diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4849 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.994413407821229,
+  "global_step": 804,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 1.0973,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.048,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.9709,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.9895,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.9636,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.9684,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.9015,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.9344,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.9279,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.9241,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.877,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.9253,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.8935,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.9063,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.9067,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.865,
+      "step": 16
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 0.8991,
+      "step": 17
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.8919,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 0.9375,
+      "step": 19
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.8823,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.9011,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8969,
+      "step": 22
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 0.9377,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.865,
+      "step": 24
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.8763,
+      "step": 25
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 0.8526,
+      "step": 26
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.8437,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 0.8529,
+      "step": 28
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 0.8917,
+      "step": 29
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.8718,
+      "step": 30
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 0.8952,
+      "step": 31
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 0.8822,
+      "step": 32
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.8738,
+      "step": 33
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999916984269715e-05,
+      "loss": 0.8835,
+      "step": 34
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999667938457175e-05,
+      "loss": 0.8765,
+      "step": 35
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999252866697328e-05,
+      "loss": 0.9078,
+      "step": 36
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998671775881668e-05,
+      "loss": 0.8881,
+      "step": 37
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997924675658134e-05,
+      "loss": 0.8477,
+      "step": 38
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997011578430938e-05,
+      "loss": 0.8506,
+      "step": 39
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999593249936037e-05,
+      "loss": 0.8646,
+      "step": 40
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994687456362532e-05,
+      "loss": 0.8204,
+      "step": 41
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993276470109054e-05,
+      "loss": 0.8339,
+      "step": 42
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991699564026755e-05,
+      "loss": 0.8299,
+      "step": 43
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989956764297232e-05,
+      "loss": 0.8368,
+      "step": 44
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9988048099856443e-05,
+      "loss": 0.8283,
+      "step": 45
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998597360239422e-05,
+      "loss": 0.8118,
+      "step": 46
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9983733306353755e-05,
+      "loss": 0.853,
+      "step": 47
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9981327248931008e-05,
+      "loss": 0.8937,
+      "step": 48
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9978755470074093e-05,
+      "loss": 0.8151,
+      "step": 49
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9976018012482642e-05,
+      "loss": 0.86,
+      "step": 50
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9973114921607055e-05,
+      "loss": 0.8584,
+      "step": 51
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9970046245647777e-05,
+      "loss": 0.8532,
+      "step": 52
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.996681203555449e-05,
+      "loss": 0.8282,
+      "step": 53
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.996341234502524e-05,
+      "loss": 0.8483,
+      "step": 54
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99598472305056e-05,
+      "loss": 0.8513,
+      "step": 55
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.995611675118768e-05,
+      "loss": 0.8674,
+      "step": 56
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9952220969009175e-05,
+      "loss": 0.8434,
+      "step": 57
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9948159948652324e-05,
+      "loss": 0.8474,
+      "step": 58
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9943933757542847e-05,
+      "loss": 0.8793,
+      "step": 59
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9939542465848808e-05,
+      "loss": 0.8488,
+      "step": 60
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934986146479457e-05,
+      "loss": 0.8291,
+      "step": 61
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9930264875084045e-05,
+      "loss": 0.857,
+      "step": 62
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9925378730050518e-05,
+      "loss": 0.8573,
+      "step": 63
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920327792504263e-05,
+      "loss": 0.8404,
+      "step": 64
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991511214630673e-05,
+      "loss": 0.8245,
+      "step": 65
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9909731878054057e-05,
+      "loss": 0.8506,
+      "step": 66
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990418707707562e-05,
+      "loss": 0.8564,
+      "step": 67
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9898477835432566e-05,
+      "loss": 0.8516,
+      "step": 68
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.989260424791626e-05,
+      "loss": 0.8626,
+      "step": 69
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988656641204674e-05,
+      "loss": 0.8331,
+      "step": 70
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9880364428071073e-05,
+      "loss": 0.8271,
+      "step": 71
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9873998398961708e-05,
+      "loss": 0.8371,
+      "step": 72
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9867468430414752e-05,
+      "loss": 0.8212,
+      "step": 73
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9860774630848228e-05,
+      "loss": 0.8712,
+      "step": 74
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.985391711140027e-05,
+      "loss": 0.8365,
+      "step": 75
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.984689598592727e-05,
+      "loss": 0.8543,
+      "step": 76
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9839711371002006e-05,
+      "loss": 0.8369,
+      "step": 77
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9832363385911682e-05,
+      "loss": 0.8333,
+      "step": 78
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.982485215265598e-05,
+      "loss": 0.8556,
+      "step": 79
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.981717779594499e-05,
+      "loss": 0.8765,
+      "step": 80
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.980934044319718e-05,
+      "loss": 0.8738,
+      "step": 81
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9801340224537266e-05,
+      "loss": 0.8279,
+      "step": 82
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9793177272794044e-05,
+      "loss": 0.8081,
+      "step": 83
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97848517234982e-05,
+      "loss": 0.818,
+      "step": 84
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9776363714880037e-05,
+      "loss": 0.7964,
+      "step": 85
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9767713387867204e-05,
+      "loss": 0.8212,
+      "step": 86
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9758900886082343e-05,
+      "loss": 0.8306,
+      "step": 87
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9749926355840716e-05,
+      "loss": 0.8317,
+      "step": 88
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9740789946147748e-05,
+      "loss": 0.8105,
+      "step": 89
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9731491808696593e-05,
+      "loss": 0.8434,
+      "step": 90
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.972203209786558e-05,
+      "loss": 0.8484,
+      "step": 91
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9712410970715674e-05,
+      "loss": 0.8463,
+      "step": 92
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9702628586987846e-05,
+      "loss": 0.8552,
+      "step": 93
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.969268510910045e-05,
+      "loss": 0.8104,
+      "step": 94
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9682580702146497e-05,
+      "loss": 0.865,
+      "step": 95
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9672315533890934e-05,
+      "loss": 0.8318,
+      "step": 96
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.966188977476785e-05,
+      "loss": 0.8648,
+      "step": 97
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.965130359787764e-05,
+      "loss": 0.8442,
+      "step": 98
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9640557178984152e-05,
+      "loss": 0.8372,
+      "step": 99
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.962965069651175e-05,
+      "loss": 0.8479,
+      "step": 100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9618584331542345e-05,
+      "loss": 0.8283,
+      "step": 101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.960735826781242e-05,
+      "loss": 0.8316,
+      "step": 102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.959597269170995e-05,
+      "loss": 0.8203,
+      "step": 103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9584427792271317e-05,
+      "loss": 0.8137,
+      "step": 104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9572723761178168e-05,
+      "loss": 0.8307,
+      "step": 105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9560860792754248e-05,
+      "loss": 0.7982,
+      "step": 106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.954883908396215e-05,
+      "loss": 0.8406,
+      "step": 107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9536658834400058e-05,
+      "loss": 0.8073,
+      "step": 108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.952432024629844e-05,
+      "loss": 0.8122,
+      "step": 109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9511823524516677e-05,
+      "loss": 0.826,
+      "step": 110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9499168876539666e-05,
+      "loss": 0.7872,
+      "step": 111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.948635651247437e-05,
+      "loss": 0.8111,
+      "step": 112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9473386645046344e-05,
+      "loss": 0.8503,
+      "step": 113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9460259489596192e-05,
+      "loss": 0.8338,
+      "step": 114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9446975264075993e-05,
+      "loss": 0.8166,
+      "step": 115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.943353418904568e-05,
+      "loss": 0.8383,
+      "step": 116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9419936487669396e-05,
+      "loss": 0.8706,
+      "step": 117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.940618238571175e-05,
+      "loss": 0.843,
+      "step": 118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.939227211153411e-05,
+      "loss": 0.8368,
+      "step": 119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9378205896090796e-05,
+      "loss": 0.8077,
+      "step": 120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.936398397292523e-05,
+      "loss": 0.7824,
+      "step": 121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.934960657816608e-05,
+      "loss": 0.8343,
+      "step": 122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9335073950523335e-05,
+      "loss": 0.7876,
+      "step": 123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9320386331284334e-05,
+      "loss": 0.7896,
+      "step": 124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9305543964309756e-05,
+      "loss": 0.8056,
+      "step": 125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.92905470960296e-05,
+      "loss": 0.8155,
+      "step": 126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9275395975439054e-05,
+      "loss": 0.8526,
+      "step": 127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9260090854094388e-05,
+      "loss": 0.8044,
+      "step": 128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9244631986108768e-05,
+      "loss": 0.7958,
+      "step": 129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.922901962814803e-05,
+      "loss": 0.8284,
+      "step": 130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.921325403942644e-05,
+      "loss": 0.8291,
+      "step": 131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9197335481702374e-05,
+      "loss": 0.8035,
+      "step": 132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9181264219273957e-05,
+      "loss": 0.7947,
+      "step": 133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.916504051897472e-05,
+      "loss": 0.8085,
+      "step": 134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9148664650169128e-05,
+      "loss": 0.8082,
+      "step": 135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9132136884748114e-05,
+      "loss": 0.801,
+      "step": 136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9115457497124588e-05,
+      "loss": 0.781,
+      "step": 137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9098626764228854e-05,
+      "loss": 0.7741,
+      "step": 138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9081644965504035e-05,
+      "loss": 0.8164,
+      "step": 139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.906451238290141e-05,
+      "loss": 0.8238,
+      "step": 140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.904722930087575e-05,
+      "loss": 0.8289,
+      "step": 141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9029796006380595e-05,
+      "loss": 0.8435,
+      "step": 142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9012212788863475e-05,
+      "loss": 0.837,
+      "step": 143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8994479940261122e-05,
+      "loss": 0.8466,
+      "step": 144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8976597754994614e-05,
+      "loss": 0.8067,
+      "step": 145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8958566529964475e-05,
+      "loss": 0.845,
+      "step": 146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8940386564545773e-05,
+      "loss": 0.8186,
+      "step": 147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8922058160583133e-05,
+      "loss": 0.8188,
+      "step": 148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.890358162238572e-05,
+      "loss": 0.8225,
+      "step": 149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8884957256722197e-05,
+      "loss": 0.8152,
+      "step": 150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.886618537281562e-05,
+      "loss": 0.8285,
+      "step": 151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8847266282338333e-05,
+      "loss": 0.7959,
+      "step": 152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8828200299406747e-05,
+      "loss": 0.7822,
+      "step": 153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8808987740576166e-05,
+      "loss": 0.8137,
+      "step": 154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8789628924835512e-05,
+      "loss": 0.825,
+      "step": 155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8770124173602026e-05,
+      "loss": 0.823,
+      "step": 156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8750473810715944e-05,
+      "loss": 0.8252,
+      "step": 157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8730678162435112e-05,
+      "loss": 0.82,
+      "step": 158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.871073755742957e-05,
+      "loss": 0.7967,
+      "step": 159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8690652326776088e-05,
+      "loss": 0.8253,
+      "step": 160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8670422803952696e-05,
+      "loss": 0.7896,
+      "step": 161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8650049324833107e-05,
+      "loss": 0.7918,
+      "step": 162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8629532227681175e-05,
+      "loss": 0.8054,
+      "step": 163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8608871853145257e-05,
+      "loss": 0.788,
+      "step": 164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8588068544252572e-05,
+      "loss": 0.7879,
+      "step": 165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.85671226464035e-05,
+      "loss": 0.7775,
+      "step": 166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8546034507365832e-05,
+      "loss": 0.7861,
+      "step": 167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.852480447726903e-05,
+      "loss": 0.8016,
+      "step": 168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.850343290859838e-05,
+      "loss": 0.817,
+      "step": 169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8481920156189157e-05,
+      "loss": 0.7894,
+      "step": 170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8460266577220733e-05,
+      "loss": 0.8207,
+      "step": 171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8438472531210645e-05,
+      "loss": 0.7807,
+      "step": 172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8416538380008616e-05,
+      "loss": 0.8199,
+      "step": 173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.839446448779057e-05,
+      "loss": 0.8094,
+      "step": 174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8372251221052556e-05,
+      "loss": 0.8002,
+      "step": 175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8349898948604686e-05,
+      "loss": 0.8008,
+      "step": 176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8327408041565013e-05,
+      "loss": 0.8159,
+      "step": 177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8304778873353345e-05,
+      "loss": 0.7965,
+      "step": 178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8282011819685064e-05,
+      "loss": 0.7865,
+      "step": 179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8259107258564897e-05,
+      "loss": 0.7781,
+      "step": 180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.823606557028061e-05,
+      "loss": 0.7835,
+      "step": 181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.821288713739672e-05,
+      "loss": 0.8132,
+      "step": 182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.818957234474813e-05,
+      "loss": 0.8205,
+      "step": 183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.816612157943376e-05,
+      "loss": 0.7924,
+      "step": 184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8142535230810078e-05,
+      "loss": 0.836,
+      "step": 185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8118813690484686e-05,
+      "loss": 0.8127,
+      "step": 186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8094957352309773e-05,
+      "loss": 0.7757,
+      "step": 187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8070966612375612e-05,
+      "loss": 0.8037,
+      "step": 188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8046841869003962e-05,
+      "loss": 0.7764,
+      "step": 189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8022583522741456e-05,
+      "loss": 0.8267,
+      "step": 190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.799819197635297e-05,
+      "loss": 0.8251,
+      "step": 191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.79736676348149e-05,
+      "loss": 0.8104,
+      "step": 192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.794901090530848e-05,
+      "loss": 0.7853,
+      "step": 193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7924222197212984e-05,
+      "loss": 0.7996,
+      "step": 194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7899301922098958e-05,
+      "loss": 0.8062,
+      "step": 195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7874250493721362e-05,
+      "loss": 0.8023,
+      "step": 196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7849068328012734e-05,
+      "loss": 0.8263,
+      "step": 197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7823755843076233e-05,
+      "loss": 0.7949,
+      "step": 198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7798313459178754e-05,
+      "loss": 0.7921,
+      "step": 199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7772741598743913e-05,
+      "loss": 0.7891,
+      "step": 200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.774704068634504e-05,
+      "loss": 0.7698,
+      "step": 201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.772121114869814e-05,
+      "loss": 0.7697,
+      "step": 202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7695253414654796e-05,
+      "loss": 0.7906,
+      "step": 203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.766916791519506e-05,
+      "loss": 0.7894,
+      "step": 204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7642955083420288e-05,
+      "loss": 0.8071,
+      "step": 205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7616615354545956e-05,
+      "loss": 0.8193,
+      "step": 206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.759014916589443e-05,
+      "loss": 0.7644,
+      "step": 207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7563556956887705e-05,
+      "loss": 0.7712,
+      "step": 208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7536839169040107e-05,
+      "loss": 0.7797,
+      "step": 209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7509996245950977e-05,
+      "loss": 0.799,
+      "step": 210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7483028633297287e-05,
+      "loss": 0.8082,
+      "step": 211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.745593677882624e-05,
+      "loss": 0.7822,
+      "step": 212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7428721132347863e-05,
+      "loss": 0.7855,
+      "step": 213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.740138214572751e-05,
+      "loss": 0.8207,
+      "step": 214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7373920272878357e-05,
+      "loss": 0.789,
+      "step": 215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.73463359697539e-05,
+      "loss": 0.8216,
+      "step": 216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.731862969434035e-05,
+      "loss": 0.7807,
+      "step": 217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.729080190664904e-05,
+      "loss": 0.7883,
+      "step": 218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7262853068708807e-05,
+      "loss": 0.7688,
+      "step": 219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7234783644558276e-05,
+      "loss": 0.7807,
+      "step": 220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72065941002382e-05,
+      "loss": 0.7771,
+      "step": 221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7178284903783698e-05,
+      "loss": 0.7987,
+      "step": 222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.71498565252165e-05,
+      "loss": 0.7784,
+      "step": 223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7121309436537125e-05,
+      "loss": 0.7796,
+      "step": 224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7092644111717052e-05,
+      "loss": 0.7717,
+      "step": 225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7063861026690863e-05,
+      "loss": 0.7564,
+      "step": 226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7034960659348314e-05,
+      "loss": 0.8182,
+      "step": 227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7005943489526432e-05,
+      "loss": 0.7806,
+      "step": 228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.697680999900152e-05,
+      "loss": 0.7406,
+      "step": 229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6947560671481182e-05,
+      "loss": 0.8077,
+      "step": 230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6918195992596274e-05,
+      "loss": 0.7813,
+      "step": 231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6888716449892843e-05,
+      "loss": 0.7902,
+      "step": 232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.685912253282405e-05,
+      "loss": 0.8099,
+      "step": 233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6829414732742013e-05,
+      "loss": 0.7749,
+      "step": 234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6799593542889686e-05,
+      "loss": 0.7565,
+      "step": 235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6769659458392636e-05,
+      "loss": 0.7424,
+      "step": 236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6739612976250836e-05,
+      "loss": 0.769,
+      "step": 237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.670945459533042e-05,
+      "loss": 0.8043,
+      "step": 238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6679184816355393e-05,
+      "loss": 0.7834,
+      "step": 239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6648804141899316e-05,
+      "loss": 0.8174,
+      "step": 240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6618313076376966e-05,
+      "loss": 0.7654,
+      "step": 241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.658771212603595e-05,
+      "loss": 0.7987,
+      "step": 242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6557001798948324e-05,
+      "loss": 0.7606,
+      "step": 243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.652618260500213e-05,
+      "loss": 0.7392,
+      "step": 244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6495255055892948e-05,
+      "loss": 0.7951,
+      "step": 245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6464219665115392e-05,
+      "loss": 0.7711,
+      "step": 246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6433076947954593e-05,
+      "loss": 0.7499,
+      "step": 247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6401827421477625e-05,
+      "loss": 0.7558,
+      "step": 248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.637047160452494e-05,
+      "loss": 0.7708,
+      "step": 249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6339010017701757e-05,
+      "loss": 0.8055,
+      "step": 250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6307443183369382e-05,
+      "loss": 0.756,
+      "step": 251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6275771625636592e-05,
+      "loss": 0.7945,
+      "step": 252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6243995870350875e-05,
+      "loss": 0.7834,
+      "step": 253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621211644508974e-05,
+      "loss": 0.716,
+      "step": 254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6180133879151943e-05,
+      "loss": 0.7796,
+      "step": 255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6148048703548696e-05,
+      "loss": 0.8073,
+      "step": 256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6115861450994855e-05,
+      "loss": 0.767,
+      "step": 257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6083572655900074e-05,
+      "loss": 0.7638,
+      "step": 258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6051182854359935e-05,
+      "loss": 0.8298,
+      "step": 259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.601869258414705e-05,
+      "loss": 0.7502,
+      "step": 260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5986102384702112e-05,
+      "loss": 0.775,
+      "step": 261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5953412797124968e-05,
+      "loss": 0.7736,
+      "step": 262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.592062436416561e-05,
+      "loss": 0.765,
+      "step": 263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5887737630215188e-05,
+      "loss": 0.7844,
+      "step": 264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5854753141296938e-05,
+      "loss": 0.7701,
+      "step": 265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5821671445057157e-05,
+      "loss": 0.7761,
+      "step": 266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5788493090756074e-05,
+      "loss": 0.7867,
+      "step": 267
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5755218629258747e-05,
+      "loss": 0.7642,
+      "step": 268
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.572184861302593e-05,
+      "loss": 0.7152,
+      "step": 269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5688383596104875e-05,
+      "loss": 0.7047,
+      "step": 270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.565482413412014e-05,
+      "loss": 0.7114,
+      "step": 271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.562117078426438e-05,
+      "loss": 0.7058,
+      "step": 272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.558742410528907e-05,
+      "loss": 0.7027,
+      "step": 273
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5553584657495257e-05,
+      "loss": 0.7172,
+      "step": 274
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5519653002724235e-05,
+      "loss": 0.7043,
+      "step": 275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5485629704348223e-05,
+      "loss": 0.7049,
+      "step": 276
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5451515327261012e-05,
+      "loss": 0.7158,
+      "step": 277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5417310437868586e-05,
+      "loss": 0.7166,
+      "step": 278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5383015604079723e-05,
+      "loss": 0.6919,
+      "step": 279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5348631395296553e-05,
+      "loss": 0.7179,
+      "step": 280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.531415838240512e-05,
+      "loss": 0.6826,
+      "step": 281
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5279597137765894e-05,
+      "loss": 0.726,
+      "step": 282
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5244948235204265e-05,
+      "loss": 0.7188,
+      "step": 283
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5210212250001024e-05,
+      "loss": 0.7284,
+      "step": 284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5175389758882803e-05,
+      "loss": 0.714,
+      "step": 285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5140481340012515e-05,
+      "loss": 0.7043,
+      "step": 286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5105487572979729e-05,
+      "loss": 0.729,
+      "step": 287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5070409038791077e-05,
+      "loss": 0.743,
+      "step": 288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5035246319860578e-05,
+      "loss": 0.7074,
+      "step": 289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7443,
+      "step": 290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4964670664409136e-05,
+      "loss": 0.7464,
+      "step": 291
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4929258899666102e-05,
+      "loss": 0.7055,
+      "step": 292
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.48937652937176e-05,
+      "loss": 0.7015,
+      "step": 293
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4858190435869155e-05,
+      "loss": 0.6848,
+      "step": 294
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.482253491677533e-05,
+      "loss": 0.688,
+      "step": 295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4786799328429915e-05,
+      "loss": 0.6706,
+      "step": 296
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4750984264156103e-05,
+      "loss": 0.7108,
+      "step": 297
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4715090318596637e-05,
+      "loss": 0.6752,
+      "step": 298
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4679118087703933e-05,
+      "loss": 0.7324,
+      "step": 299
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4643068168730203e-05,
+      "loss": 0.7175,
+      "step": 300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4606941160217511e-05,
+      "loss": 0.7064,
+      "step": 301
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4570737661987862e-05,
+      "loss": 0.7336,
+      "step": 302
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4534458275133214e-05,
+      "loss": 0.7196,
+      "step": 303
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4498103602005537e-05,
+      "loss": 0.7266,
+      "step": 304
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.446167424620677e-05,
+      "loss": 0.7367,
+      "step": 305
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4425170812578837e-05,
+      "loss": 0.7055,
+      "step": 306
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4388593907193571e-05,
+      "loss": 0.7063,
+      "step": 307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.435194413734268e-05,
+      "loss": 0.6754,
+      "step": 308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.431522211152764e-05,
+      "loss": 0.7158,
+      "step": 309
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4278428439449613e-05,
+      "loss": 0.6712,
+      "step": 310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.424156373199931e-05,
+      "loss": 0.6762,
+      "step": 311
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.420462860124685e-05,
+      "loss": 0.682,
+      "step": 312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.416762366043161e-05,
+      "loss": 0.6596,
+      "step": 313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4130549523952022e-05,
+      "loss": 0.6997,
+      "step": 314
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4093406807355389e-05,
+      "loss": 0.6553,
+      "step": 315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4056196127327661e-05,
+      "loss": 0.7142,
+      "step": 316
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4018918101683191e-05,
+      "loss": 0.6933,
+      "step": 317
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3981573349354492e-05,
+      "loss": 0.7041,
+      "step": 318
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3944162490381931e-05,
+      "loss": 0.7147,
+      "step": 319
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3906686145903467e-05,
+      "loss": 0.6945,
+      "step": 320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3869144938144325e-05,
+      "loss": 0.7096,
+      "step": 321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3831539490406658e-05,
+      "loss": 0.6956,
+      "step": 322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3793870427059204e-05,
+      "loss": 0.6735,
+      "step": 323
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3756138373526931e-05,
+      "loss": 0.7222,
+      "step": 324
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3718343956280624e-05,
+      "loss": 0.6984,
+      "step": 325
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3680487802826523e-05,
+      "loss": 0.6959,
+      "step": 326
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3642570541695867e-05,
+      "loss": 0.708,
+      "step": 327
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3604592802434486e-05,
+      "loss": 0.7124,
+      "step": 328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3566555215592323e-05,
+      "loss": 0.7126,
+      "step": 329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3528458412712996e-05,
+      "loss": 0.6948,
+      "step": 330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3490303026323289e-05,
+      "loss": 0.6973,
+      "step": 331
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3452089689922646e-05,
+      "loss": 0.6928,
+      "step": 332
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3413819037972682e-05,
+      "loss": 0.7048,
+      "step": 333
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3375491705886614e-05,
+      "loss": 0.6836,
+      "step": 334
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3337108330018737e-05,
+      "loss": 0.7153,
+      "step": 335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3298669547653848e-05,
+      "loss": 0.7103,
+      "step": 336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3260175996996667e-05,
+      "loss": 0.7268,
+      "step": 337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.322162831716123e-05,
+      "loss": 0.7019,
+      "step": 338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3183027148160304e-05,
+      "loss": 0.6769,
+      "step": 339
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3144373130894729e-05,
+      "loss": 0.6861,
+      "step": 340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3105666907142804e-05,
+      "loss": 0.6948,
+      "step": 341
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3066909119549615e-05,
+      "loss": 0.6713,
+      "step": 342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3028100411616369e-05,
+      "loss": 0.7094,
+      "step": 343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2989241427689712e-05,
+      "loss": 0.6863,
+      "step": 344
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.295033281295103e-05,
+      "loss": 0.7025,
+      "step": 345
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2911375213405738e-05,
+      "loss": 0.6959,
+      "step": 346
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2872369275872547e-05,
+      "loss": 0.7007,
+      "step": 347
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.283331564797274e-05,
+      "loss": 0.7085,
+      "step": 348
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2794214978119399e-05,
+      "loss": 0.7446,
+      "step": 349
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2755067915506664e-05,
+      "loss": 0.7093,
+      "step": 350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.271587511009893e-05,
+      "loss": 0.6597,
+      "step": 351
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2676637212620069e-05,
+      "loss": 0.6855,
+      "step": 352
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.263735487454262e-05,
+      "loss": 0.6607,
+      "step": 353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2598028748076987e-05,
+      "loss": 0.6527,
+      "step": 354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.255865948616059e-05,
+      "loss": 0.6636,
+      "step": 355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2519247742447029e-05,
+      "loss": 0.717,
+      "step": 356
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2479794171295248e-05,
+      "loss": 0.6842,
+      "step": 357
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2440299427758646e-05,
+      "loss": 0.6855,
+      "step": 358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2400764167574225e-05,
+      "loss": 0.728,
+      "step": 359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2361189047151676e-05,
+      "loss": 0.6984,
+      "step": 360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2321574723562511e-05,
+      "loss": 0.7033,
+      "step": 361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2281921854529127e-05,
+      "loss": 0.6982,
+      "step": 362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.22422310984139e-05,
+      "loss": 0.6974,
+      "step": 363
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2202503114208251e-05,
+      "loss": 0.7017,
+      "step": 364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2162738561521704e-05,
+      "loss": 0.722,
+      "step": 365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2122938100570938e-05,
+      "loss": 0.7111,
+      "step": 366
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2083102392168814e-05,
+      "loss": 0.6866,
+      "step": 367
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2043232097713424e-05,
+      "loss": 0.7032,
+      "step": 368
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2003327879177085e-05,
+      "loss": 0.6944,
+      "step": 369
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1963390399095368e-05,
+      "loss": 0.7073,
+      "step": 370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1923420320556085e-05,
+      "loss": 0.6594,
+      "step": 371
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1883418307188292e-05,
+      "loss": 0.698,
+      "step": 372
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1843385023151255e-05,
+      "loss": 0.6736,
+      "step": 373
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1803321133123442e-05,
+      "loss": 0.6595,
+      "step": 374
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1763227302291464e-05,
+      "loss": 0.6944,
+      "step": 375
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1723104196339055e-05,
+      "loss": 0.7015,
+      "step": 376
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1682952481436e-05,
+      "loss": 0.6616,
+      "step": 377
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1642772824227091e-05,
+      "loss": 0.6924,
+      "step": 378
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.160256589182104e-05,
+      "loss": 0.6467,
+      "step": 379
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1562332351779422e-05,
+      "loss": 0.6802,
+      "step": 380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1522072872105576e-05,
+      "loss": 0.6882,
+      "step": 381
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1481788121233523e-05,
+      "loss": 0.7198,
+      "step": 382
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1441478768016871e-05,
+      "loss": 0.7014,
+      "step": 383
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1401145481717695e-05,
+      "loss": 0.7074,
+      "step": 384
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1360788931995434e-05,
+      "loss": 0.6883,
+      "step": 385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1320409788895786e-05,
+      "loss": 0.7337,
+      "step": 386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1280008722839552e-05,
+      "loss": 0.6999,
+      "step": 387
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1239586404611542e-05,
+      "loss": 0.688,
+      "step": 388
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1199143505349401e-05,
+      "loss": 0.6497,
+      "step": 389
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1158680696532502e-05,
+      "loss": 0.667,
+      "step": 390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1118198649970762e-05,
+      "loss": 0.6805,
+      "step": 391
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1077698037793516e-05,
+      "loss": 0.6681,
+      "step": 392
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1037179532438345e-05,
+      "loss": 0.6398,
+      "step": 393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0996643806639904e-05,
+      "loss": 0.6877,
+      "step": 394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0956091533418777e-05,
+      "loss": 0.6758,
+      "step": 395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0915523386070277e-05,
+      "loss": 0.6901,
+      "step": 396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0874940038153279e-05,
+      "loss": 0.6693,
+      "step": 397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0834342163479035e-05,
+      "loss": 0.6775,
+      "step": 398
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.079373043609999e-05,
+      "loss": 0.6905,
+      "step": 399
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0753105530298582e-05,
+      "loss": 0.6888,
+      "step": 400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0712468120576059e-05,
+      "loss": 0.6849,
+      "step": 401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.067181888164127e-05,
+      "loss": 0.66,
+      "step": 402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0631158488399463e-05,
+      "loss": 0.6826,
+      "step": 403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0590487615941088e-05,
+      "loss": 0.673,
+      "step": 404
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.054980693953058e-05,
+      "loss": 0.6772,
+      "step": 405
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0509117134595152e-05,
+      "loss": 0.6454,
+      "step": 406
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.046841887671358e-05,
+      "loss": 0.6422,
+      "step": 407
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.042771284160498e-05,
+      "loss": 0.7042,
+      "step": 408
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.03869997051176e-05,
+      "loss": 0.6828,
+      "step": 409
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0346280143217593e-05,
+      "loss": 0.69,
+      "step": 410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0305554831977788e-05,
+      "loss": 0.7156,
+      "step": 411
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0264824447566484e-05,
+      "loss": 0.6925,
+      "step": 412
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0224089666236195e-05,
+      "loss": 0.7213,
+      "step": 413
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0183351164312448e-05,
+      "loss": 0.6712,
+      "step": 414
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0142609618182539e-05,
+      "loss": 0.7007,
+      "step": 415
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.010186570428431e-05,
+      "loss": 0.6874,
+      "step": 416
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0061120099094917e-05,
+      "loss": 0.6735,
+      "step": 417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.002037347911959e-05,
+      "loss": 0.6991,
+      "step": 418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.979626520880413e-06,
+      "loss": 0.6606,
+      "step": 419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.938879900905086e-06,
+      "loss": 0.6891,
+      "step": 420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.898134295715693e-06,
+      "loss": 0.6594,
+      "step": 421
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.857390381817463e-06,
+      "loss": 0.6597,
+      "step": 422
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.816648835687557e-06,
+      "loss": 0.6806,
+      "step": 423
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.775910333763809e-06,
+      "loss": 0.6837,
+      "step": 424
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.73517555243352e-06,
+      "loss": 0.6813,
+      "step": 425
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.694445168022214e-06,
+      "loss": 0.7019,
+      "step": 426
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.653719856782412e-06,
+      "loss": 0.6624,
+      "step": 427
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.613000294882406e-06,
+      "loss": 0.6842,
+      "step": 428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.572287158395025e-06,
+      "loss": 0.6762,
+      "step": 429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.531581123286426e-06,
+      "loss": 0.6414,
+      "step": 430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.49088286540485e-06,
+      "loss": 0.6768,
+      "step": 431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.45019306046942e-06,
+      "loss": 0.6537,
+      "step": 432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.409512384058916e-06,
+      "loss": 0.6547,
+      "step": 433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.368841511600539e-06,
+      "loss": 0.6429,
+      "step": 434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.328181118358734e-06,
+      "loss": 0.6631,
+      "step": 435
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.287531879423943e-06,
+      "loss": 0.6514,
+      "step": 436
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.246894469701423e-06,
+      "loss": 0.6907,
+      "step": 437
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.206269563900013e-06,
+      "loss": 0.6611,
+      "step": 438
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.165657836520969e-06,
+      "loss": 0.6811,
+      "step": 439
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.125059961846725e-06,
+      "loss": 0.6474,
+      "step": 440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.084476613929726e-06,
+      "loss": 0.6717,
+      "step": 441
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.043908466581225e-06,
+      "loss": 0.6683,
+      "step": 442
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.003356193360097e-06,
+      "loss": 0.6768,
+      "step": 443
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.96282046756166e-06,
+      "loss": 0.6776,
+      "step": 444
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.922301962206487e-06,
+      "loss": 0.6714,
+      "step": 445
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.881801350029242e-06,
+      "loss": 0.641,
+      "step": 446
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.841319303467502e-06,
+      "loss": 0.6795,
+      "step": 447
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.8008564946506e-06,
+      "loss": 0.6439,
+      "step": 448
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.760413595388463e-06,
+      "loss": 0.6314,
+      "step": 449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.719991277160453e-06,
+      "loss": 0.6687,
+      "step": 450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.67959021110422e-06,
+      "loss": 0.6709,
+      "step": 451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.63921106800457e-06,
+      "loss": 0.6532,
+      "step": 452
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.59885451828231e-06,
+      "loss": 0.7103,
+      "step": 453
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.558521231983134e-06,
+      "loss": 0.7086,
+      "step": 454
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.518211878766475e-06,
+      "loss": 0.6436,
+      "step": 455
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.477927127894424e-06,
+      "loss": 0.6654,
+      "step": 456
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.43766764822058e-06,
+      "loss": 0.6417,
+      "step": 457
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.39743410817896e-06,
+      "loss": 0.6835,
+      "step": 458
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.35722717577291e-06,
+      "loss": 0.6904,
+      "step": 459
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.317047518564e-06,
+      "loss": 0.6609,
+      "step": 460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.276895803660947e-06,
+      "loss": 0.6907,
+      "step": 461
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.236772697708537e-06,
+      "loss": 0.6465,
+      "step": 462
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.196678866876561e-06,
+      "loss": 0.6563,
+      "step": 463
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.156614976848748e-06,
+      "loss": 0.6752,
+      "step": 464
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.116581692811711e-06,
+      "loss": 0.6844,
+      "step": 465
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.076579679443919e-06,
+      "loss": 0.7005,
+      "step": 466
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.036609600904636e-06,
+      "loss": 0.6601,
+      "step": 467
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.996672120822918e-06,
+      "loss": 0.6684,
+      "step": 468
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.956767902286578e-06,
+      "loss": 0.6402,
+      "step": 469
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.916897607831189e-06,
+      "loss": 0.6438,
+      "step": 470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.877061899429067e-06,
+      "loss": 0.6256,
+      "step": 471
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.8372614384783e-06,
+      "loss": 0.644,
+      "step": 472
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.797496885791754e-06,
+      "loss": 0.6987,
+      "step": 473
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.757768901586106e-06,
+      "loss": 0.6808,
+      "step": 474
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.718078145470878e-06,
+      "loss": 0.6762,
+      "step": 475
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.678425276437494e-06,
+      "loss": 0.6372,
+      "step": 476
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.638810952848328e-06,
+      "loss": 0.6541,
+      "step": 477
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.599235832425781e-06,
+      "loss": 0.6541,
+      "step": 478
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.559700572241354e-06,
+      "loss": 0.6985,
+      "step": 479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.520205828704753e-06,
+      "loss": 0.6441,
+      "step": 480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.480752257552972e-06,
+      "loss": 0.6647,
+      "step": 481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.441340513839412e-06,
+      "loss": 0.6855,
+      "step": 482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.401971251923015e-06,
+      "loss": 0.6828,
+      "step": 483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.362645125457381e-06,
+      "loss": 0.6739,
+      "step": 484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.323362787379935e-06,
+      "loss": 0.6812,
+      "step": 485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.284124889901074e-06,
+      "loss": 0.6458,
+      "step": 486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.244932084493338e-06,
+      "loss": 0.6302,
+      "step": 487
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.205785021880603e-06,
+      "loss": 0.6504,
+      "step": 488
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.166684352027265e-06,
+      "loss": 0.6717,
+      "step": 489
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.127630724127457e-06,
+      "loss": 0.6382,
+      "step": 490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.088624786594266e-06,
+      "loss": 0.6781,
+      "step": 491
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.049667187048974e-06,
+      "loss": 0.6379,
+      "step": 492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.01075857231029e-06,
+      "loss": 0.6438,
+      "step": 493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.971899588383635e-06,
+      "loss": 0.6584,
+      "step": 494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9330908804503874e-06,
+      "loss": 0.6621,
+      "step": 495
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.894333092857199e-06,
+      "loss": 0.6591,
+      "step": 496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.855626869105273e-06,
+      "loss": 0.6366,
+      "step": 497
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8169728518397e-06,
+      "loss": 0.6509,
+      "step": 498
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.778371682838775e-06,
+      "loss": 0.6809,
+      "step": 499
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.739824003003339e-06,
+      "loss": 0.6636,
+      "step": 500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.701330452346156e-06,
+      "loss": 0.6704,
+      "step": 501
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.662891669981266e-06,
+      "loss": 0.6648,
+      "step": 502
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.624508294113388e-06,
+      "loss": 0.6388,
+      "step": 503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.586180962027322e-06,
+      "loss": 0.6193,
+      "step": 504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.547910310077354e-06,
+      "loss": 0.6301,
+      "step": 505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.509696973676714e-06,
+      "loss": 0.6553,
+      "step": 506
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.471541587287003e-06,
+      "loss": 0.6595,
+      "step": 507
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.433444784407679e-06,
+      "loss": 0.7049,
+      "step": 508
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.395407197565518e-06,
+      "loss": 0.6673,
+      "step": 509
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.357429458304135e-06,
+      "loss": 0.668,
+      "step": 510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.319512197173479e-06,
+      "loss": 0.6426,
+      "step": 511
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2816560437193786e-06,
+      "loss": 0.6404,
+      "step": 512
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.243861626473073e-06,
+      "loss": 0.6102,
+      "step": 513
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.206129572940798e-06,
+      "loss": 0.6502,
+      "step": 514
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.168460509593343e-06,
+      "loss": 0.6352,
+      "step": 515
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1308550618556765e-06,
+      "loss": 0.6234,
+      "step": 516
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.093313854096534e-06,
+      "loss": 0.6304,
+      "step": 517
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0558375096180724e-06,
+      "loss": 0.6817,
+      "step": 518
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0184266506455125e-06,
+      "loss": 0.6517,
+      "step": 519
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.981081898316809e-06,
+      "loss": 0.639,
+      "step": 520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.943803872672344e-06,
+      "loss": 0.681,
+      "step": 521
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906593192644615e-06,
+      "loss": 0.6147,
+      "step": 522
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.869450476047984e-06,
+      "loss": 0.6364,
+      "step": 523
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.832376339568394e-06,
+      "loss": 0.6765,
+      "step": 524
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.795371398753153e-06,
+      "loss": 0.658,
+      "step": 525
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.758436268000696e-06,
+      "loss": 0.6379,
+      "step": 526
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.72157156055039e-06,
+      "loss": 0.6504,
+      "step": 527
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.684777888472359e-06,
+      "loss": 0.6907,
+      "step": 528
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.648055862657321e-06,
+      "loss": 0.6411,
+      "step": 529
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611406092806429e-06,
+      "loss": 0.656,
+      "step": 530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.574829187421166e-06,
+      "loss": 0.641,
+      "step": 531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.538325753793227e-06,
+      "loss": 0.6668,
+      "step": 532
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.501896397994465e-06,
+      "loss": 0.6293,
+      "step": 533
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.465541724866787e-06,
+      "loss": 0.6394,
+      "step": 534
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.429262338012144e-06,
+      "loss": 0.6625,
+      "step": 535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.393058839782492e-06,
+      "loss": 0.6709,
+      "step": 536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.356931831269798e-06,
+      "loss": 0.5811,
+      "step": 537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3208819122960674e-06,
+      "loss": 0.5701,
+      "step": 538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.284909681403367e-06,
+      "loss": 0.5856,
+      "step": 539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.249015735843901e-06,
+      "loss": 0.5834,
+      "step": 540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.213200671570085e-06,
+      "loss": 0.5851,
+      "step": 541
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.177465083224671e-06,
+      "loss": 0.5928,
+      "step": 542
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.141809564130847e-06,
+      "loss": 0.6236,
+      "step": 543
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.106234706282405e-06,
+      "loss": 0.5716,
+      "step": 544
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.070741100333901e-06,
+      "loss": 0.5997,
+      "step": 545
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.035329335590868e-06,
+      "loss": 0.6034,
+      "step": 546
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.5754,
+      "step": 547
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.964753680139425e-06,
+      "loss": 0.5561,
+      "step": 548
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9295909612089265e-06,
+      "loss": 0.6011,
+      "step": 549
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.894512427020275e-06,
+      "loss": 0.5922,
+      "step": 550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.859518659987491e-06,
+      "loss": 0.6,
+      "step": 551
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.824610241117198e-06,
+      "loss": 0.6257,
+      "step": 552
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.789787749998979e-06,
+      "loss": 0.5812,
+      "step": 553
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.755051764795737e-06,
+      "loss": 0.5979,
+      "step": 554
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.720402862234105e-06,
+      "loss": 0.6245,
+      "step": 555
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6858416175948795e-06,
+      "loss": 0.6298,
+      "step": 556
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651368604703449e-06,
+      "loss": 0.5722,
+      "step": 557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.616984395920282e-06,
+      "loss": 0.6015,
+      "step": 558
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.582689562131416e-06,
+      "loss": 0.6211,
+      "step": 559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.548484672738992e-06,
+      "loss": 0.638,
+      "step": 560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.514370295651781e-06,
+      "loss": 0.5722,
+      "step": 561
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.480346997275767e-06,
+      "loss": 0.5818,
+      "step": 562
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.446415342504742e-06,
+      "loss": 0.5717,
+      "step": 563
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.412575894710931e-06,
+      "loss": 0.551,
+      "step": 564
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378829215735624e-06,
+      "loss": 0.5558,
+      "step": 565
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.345175865879863e-06,
+      "loss": 0.5926,
+      "step": 566
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.311616403895126e-06,
+      "loss": 0.5676,
+      "step": 567
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.27815138697407e-06,
+      "loss": 0.6067,
+      "step": 568
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.244781370741254e-06,
+      "loss": 0.5916,
+      "step": 569
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.211506909243932e-06,
+      "loss": 0.595,
+      "step": 570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.17832855494285e-06,
+      "loss": 0.6141,
+      "step": 571
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.145246858703064e-06,
+      "loss": 0.6039,
+      "step": 572
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1122623697848164e-06,
+      "loss": 0.6137,
+      "step": 573
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.079375635834392e-06,
+      "loss": 0.6145,
+      "step": 574
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.046587202875036e-06,
+      "loss": 0.5684,
+      "step": 575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.013897615297889e-06,
+      "loss": 0.5587,
+      "step": 576
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.981307415852949e-06,
+      "loss": 0.5996,
+      "step": 577
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948817145640062e-06,
+      "loss": 0.5766,
+      "step": 578
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.916427344099928e-06,
+      "loss": 0.5676,
+      "step": 579
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.884138549005149e-06,
+      "loss": 0.5692,
+      "step": 580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.851951296451305e-06,
+      "loss": 0.5753,
+      "step": 581
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.819866120848058e-06,
+      "loss": 0.5462,
+      "step": 582
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7878835549102623e-06,
+      "loss": 0.5613,
+      "step": 583
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.756004129649129e-06,
+      "loss": 0.5663,
+      "step": 584
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.72422837436341e-06,
+      "loss": 0.5799,
+      "step": 585
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6925568166306182e-06,
+      "loss": 0.5639,
+      "step": 586
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.660989982298249e-06,
+      "loss": 0.5945,
+      "step": 587
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6295283954750626e-06,
+      "loss": 0.5932,
+      "step": 588
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.598172578522382e-06,
+      "loss": 0.5985,
+      "step": 589
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5669230520454114e-06,
+      "loss": 0.5812,
+      "step": 590
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5357803348846087e-06,
+      "loss": 0.5743,
+      "step": 591
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504744944107055e-06,
+      "loss": 0.5741,
+      "step": 592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4738173949978737e-06,
+      "loss": 0.6024,
+      "step": 593
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.442998201051677e-06,
+      "loss": 0.5778,
+      "step": 594
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.412287873964052e-06,
+      "loss": 0.5701,
+      "step": 595
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3816869236230387e-06,
+      "loss": 0.6132,
+      "step": 596
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3511958581006874e-06,
+      "loss": 0.5816,
+      "step": 597
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.320815183644608e-06,
+      "loss": 0.5802,
+      "step": 598
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2905454046695817e-06,
+      "loss": 0.5927,
+      "step": 599
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2603870237491653e-06,
+      "loss": 0.5895,
+      "step": 600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2303405416073673e-06,
+      "loss": 0.5768,
+      "step": 601
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2004064571103154e-06,
+      "loss": 0.583,
+      "step": 602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1705852672579853e-06,
+      "loss": 0.5778,
+      "step": 603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1408774671759524e-06,
+      "loss": 0.5905,
+      "step": 604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.111283550107158e-06,
+      "loss": 0.6088,
+      "step": 605
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0818040074037303e-06,
+      "loss": 0.6004,
+      "step": 606
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052439328518817e-06,
+      "loss": 0.597,
+      "step": 607
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0231900009984793e-06,
+      "loss": 0.5485,
+      "step": 608
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.994056510473571e-06,
+      "loss": 0.5902,
+      "step": 609
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9650393406516885e-06,
+      "loss": 0.5635,
+      "step": 610
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9361389733091416e-06,
+      "loss": 0.5791,
+      "step": 611
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.907355888282948e-06,
+      "loss": 0.5661,
+      "step": 612
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.878690563462877e-06,
+      "loss": 0.596,
+      "step": 613
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8501434747835012e-06,
+      "loss": 0.561,
+      "step": 614
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8217150962163044e-06,
+      "loss": 0.5744,
+      "step": 615
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.793405899761803e-06,
+      "loss": 0.6,
+      "step": 616
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7652163554417276e-06,
+      "loss": 0.6088,
+      "step": 617
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.737146931291198e-06,
+      "loss": 0.5956,
+      "step": 618
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.709198093350961e-06,
+      "loss": 0.5821,
+      "step": 619
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6813703056596528e-06,
+      "loss": 0.5486,
+      "step": 620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6536640302461036e-06,
+      "loss": 0.5498,
+      "step": 621
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6260797271216475e-06,
+      "loss": 0.5392,
+      "step": 622
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5986178542724983e-06,
+      "loss": 0.5426,
+      "step": 623
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5712788676521382e-06,
+      "loss": 0.5824,
+      "step": 624
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.544063221173758e-06,
+      "loss": 0.5935,
+      "step": 625
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5169713667027162e-06,
+      "loss": 0.5529,
+      "step": 626
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.490003754049024e-06,
+      "loss": 0.5908,
+      "step": 627
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.463160830959894e-06,
+      "loss": 0.5749,
+      "step": 628
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4364430431122997e-06,
+      "loss": 0.5887,
+      "step": 629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4098508341055714e-06,
+      "loss": 0.6168,
+      "step": 630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3833846454540457e-06,
+      "loss": 0.5605,
+      "step": 631
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.357044916579714e-06,
+      "loss": 0.6113,
+      "step": 632
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3308320848049436e-06,
+      "loss": 0.5886,
+      "step": 633
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3047465853452058e-06,
+      "loss": 0.6024,
+      "step": 634
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.278788851301862e-06,
+      "loss": 0.5963,
+      "step": 635
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2529593136549623e-06,
+      "loss": 0.5684,
+      "step": 636
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2272584012560894e-06,
+      "loss": 0.5925,
+      "step": 637
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.201686540821245e-06,
+      "loss": 0.5529,
+      "step": 638
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.176244156923768e-06,
+      "loss": 0.5924,
+      "step": 639
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1509316719872718e-06,
+      "loss": 0.5543,
+      "step": 640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1257495062786404e-06,
+      "loss": 0.5518,
+      "step": 641
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100698077901049e-06,
+      "loss": 0.5693,
+      "step": 642
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0757778027870192e-06,
+      "loss": 0.5525,
+      "step": 643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0509890946915244e-06,
+      "loss": 0.5783,
+      "step": 644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.026332365185102e-06,
+      "loss": 0.5564,
+      "step": 645
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.001808023647035e-06,
+      "loss": 0.5712,
+      "step": 646
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9774164772585435e-06,
+      "loss": 0.5682,
+      "step": 647
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953158130996039e-06,
+      "loss": 0.533,
+      "step": 648
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9290333876243892e-06,
+      "loss": 0.5506,
+      "step": 649
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9050426476902296e-06,
+      "loss": 0.6095,
+      "step": 650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8811863095153182e-06,
+      "loss": 0.5872,
+      "step": 651
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8574647691899217e-06,
+      "loss": 0.5962,
+      "step": 652
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.833878420566243e-06,
+      "loss": 0.5853,
+      "step": 653
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8104276552518696e-06,
+      "loss": 0.616,
+      "step": 654
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7871128626032851e-06,
+      "loss": 0.5858,
+      "step": 655
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7639344297193927e-06,
+      "loss": 0.5746,
+      "step": 656
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7408927414351051e-06,
+      "loss": 0.5594,
+      "step": 657
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7179881803149355e-06,
+      "loss": 0.5367,
+      "step": 658
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.695221126646659e-06,
+      "loss": 0.5795,
+      "step": 659
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6725919584349882e-06,
+      "loss": 0.5415,
+      "step": 660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.650101051395313e-06,
+      "loss": 0.5419,
+      "step": 661
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6277487789474477e-06,
+      "loss": 0.5623,
+      "step": 662
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6055355122094352e-06,
+      "loss": 0.555,
+      "step": 663
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5834616199913877e-06,
+      "loss": 0.5766,
+      "step": 664
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5615274687893588e-06,
+      "loss": 0.5518,
+      "step": 665
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5397334227792692e-06,
+      "loss": 0.5426,
+      "step": 666
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5180798438108468e-06,
+      "loss": 0.5697,
+      "step": 667
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.496567091401625e-06,
+      "loss": 0.5884,
+      "step": 668
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4751955227309722e-06,
+      "loss": 0.5664,
+      "step": 669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4539654926341695e-06,
+      "loss": 0.5626,
+      "step": 670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4328773535965045e-06,
+      "loss": 0.5791,
+      "step": 671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4119314557474272e-06,
+      "loss": 0.5769,
+      "step": 672
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3911281468547432e-06,
+      "loss": 0.5395,
+      "step": 673
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3704677723188254e-06,
+      "loss": 0.5463,
+      "step": 674
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3499506751668933e-06,
+      "loss": 0.5307,
+      "step": 675
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3295771960473057e-06,
+      "loss": 0.5629,
+      "step": 676
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3093476732239129e-06,
+      "loss": 0.5902,
+      "step": 677
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2892624425704324e-06,
+      "loss": 0.5644,
+      "step": 678
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2693218375648886e-06,
+      "loss": 0.5903,
+      "step": 679
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249526189284057e-06,
+      "loss": 0.5831,
+      "step": 680
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.229875826397976e-06,
+      "loss": 0.5983,
+      "step": 681
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2103710751644916e-06,
+      "loss": 0.554,
+      "step": 682
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.191012259423835e-06,
+      "loss": 0.5906,
+      "step": 683
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1717997005932546e-06,
+      "loss": 0.5605,
+      "step": 684
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1527337176616704e-06,
+      "loss": 0.5725,
+      "step": 685
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.133814627184382e-06,
+      "loss": 0.5659,
+      "step": 686
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1150427432778078e-06,
+      "loss": 0.552,
+      "step": 687
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0964183776142834e-06,
+      "loss": 0.5884,
+      "step": 688
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0779418394168695e-06,
+      "loss": 0.5384,
+      "step": 689
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0596134354542287e-06,
+      "loss": 0.5287,
+      "step": 690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0414334700355277e-06,
+      "loss": 0.5616,
+      "step": 691
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0234022450053916e-06,
+      "loss": 0.5769,
+      "step": 692
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0055200597388793e-06,
+      "loss": 0.5718,
+      "step": 693
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.877872111365273e-07,
+      "loss": 0.5537,
+      "step": 694
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.702039936194075e-07,
+      "loss": 0.5825,
+      "step": 695
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.527706991242502e-07,
+      "loss": 0.5353,
+      "step": 696
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.354876170985927e-07,
+      "loss": 0.5586,
+      "step": 697
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.18355034495968e-07,
+      "loss": 0.5504,
+      "step": 698
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.013732357711469e-07,
+      "loss": 0.5497,
+      "step": 699
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.845425028754151e-07,
+      "loss": 0.5369,
+      "step": 700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.678631152518869e-07,
+      "loss": 0.5419,
+      "step": 701
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.513353498308741e-07,
+      "loss": 0.5273,
+      "step": 702
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.349594810252792e-07,
+      "loss": 0.5392,
+      "step": 703
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.187357807260432e-07,
+      "loss": 0.5411,
+      "step": 704
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.0266451829763e-07,
+      "loss": 0.5713,
+      "step": 705
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.867459605735606e-07,
+      "loss": 0.5693,
+      "step": 706
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.709803718519726e-07,
+      "loss": 0.5477,
+      "step": 707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.553680138912378e-07,
+      "loss": 0.5572,
+      "step": 708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.39909145905614e-07,
+      "loss": 0.5072,
+      "step": 709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.246040245609465e-07,
+      "loss": 0.5501,
+      "step": 710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.094529039704013e-07,
+      "loss": 0.5528,
+      "step": 711
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.944560356902452e-07,
+      "loss": 0.5601,
+      "step": 712
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.796136687156696e-07,
+      "loss": 0.5736,
+      "step": 713
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.649260494766674e-07,
+      "loss": 0.5616,
+      "step": 714
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.503934218339214e-07,
+      "loss": 0.5281,
+      "step": 715
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.36016027074774e-07,
+      "loss": 0.5233,
+      "step": 716
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.217941039092068e-07,
+      "loss": 0.5144,
+      "step": 717
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.077278884658888e-07,
+      "loss": 0.5313,
+      "step": 718
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.938176142882512e-07,
+      "loss": 0.5633,
+      "step": 719
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.800635123306053e-07,
+      "loss": 0.5463,
+      "step": 720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.664658109543186e-07,
+      "loss": 0.5405,
+      "step": 721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.530247359240093e-07,
+      "loss": 0.5892,
+      "step": 722
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.39740510403809e-07,
+      "loss": 0.5868,
+      "step": 723
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.266133549536578e-07,
+      "loss": 0.546,
+      "step": 724
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.136434875256335e-07,
+      "loss": 0.5413,
+      "step": 725
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.00831123460338e-07,
+      "loss": 0.5368,
+      "step": 726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.881764754833229e-07,
+      "loss": 0.5512,
+      "step": 727
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7567975370155894e-07,
+      "loss": 0.5552,
+      "step": 728
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.633411655999431e-07,
+      "loss": 0.5737,
+      "step": 729
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.511609160378549e-07,
+      "loss": 0.5205,
+      "step": 730
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.391392072457545e-07,
+      "loss": 0.5367,
+      "step": 731
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.272762388218332e-07,
+      "loss": 0.5573,
+      "step": 732
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.155722077286872e-07,
+      "loss": 0.551,
+      "step": 733
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0402730829005254e-07,
+      "loss": 0.5845,
+      "step": 734
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9264173218758083e-07,
+      "loss": 0.557,
+      "step": 735
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8141566845765645e-07,
+      "loss": 0.5567,
+      "step": 736
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.703493034882544e-07,
+      "loss": 0.5477,
+      "step": 737
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.594428210158496e-07,
+      "loss": 0.5048,
+      "step": 738
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4869640212236155e-07,
+      "loss": 0.5243,
+      "step": 739
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3811022523215333e-07,
+      "loss": 0.5233,
+      "step": 740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2768446610906834e-07,
+      "loss": 0.5494,
+      "step": 741
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.174192978535051e-07,
+      "loss": 0.581,
+      "step": 742
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.073148908995538e-07,
+      "loss": 0.5739,
+      "step": 743
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.973714130121563e-07,
+      "loss": 0.5195,
+      "step": 744
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8758902928432976e-07,
+      "loss": 0.5319,
+      "step": 745
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.779679021344206e-07,
+      "loss": 0.5501,
+      "step": 746
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.685081913034082e-07,
+      "loss": 0.5707,
+      "step": 747
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.592100538522524e-07,
+      "loss": 0.5638,
+      "step": 748
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.500736441592866e-07,
+      "loss": 0.5396,
+      "step": 749
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.410991139176566e-07,
+      "loss": 0.5503,
+      "step": 750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3228661213279824e-07,
+      "loss": 0.5876,
+      "step": 751
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2363628511996692e-07,
+      "loss": 0.551,
+      "step": 752
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1514827650180425e-07,
+      "loss": 0.5742,
+      "step": 753
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0682272720595642e-07,
+      "loss": 0.5494,
+      "step": 754
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9865977546273506e-07,
+      "loss": 0.5381,
+      "step": 755
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9065955680282245e-07,
+      "loss": 0.5296,
+      "step": 756
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8282220405501428e-07,
+      "loss": 0.5507,
+      "step": 757
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7514784734402267e-07,
+      "loss": 0.5407,
+      "step": 758
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6763661408831677e-07,
+      "loss": 0.557,
+      "step": 759
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6028862899799657e-07,
+      "loss": 0.5233,
+      "step": 760
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5310401407273246e-07,
+      "loss": 0.5374,
+      "step": 761
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.460828885997334e-07,
+      "loss": 0.5456,
+      "step": 762
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3922536915177533e-07,
+      "loss": 0.5199,
+      "step": 763
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3253156958525136e-07,
+      "loss": 0.5828,
+      "step": 764
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2600160103829584e-07,
+      "loss": 0.5235,
+      "step": 765
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1963557192892772e-07,
+      "loss": 0.5223,
+      "step": 766
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1343358795326109e-07,
+      "loss": 0.5694,
+      "step": 767
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0739575208373987e-07,
+      "loss": 0.5463,
+      "step": 768
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.015221645674358e-07,
+      "loss": 0.5422,
+      "step": 769
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.581292292437982e-08,
+      "loss": 0.5672,
+      "step": 770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.026812194594448e-08,
+      "loss": 0.5477,
+      "step": 771
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.488785369327068e-08,
+      "loss": 0.5307,
+      "step": 772
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.96722074957379e-08,
+      "loss": 0.4877,
+      "step": 773
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.462126994948304e-08,
+      "loss": 0.5247,
+      "step": 774
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.973512491595724e-08,
+      "loss": 0.5749,
+      "step": 775
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.501385352054246e-08,
+      "loss": 0.5419,
+      "step": 776
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.045753415119593e-08,
+      "loss": 0.5818,
+      "step": 777
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.606624245715453e-08,
+      "loss": 0.5378,
+      "step": 778
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.1840051347675777e-08,
+      "loss": 0.5655,
+      "step": 779
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7779030990826594e-08,
+      "loss": 0.5387,
+      "step": 780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.388324881232198e-08,
+      "loss": 0.4912,
+      "step": 781
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.015276949440261e-08,
+      "loss": 0.5372,
+      "step": 782
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6587654974761246e-08,
+      "loss": 0.5262,
+      "step": 783
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.318796444551464e-08,
+      "loss": 0.4982,
+      "step": 784
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.995375435222214e-08,
+      "loss": 0.5247,
+      "step": 785
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6885078392945295e-08,
+      "loss": 0.5299,
+      "step": 786
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3981987517359696e-08,
+      "loss": 0.566,
+      "step": 787
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1244529925906754e-08,
+      "loss": 0.5199,
+      "step": 788
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8672751068995464e-08,
+      "loss": 0.5561,
+      "step": 789
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6266693646245224e-08,
+      "loss": 0.5584,
+      "step": 790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.402639760577862e-08,
+      "loss": 0.5043,
+      "step": 791
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1951900143558626e-08,
+      "loss": 0.5549,
+      "step": 792
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0043235702770215e-08,
+      "loss": 0.5605,
+      "step": 793
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.300435973246368e-09,
+      "loss": 0.5288,
+      "step": 794
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.723529890946268e-09,
+      "loss": 0.5083,
+      "step": 795
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.312543637471246e-09,
+      "loss": 0.5909,
+      "step": 796
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.067500639634014e-09,
+      "loss": 0.5326,
+      "step": 797
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9884215690634265e-09,
+      "loss": 0.5523,
+      "step": 798
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.075324341868079e-09,
+      "loss": 0.5396,
+      "step": 799
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.328224118333221e-09,
+      "loss": 0.5256,
+      "step": 800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.471333026742856e-10,
+      "loss": 0.5419,
+      "step": 801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.320615428270557e-10,
+      "loss": 0.5162,
+      "step": 802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.30157302877943e-11,
+      "loss": 0.54,
+      "step": 803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0,
+      "loss": 0.5478,
+      "step": 804
+    },
+    {
+      "epoch": 2.99,
+      "step": 804,
+      "total_flos": 4.313168111122514e+17,
+      "train_loss": 0.6899483537866702,
+      "train_runtime": 4131.3363,
+      "train_samples_per_second": 49.831,
+      "train_steps_per_second": 0.195
+    }
+  ],
+  "max_steps": 804,
+  "num_train_epochs": 3,
+  "total_flos": 4.313168111122514e+17,
+  "trial_name": null,
+  "trial_params": null
+}