374591245238204e-08,
+      "loss": 0.0,
+      "step": 1301
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.106491257721114e-08,
+      "loss": 0.0001,
+      "step": 1302
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.844133601108359e-08,
+      "loss": 0.0003,
+      "step": 1303
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.587519791200868e-08,
+      "loss": 0.0001,
+      "step": 1304
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.3366513106134496e-08,
+      "loss": 0.0002,
+      "step": 1305
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.0915296087670204e-08,
+      "loss": 0.0,
+      "step": 1306
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8521561018793906e-08,
+      "loss": 0.0001,
+      "step": 1307
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.618532172957935e-08,
+      "loss": 0.0001,
+      "step": 1308
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.390659171790934e-08,
+      "loss": 0.0001,
+      "step": 1309
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.1685384149402486e-08,
+      "loss": 0.0003,
+      "step": 1310
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.952171185733211e-08,
+      "loss": 0.0,
+      "step": 1311
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7415587342557455e-08,
+      "loss": 0.0001,
+      "step": 1312
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.536702277345039e-08,
+      "loss": 0.0005,
+      "step": 1313
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.3376029985819904e-08,
+      "loss": 0.0,
+      "step": 1314
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.144262048284885e-08,
+      "loss": 0.0002,
+      "step": 1315
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.9566805435029544e-08,
+      "loss": 0.0001,
+      "step": 1316
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7748595680091583e-08,
+      "loss": 0.0001,
+      "step": 1317
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.5988001722944134e-08,
+      "loss": 0.0005,
+      "step": 1318
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.428503373561708e-08,
+      "loss": 0.0001,
+      "step": 1319
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2639701557198856e-08,
+      "loss": 0.0,
+      "step": 1320
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.105201469377871e-08,
+      "loss": 0.0001,
+      "step": 1321
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9521982318395637e-08,
+      "loss": 0.0001,
+      "step": 1322
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.804961327098398e-08,
+      "loss": 0.0,
+      "step": 1323
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.663491605831902e-08,
+      "loss": 0.0,
+      "step": 1324
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5277898853972573e-08,
+      "loss": 0.0,
+      "step": 1325
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3978569498263039e-08,
+      "loss": 0.0001,
+      "step": 1326
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2736935498212088e-08,
+      "loss": 0.0,
+      "step": 1327
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1553004027498038e-08,
+      "loss": 0.0003,
+      "step": 1328
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0426781926416995e-08,
+      "loss": 0.0001,
+      "step": 1329
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.35827570184511e-09,
+      "loss": 0.0001,
+      "step": 1330
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.347491527195273e-09,
+      "loss": 0.0,
+      "step": 1331
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.394435242384923e-09,
+      "loss": 0.0001,
+      "step": 1332
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.499112353804959e-09,
+      "loss": 0.0009,
+      "step": 1333
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.661528034284214e-09,
+      "loss": 0.0001,
+      "step": 1334
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.881687123057255e-09,
+      "loss": 0.0009,
+      "step": 1335
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.159594125744404e-09,
+      "loss": 0.0,
+      "step": 1336
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.4952532143173178e-09,
+      "loss": 0.0002,
+      "step": 1337
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.888668227081226e-09,
+      "loss": 0.0007,
+      "step": 1338
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.3398426686471744e-09,
+      "loss": 0.0001,
+      "step": 1339
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.848779709917592e-09,
+      "loss": 0.0001,
+      "step": 1340
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.4154821880618674e-09,
+      "loss": 0.0013,
+      "step": 1341
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0399526065074662e-09,
+      "loss": 0.0002,
+      "step": 1342
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.221931349166156e-10,
+      "loss": 0.0002,
+      "step": 1343
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.622056091807547e-10,
+      "loss": 0.0,
+      "step": 1344
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.599915314061008e-10,
+      "loss": 0.0,
+      "step": 1345
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.155520699092083e-10,
+      "loss": 0.0,
+      "step": 1346
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8888059203646678e-11,
+      "loss": 0.0002,
+      "step": 1347
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0,
+      "loss": 0.007,
+      "step": 1348
+    },
+    {
+      "epoch": 4.0,
+      "step": 1348,
+      "total_flos": 1.148849645628162e+18,
+      "train_loss": 0.034755720196650006,
+      "train_runtime": 9562.0276,
+      "train_samples_per_second": 4.511,
+      "train_steps_per_second": 0.141
+    }
+  ],
+  "max_steps": 1348,
+  "num_train_epochs": 4,
+  "total_flos": 1.148849645628162e+18,
+  "trial_name": null,
+  "trial_params": null
+}