diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,46902 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999936,
+  "eval_steps": 500,
+  "global_step": 7812,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-07,
+      "loss": 7.2337,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 7.1658,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-07,
+      "loss": 7.2814,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 7.2089,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 7.4849,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 7.4582,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4893617021276598e-06,
+      "loss": 7.215,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 7.3466,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9148936170212767e-06,
+      "loss": 7.3562,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 7.0391,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3404255319148935e-06,
+      "loss": 7.196,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 7.3886,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7659574468085106e-06,
+      "loss": 7.1922,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9787234042553196e-06,
+      "loss": 6.9816,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1914893617021277e-06,
+      "loss": 6.9384,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 7.0205,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6170212765957445e-06,
+      "loss": 6.8183,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 6.852,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0425531914893625e-06,
+      "loss": 6.4585,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 6.4334,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.468085106382979e-06,
+      "loss": 6.4218,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.680851063829787e-06,
+      "loss": 6.4161,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.893617021276596e-06,
+      "loss": 6.2431,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 6.0215,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 6.3608,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.531914893617021e-06,
+      "loss": 6.1681,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.74468085106383e-06,
+      "loss": 5.8055,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.957446808510639e-06,
+      "loss": 5.856,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.170212765957447e-06,
+      "loss": 5.7973,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.3829787234042555e-06,
+      "loss": 5.8657,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.595744680851064e-06,
+      "loss": 5.7734,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 5.7149,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.021276595744682e-06,
+      "loss": 5.5112,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.234042553191489e-06,
+      "loss": 5.4984,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 5.3471,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 5.4807,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 5.3411,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.085106382978725e-06,
+      "loss": 5.4824,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.297872340425532e-06,
+      "loss": 5.324,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 5.3449,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.72340425531915e-06,
+      "loss": 5.0277,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 5.3652,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.148936170212767e-06,
+      "loss": 5.3244,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.361702127659574e-06,
+      "loss": 5.1823,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.574468085106383e-06,
+      "loss": 5.1715,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 5.1764,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 5.201,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0212765957446808e-05,
+      "loss": 5.1505,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0425531914893617e-05,
+      "loss": 5.0963,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 4.9216,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0851063829787235e-05,
+      "loss": 5.0673,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1063829787234042e-05,
+      "loss": 4.9394,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1276595744680851e-05,
+      "loss": 5.1486,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.148936170212766e-05,
+      "loss": 5.1486,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.170212765957447e-05,
+      "loss": 5.0461,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1914893617021278e-05,
+      "loss": 5.0349,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2127659574468086e-05,
+      "loss": 5.0316,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2340425531914895e-05,
+      "loss": 4.9805,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2553191489361702e-05,
+      "loss": 4.8778,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 4.8611,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2978723404255318e-05,
+      "loss": 5.0053,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3191489361702127e-05,
+      "loss": 4.9357,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3404255319148936e-05,
+      "loss": 4.9937,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3617021276595745e-05,
+      "loss": 4.8656,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3829787234042554e-05,
+      "loss": 4.9931,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4042553191489363e-05,
+      "loss": 4.817,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4255319148936172e-05,
+      "loss": 4.9878,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4468085106382978e-05,
+      "loss": 4.9053,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4680851063829787e-05,
+      "loss": 4.7607,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4893617021276596e-05,
+      "loss": 4.9439,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5106382978723405e-05,
+      "loss": 4.9002,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5319148936170214e-05,
+      "loss": 4.6928,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5531914893617023e-05,
+      "loss": 4.9673,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5744680851063832e-05,
+      "loss": 4.7843,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.595744680851064e-05,
+      "loss": 4.8585,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.617021276595745e-05,
+      "loss": 4.9782,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6382978723404255e-05,
+      "loss": 4.8606,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6595744680851064e-05,
+      "loss": 4.8019,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6808510638297873e-05,
+      "loss": 4.838,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 4.9076,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.723404255319149e-05,
+      "loss": 4.8275,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74468085106383e-05,
+      "loss": 4.9279,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7659574468085106e-05,
+      "loss": 4.8378,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7872340425531915e-05,
+      "loss": 4.6865,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8085106382978724e-05,
+      "loss": 4.894,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8297872340425533e-05,
+      "loss": 4.8122,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.851063829787234e-05,
+      "loss": 4.9217,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8723404255319148e-05,
+      "loss": 4.9306,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8936170212765957e-05,
+      "loss": 4.9778,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9148936170212766e-05,
+      "loss": 4.9594,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9361702127659575e-05,
+      "loss": 4.9742,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9574468085106384e-05,
+      "loss": 4.8013,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9787234042553193e-05,
+      "loss": 4.7021,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 4.7574,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0212765957446807e-05,
+      "loss": 4.7327,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0425531914893616e-05,
+      "loss": 4.7573,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0638297872340425e-05,
+      "loss": 4.6812,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0851063829787234e-05,
+      "loss": 4.8831,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1063829787234043e-05,
+      "loss": 4.6588,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 4.8483,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.148936170212766e-05,
+      "loss": 4.7552,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.170212765957447e-05,
+      "loss": 4.6502,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191489361702128e-05,
+      "loss": 4.8011,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2127659574468085e-05,
+      "loss": 4.5011,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2340425531914894e-05,
+      "loss": 4.693,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2553191489361703e-05,
+      "loss": 4.8079,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2765957446808512e-05,
+      "loss": 4.7065,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.297872340425532e-05,
+      "loss": 4.7952,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.319148936170213e-05,
+      "loss": 4.7126,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.340425531914894e-05,
+      "loss": 4.8334,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3617021276595748e-05,
+      "loss": 4.6695,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3829787234042557e-05,
+      "loss": 4.592,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4042553191489362e-05,
+      "loss": 4.81,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.425531914893617e-05,
+      "loss": 4.9104,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.446808510638298e-05,
+      "loss": 4.8179,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.468085106382979e-05,
+      "loss": 4.6547,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.48936170212766e-05,
+      "loss": 4.864,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5106382978723404e-05,
+      "loss": 4.8087,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5319148936170213e-05,
+      "loss": 4.7246,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5531914893617022e-05,
+      "loss": 4.6586,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.574468085106383e-05,
+      "loss": 4.7445,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5957446808510637e-05,
+      "loss": 4.6976,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6170212765957446e-05,
+      "loss": 4.852,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6382978723404255e-05,
+      "loss": 4.8769,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6595744680851064e-05,
+      "loss": 4.5629,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6808510638297873e-05,
+      "loss": 4.633,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.702127659574468e-05,
+      "loss": 4.8485,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.723404255319149e-05,
+      "loss": 4.917,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.74468085106383e-05,
+      "loss": 4.7839,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.765957446808511e-05,
+      "loss": 4.7073,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7872340425531918e-05,
+      "loss": 4.7181,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8085106382978727e-05,
+      "loss": 4.7087,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8297872340425536e-05,
+      "loss": 4.8666,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8510638297872345e-05,
+      "loss": 4.8129,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8723404255319154e-05,
+      "loss": 4.6487,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8936170212765956e-05,
+      "loss": 4.7248,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9148936170212765e-05,
+      "loss": 4.7094,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9361702127659574e-05,
+      "loss": 4.6201,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9574468085106383e-05,
+      "loss": 4.7941,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9787234042553192e-05,
+      "loss": 4.7119,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-05,
+      "loss": 4.7826,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.021276595744681e-05,
+      "loss": 4.8585,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.042553191489362e-05,
+      "loss": 4.7802,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.063829787234043e-05,
+      "loss": 4.7089,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.085106382978723e-05,
+      "loss": 4.5336,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1063829787234046e-05,
+      "loss": 4.7329,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.127659574468085e-05,
+      "loss": 4.5218,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1489361702127664e-05,
+      "loss": 4.6917,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.170212765957447e-05,
+      "loss": 4.7,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.191489361702128e-05,
+      "loss": 4.65,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.212765957446809e-05,
+      "loss": 4.7023,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.23404255319149e-05,
+      "loss": 4.6216,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2553191489361705e-05,
+      "loss": 4.5945,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.276595744680851e-05,
+      "loss": 4.6735,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2978723404255317e-05,
+      "loss": 4.6966,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.319148936170213e-05,
+      "loss": 4.707,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3404255319148935e-05,
+      "loss": 4.554,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.361702127659575e-05,
+      "loss": 4.5863,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382978723404255e-05,
+      "loss": 4.8523,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4042553191489365e-05,
+      "loss": 4.6818,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.425531914893617e-05,
+      "loss": 4.5766,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.446808510638298e-05,
+      "loss": 4.787,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.468085106382979e-05,
+      "loss": 4.7453,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48936170212766e-05,
+      "loss": 4.6623,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5106382978723407e-05,
+      "loss": 4.672,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.531914893617021e-05,
+      "loss": 4.6886,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5531914893617025e-05,
+      "loss": 4.7795,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.574468085106383e-05,
+      "loss": 4.5524,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.595744680851064e-05,
+      "loss": 4.8248,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.617021276595745e-05,
+      "loss": 4.5389,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.638297872340426e-05,
+      "loss": 4.7269,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6595744680851066e-05,
+      "loss": 4.6433,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.680851063829787e-05,
+      "loss": 4.5755,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.702127659574468e-05,
+      "loss": 4.6813,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.723404255319149e-05,
+      "loss": 4.7619,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7446808510638295e-05,
+      "loss": 4.5635,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.765957446808511e-05,
+      "loss": 4.7317,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.787234042553191e-05,
+      "loss": 4.5974,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8085106382978726e-05,
+      "loss": 4.63,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.829787234042553e-05,
+      "loss": 4.6693,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8510638297872344e-05,
+      "loss": 4.6714,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.872340425531915e-05,
+      "loss": 4.6406,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.893617021276596e-05,
+      "loss": 4.6028,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.914893617021277e-05,
+      "loss": 4.6118,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.936170212765958e-05,
+      "loss": 4.7598,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9574468085106385e-05,
+      "loss": 4.7055,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.97872340425532e-05,
+      "loss": 4.5894,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 4.5895,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0212765957446816e-05,
+      "loss": 4.6393,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0425531914893614e-05,
+      "loss": 4.6297,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.063829787234043e-05,
+      "loss": 4.5015,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.085106382978723e-05,
+      "loss": 4.586,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1063829787234045e-05,
+      "loss": 4.7105,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.127659574468085e-05,
+      "loss": 4.5832,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.148936170212766e-05,
+      "loss": 4.8557,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.170212765957447e-05,
+      "loss": 4.6022,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.191489361702128e-05,
+      "loss": 4.58,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2127659574468086e-05,
+      "loss": 4.5312,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.23404255319149e-05,
+      "loss": 4.6516,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 4.6527,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.276595744680851e-05,
+      "loss": 4.5332,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.297872340425532e-05,
+      "loss": 4.53,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.319148936170213e-05,
+      "loss": 4.6744,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.340425531914894e-05,
+      "loss": 4.6157,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3617021276595746e-05,
+      "loss": 4.5896,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.382978723404256e-05,
+      "loss": 4.5386,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4042553191489364e-05,
+      "loss": 4.5069,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.425531914893617e-05,
+      "loss": 4.6042,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4468085106382975e-05,
+      "loss": 4.5368,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.468085106382979e-05,
+      "loss": 4.6522,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.489361702127659e-05,
+      "loss": 4.4118,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5106382978723406e-05,
+      "loss": 4.578,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.531914893617021e-05,
+      "loss": 4.6079,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5531914893617024e-05,
+      "loss": 4.4768,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.574468085106383e-05,
+      "loss": 4.4573,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.595744680851064e-05,
+      "loss": 4.4481,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.617021276595745e-05,
+      "loss": 4.5604,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.638297872340426e-05,
+      "loss": 4.52,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6595744680851065e-05,
+      "loss": 4.5189,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.680851063829788e-05,
+      "loss": 4.6628,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.702127659574468e-05,
+      "loss": 4.5661,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7234042553191496e-05,
+      "loss": 4.5492,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74468085106383e-05,
+      "loss": 4.4612,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7659574468085114e-05,
+      "loss": 4.5493,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.787234042553192e-05,
+      "loss": 4.4688,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8085106382978725e-05,
+      "loss": 4.5306,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.829787234042553e-05,
+      "loss": 4.4831,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.851063829787234e-05,
+      "loss": 4.5474,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.872340425531915e-05,
+      "loss": 4.6285,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.893617021276596e-05,
+      "loss": 4.3553,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9148936170212766e-05,
+      "loss": 4.5119,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.936170212765958e-05,
+      "loss": 4.5199,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9574468085106384e-05,
+      "loss": 4.5668,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97872340425532e-05,
+      "loss": 4.395,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-05,
+      "loss": 4.5401,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999999785110509e-05,
+      "loss": 4.4169,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999991404420723e-05,
+      "loss": 4.4615,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999998065994801e-05,
+      "loss": 4.6885,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999965617688795e-05,
+      "loss": 4.5812,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999994627764566e-05,
+      "loss": 4.4672,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999922639821945e-05,
+      "loss": 4.3989,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99998947042217e-05,
+      "loss": 4.2842,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999986247084974e-05,
+      "loss": 4.5529,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999982593971158e-05,
+      "loss": 4.514,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999978511081353e-05,
+      "loss": 4.4693,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999739984162596e-05,
+      "loss": 4.471,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999690559766535e-05,
+      "loss": 4.5484,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999963683763384e-05,
+      "loss": 4.34,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999957881777377e-05,
+      "loss": 4.3423,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999516500196264e-05,
+      "loss": 4.6455,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999449884912066e-05,
+      "loss": 4.488,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99993789719326e-05,
+      "loss": 4.6493,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999930376127008e-05,
+      "loss": 4.5202,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999922425293743e-05,
+      "loss": 4.4861,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99991404469483e-05,
+      "loss": 4.5522,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999905234331712e-05,
+      "loss": 4.4995,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9998959942059034e-05,
+      "loss": 4.5491,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999886324318992e-05,
+      "loss": 4.4177,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9998762246726395e-05,
+      "loss": 4.459,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999865695268584e-05,
+      "loss": 4.4436,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999854736108633e-05,
+      "loss": 4.378,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999843347194674e-05,
+      "loss": 4.4882,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999831528528662e-05,
+      "loss": 4.506,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999819280112629e-05,
+      "loss": 4.5786,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9998066019486825e-05,
+      "loss": 4.3976,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999793494039e-05,
+      "loss": 4.4671,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999779956385837e-05,
+      "loss": 4.4886,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999765988991518e-05,
+      "loss": 4.3905,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999751591858447e-05,
+      "loss": 4.3452,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999736764989097e-05,
+      "loss": 4.2806,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999721508386018e-05,
+      "loss": 4.5325,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999705822051832e-05,
+      "loss": 4.5635,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9996897059892366e-05,
+      "loss": 4.5472,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999673160201002e-05,
+      "loss": 4.3622,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999656184689972e-05,
+      "loss": 4.4795,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9996387794590656e-05,
+      "loss": 4.4696,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999620944511274e-05,
+      "loss": 4.4223,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9996026798496656e-05,
+      "loss": 4.3658,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999583985477377e-05,
+      "loss": 4.4523,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999564861397624e-05,
+      "loss": 4.3669,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999545307613694e-05,
+      "loss": 4.5325,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999525324128949e-05,
+      "loss": 4.3615,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999504910946824e-05,
+      "loss": 4.485,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999484068070827e-05,
+      "loss": 4.3723,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999462795504543e-05,
+      "loss": 4.4909,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9994410932516276e-05,
+      "loss": 4.461,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9994189613158124e-05,
+      "loss": 4.4414,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9993963997009016e-05,
+      "loss": 4.4895,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999373408410775e-05,
+      "loss": 4.5494,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999349987449384e-05,
+      "loss": 4.5011,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999326136820755e-05,
+      "loss": 4.5134,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9993018565289884e-05,
+      "loss": 4.5029,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9992771465782585e-05,
+      "loss": 4.3493,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999252006972813e-05,
+      "loss": 4.2876,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999226437716974e-05,
+      "loss": 4.3559,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999200438815136e-05,
+      "loss": 4.3004,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99917401027177e-05,
+      "loss": 4.2749,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999147152091419e-05,
+      "loss": 4.4462,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999119864278699e-05,
+      "loss": 4.3016,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999092146838302e-05,
+      "loss": 4.3309,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9990639997749947e-05,
+      "loss": 4.547,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999035423093612e-05,
+      "loss": 4.4396,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9990064167990705e-05,
+      "loss": 4.4364,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9989769808963535e-05,
+      "loss": 4.5768,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998947115390524e-05,
+      "loss": 4.4782,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998916820286715e-05,
+      "loss": 4.4633,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998886095590134e-05,
+      "loss": 4.5097,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9988549413060636e-05,
+      "loss": 4.3728,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99882335743986e-05,
+      "loss": 4.4206,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9987913439969525e-05,
+      "loss": 4.3592,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998758900982845e-05,
+      "loss": 4.2701,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998726028403113e-05,
+      "loss": 4.4421,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99869272626341e-05,
+      "loss": 4.3912,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9986589945694596e-05,
+      "loss": 4.5098,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998624833327061e-05,
+      "loss": 4.4267,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998590242542087e-05,
+      "loss": 4.4087,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998555222220484e-05,
+      "loss": 4.4655,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998519772368273e-05,
+      "loss": 4.4328,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998483892991549e-05,
+      "loss": 4.2626,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998447584096477e-05,
+      "loss": 4.4173,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998410845689301e-05,
+      "loss": 4.3162,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998373677776337e-05,
+      "loss": 4.3414,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9983360803639745e-05,
+      "loss": 4.4547,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998298053458676e-05,
+      "loss": 4.4641,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99825959706698e-05,
+      "loss": 4.2699,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9982207111954963e-05,
+      "loss": 4.3979,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99818139585091e-05,
+      "loss": 4.4327,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998141651039982e-05,
+      "loss": 4.4638,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998101476769542e-05,
+      "loss": 4.4299,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998060873046499e-05,
+      "loss": 4.4879,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99801983987783e-05,
+      "loss": 4.3727,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997978377270592e-05,
+      "loss": 4.3434,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997936485231912e-05,
+      "loss": 4.5021,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9978941637689915e-05,
+      "loss": 4.3693,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997851412889106e-05,
+      "loss": 4.4021,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9978082325996045e-05,
+      "loss": 4.2811,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997764622907912e-05,
+      "loss": 4.274,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997720583821523e-05,
+      "loss": 4.3155,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9976761153480106e-05,
+      "loss": 4.3461,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997631217495018e-05,
+      "loss": 4.3568,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9975858902702645e-05,
+      "loss": 4.4546,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9975401336815406e-05,
+      "loss": 4.3216,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997493947736716e-05,
+      "loss": 4.3728,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997447332443727e-05,
+      "loss": 4.3644,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997400287810588e-05,
+      "loss": 4.4836,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997352813845388e-05,
+      "loss": 4.3785,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9973049105562875e-05,
+      "loss": 4.322,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997256577951521e-05,
+      "loss": 4.4221,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9972078160393985e-05,
+      "loss": 4.3441,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997158624828303e-05,
+      "loss": 4.3138,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9971090043266896e-05,
+      "loss": 4.2845,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99705895454309e-05,
+      "loss": 4.4062,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9970084754861066e-05,
+      "loss": 4.3816,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99695756716442e-05,
+      "loss": 4.2521,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996906229586779e-05,
+      "loss": 4.3857,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996854462762011e-05,
+      "loss": 4.2496,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9968022666990145e-05,
+      "loss": 4.3496,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996749641406764e-05,
+      "loss": 4.2856,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9966965868943035e-05,
+      "loss": 4.5328,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9966431031707575e-05,
+      "loss": 4.4632,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996589190245318e-05,
+      "loss": 4.5417,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9965348481272525e-05,
+      "loss": 4.3207,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996480076825906e-05,
+      "loss": 4.5006,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9964248763506916e-05,
+      "loss": 4.4376,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9963692467111e-05,
+      "loss": 4.1864,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9963131879166945e-05,
+      "loss": 4.3036,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996256699977112e-05,
+      "loss": 4.3183,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996199782902065e-05,
+      "loss": 4.2006,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996142436701336e-05,
+      "loss": 4.3062,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996084661384784e-05,
+      "loss": 4.4533,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9960264569623414e-05,
+      "loss": 4.4506,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9959678234440155e-05,
+      "loss": 4.2794,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995908760839884e-05,
+      "loss": 4.4515,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9958492691601023e-05,
+      "loss": 4.3762,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995789348414897e-05,
+      "loss": 4.2043,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9957289986145675e-05,
+      "loss": 4.3852,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9956682197694915e-05,
+      "loss": 4.3161,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995607011890116e-05,
+      "loss": 4.1823,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9955453749869626e-05,
+      "loss": 4.3139,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99548330907063e-05,
+      "loss": 4.4216,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995420814151785e-05,
+      "loss": 4.3649,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995357890241174e-05,
+      "loss": 4.2391,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9952945373496126e-05,
+      "loss": 4.3629,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995230755487992e-05,
+      "loss": 4.3374,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995166544667277e-05,
+      "loss": 4.2797,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995101904898507e-05,
+      "loss": 4.2669,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9950368361927943e-05,
+      "loss": 4.2824,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9949713385613236e-05,
+      "loss": 4.2599,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994905412015356e-05,
+      "loss": 4.3103,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994839056566225e-05,
+      "loss": 4.4237,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994772272225337e-05,
+      "loss": 4.4225,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994705059004174e-05,
+      "loss": 4.1232,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9946374169142893e-05,
+      "loss": 4.2917,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994569345967314e-05,
+      "loss": 4.3722,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994500846174947e-05,
+      "loss": 4.4237,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994431917548966e-05,
+      "loss": 4.2891,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994362560101221e-05,
+      "loss": 4.2416,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994292773843635e-05,
+      "loss": 4.2284,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9942225587882044e-05,
+      "loss": 4.3645,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9941519149470004e-05,
+      "loss": 4.1656,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994080842332168e-05,
+      "loss": 4.383,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994009340955924e-05,
+      "loss": 4.4034,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993937410830562e-05,
+      "loss": 4.1883,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9938650519684475e-05,
+      "loss": 4.2756,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993792264382018e-05,
+      "loss": 4.3285,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993719048083788e-05,
+      "loss": 4.3822,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993645403086344e-05,
+      "loss": 4.2983,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9935713294023465e-05,
+      "loss": 4.4615,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9934968270445295e-05,
+      "loss": 4.3054,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993421896025701e-05,
+      "loss": 4.3968,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993346536358742e-05,
+      "loss": 4.252,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993270748056608e-05,
+      "loss": 4.3086,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993194531132328e-05,
+      "loss": 4.2844,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993117885599004e-05,
+      "loss": 4.2721,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993040811469815e-05,
+      "loss": 4.279,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992963308758006e-05,
+      "loss": 4.363,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992885377476904e-05,
+      "loss": 4.4436,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992807017639907e-05,
+      "loss": 4.2437,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992728229260484e-05,
+      "loss": 4.3669,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9926490123521795e-05,
+      "loss": 4.2752,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992569366928612e-05,
+      "loss": 4.182,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9924892930034755e-05,
+      "loss": 4.2939,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992408790590534e-05,
+      "loss": 4.3078,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992327859703626e-05,
+      "loss": 4.3685,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992246500356665e-05,
+      "loss": 4.2438,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9921647125636394e-05,
+      "loss": 4.3564,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992082496338607e-05,
+      "loss": 4.3453,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991999851695703e-05,
+      "loss": 4.1605,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991916778649135e-05,
+      "loss": 4.3278,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9918332772131837e-05,
+      "loss": 4.3776,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9917493474022045e-05,
+      "loss": 4.4306,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991664989230626e-05,
+      "loss": 4.4161,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9915802027129496e-05,
+      "loss": 4.3106,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991494987863751e-05,
+      "loss": 4.285,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9914093446976804e-05,
+      "loss": 4.3871,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991323273229461e-05,
+      "loss": 4.1874,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9912367734738886e-05,
+      "loss": 4.254,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9911498454458336e-05,
+      "loss": 4.2568,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9910624891602407e-05,
+      "loss": 4.2675,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9909747046321273e-05,
+      "loss": 4.3181,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9908864918765844e-05,
+      "loss": 4.3337,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990797850908776e-05,
+      "loss": 4.2493,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990708781743942e-05,
+      "loss": 4.2586,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990619284397393e-05,
+      "loss": 4.2823,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990529358884516e-05,
+      "loss": 4.1929,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990439005220769e-05,
+      "loss": 4.339,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990348223421685e-05,
+      "loss": 4.2606,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990257013502872e-05,
+      "loss": 4.2809,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9901653754800073e-05,
+      "loss": 4.2727,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990073309368847e-05,
+      "loss": 4.4037,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989980815185217e-05,
+      "loss": 4.3027,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9898878929450186e-05,
+      "loss": 4.3533,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989794542664227e-05,
+      "loss": 4.3777,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989700764358888e-05,
+      "loss": 4.2983,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989606558045126e-05,
+      "loss": 4.1589,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989511923739134e-05,
+      "loss": 4.2224,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989416861457181e-05,
+      "loss": 4.3508,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98932137121561e-05,
+      "loss": 4.4265,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9892254530308366e-05,
+      "loss": 4.2358,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98912910691935e-05,
+      "loss": 4.268,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989032332897714e-05,
+      "loss": 4.2914,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988935130982564e-05,
+      "loss": 4.3437,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9888375011906105e-05,
+      "loss": 4.2451,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988739443538638e-05,
+      "loss": 4.3485,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988640958043504e-05,
+      "loss": 4.2904,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988542044722138e-05,
+      "loss": 4.2685,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9884427035915444e-05,
+      "loss": 4.31,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9883429346688014e-05,
+      "loss": 4.2974,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9882427379710616e-05,
+      "loss": 4.2759,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988142113515548e-05,
+      "loss": 4.2475,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98804106131956e-05,
+      "loss": 4.2895,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98793958140047e-05,
+      "loss": 4.3075,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9878376737757235e-05,
+      "loss": 4.2837,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98773533846284e-05,
+      "loss": 4.2801,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9876325754794106e-05,
+      "loss": 4.3415,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987529384843103e-05,
+      "loss": 4.1331,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987425766571655e-05,
+      "loss": 4.2245,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9873217206828826e-05,
+      "loss": 4.276,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98721724719467e-05,
+      "loss": 4.37,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987112346124978e-05,
+      "loss": 4.1689,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9870070174918416e-05,
+      "loss": 4.1774,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986901261313366e-05,
+      "loss": 4.3389,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986795077607733e-05,
+      "loss": 4.375,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986688466393198e-05,
+      "loss": 4.2044,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9865814276880865e-05,
+      "loss": 4.3221,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9864739615108005e-05,
+      "loss": 4.3346,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986366067879815e-05,
+      "loss": 4.3673,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986257746813678e-05,
+      "loss": 4.2907,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986148998331012e-05,
+      "loss": 4.224,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98603982245051e-05,
+      "loss": 4.286,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985930219190943e-05,
+      "loss": 4.335,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9858201885711507e-05,
+      "loss": 4.1796,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985709730610051e-05,
+      "loss": 4.2744,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985598845326631e-05,
+      "loss": 4.2341,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9854875327399546e-05,
+      "loss": 4.3148,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985375792869157e-05,
+      "loss": 4.2244,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9852636257334474e-05,
+      "loss": 4.1571,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985151031352109e-05,
+      "loss": 4.4269,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985038009744498e-05,
+      "loss": 4.2362,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984924560930044e-05,
+      "loss": 4.2689,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984810684928251e-05,
+      "loss": 4.3375,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9846963817586936e-05,
+      "loss": 4.3264,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984581651441024e-05,
+      "loss": 4.2728,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984466493994964e-05,
+      "loss": 4.3042,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984350909440312e-05,
+      "loss": 4.182,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984234897796937e-05,
+      "loss": 4.336,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9841184590847835e-05,
+      "loss": 4.2538,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.984001593323868e-05,
+      "loss": 4.1623,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9838843005342825e-05,
+      "loss": 4.2254,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98376658073619e-05,
+      "loss": 4.1158,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983648433949828e-05,
+      "loss": 4.2286,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983529860195507e-05,
+      "loss": 4.193,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9834108594936114e-05,
+      "loss": 4.2629,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983291431864599e-05,
+      "loss": 4.2207,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983171577329001e-05,
+      "loss": 4.1816,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983051295907421e-05,
+      "loss": 4.3182,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982930587620537e-05,
+      "loss": 4.2965,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982809452489101e-05,
+      "loss": 4.3496,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982687890533937e-05,
+      "loss": 4.1924,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982565901775942e-05,
+      "loss": 4.3465,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982443486236089e-05,
+      "loss": 4.2537,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982320643935421e-05,
+      "loss": 4.3201,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982197374895058e-05,
+      "loss": 4.1764,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982073679136189e-05,
+      "loss": 4.2542,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9819495566800803e-05,
+      "loss": 4.1516,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98182500754807e-05,
+      "loss": 4.182,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981700031761568e-05,
+      "loss": 4.1636,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9815746293420606e-05,
+      "loss": 4.3477,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981448800311106e-05,
+      "loss": 4.2706,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981322544690335e-05,
+      "loss": 4.3059,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981195862501453e-05,
+      "loss": 4.2529,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981068753766237e-05,
+      "loss": 4.3105,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9809412185065393e-05,
+      "loss": 4.2473,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980813256744286e-05,
+      "loss": 4.2805,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980684868501472e-05,
+      "loss": 4.2716,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980556053800171e-05,
+      "loss": 4.2991,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980426812662527e-05,
+      "loss": 4.1023,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9802971451107594e-05,
+      "loss": 4.1849,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9801670511671576e-05,
+      "loss": 4.3856,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9800365308540875e-05,
+      "loss": 4.1176,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9799055841939866e-05,
+      "loss": 4.2983,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9797742112093664e-05,
+      "loss": 4.1224,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9796424119228116e-05,
+      "loss": 4.121,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9795101863569785e-05,
+      "loss": 4.3151,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979377534534601e-05,
+      "loss": 4.2109,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97924445647848e-05,
+      "loss": 4.2199,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979110952211497e-05,
+      "loss": 4.2667,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9789770217566e-05,
+      "loss": 4.0388,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9788426651368144e-05,
+      "loss": 4.256,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978707882375237e-05,
+      "loss": 4.2187,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9785726734950395e-05,
+      "loss": 4.292,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978437038519465e-05,
+      "loss": 4.2144,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9783009774718315e-05,
+      "loss": 4.3402,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9781644903755286e-05,
+      "loss": 4.2269,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978027577254021e-05,
+      "loss": 4.3,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977890238130845e-05,
+      "loss": 4.2131,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97775247302961e-05,
+      "loss": 4.2942,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9776142819740015e-05,
+      "loss": 4.3464,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977475664987774e-05,
+      "loss": 4.271,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977336622094759e-05,
+      "loss": 4.1769,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977197153318859e-05,
+      "loss": 4.1588,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97705725868405e-05,
+      "loss": 4.1968,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976916938214381e-05,
+      "loss": 4.2256,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976776191933976e-05,
+      "loss": 4.1985,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9766350198670306e-05,
+      "loss": 4.1972,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9764934220378134e-05,
+      "loss": 4.2437,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9763513984706675e-05,
+      "loss": 4.2647,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976208949190007e-05,
+      "loss": 4.2404,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976066074220322e-05,
+      "loss": 4.2902,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975922773586173e-05,
+      "loss": 4.2262,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9757790473121975e-05,
+      "loss": 4.2081,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975634895423101e-05,
+      "loss": 4.3114,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975490317943666e-05,
+      "loss": 4.2836,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975345314898747e-05,
+      "loss": 4.2095,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975199886313272e-05,
+      "loss": 4.2034,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9750540322122415e-05,
+      "loss": 4.2973,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974907752620729e-05,
+      "loss": 4.2504,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974761047563883e-05,
+      "loss": 4.0805,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9746139170669226e-05,
+      "loss": 4.2461,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9744663611551415e-05,
+      "loss": 4.2436,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974318379853907e-05,
+      "loss": 4.2598,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974169973188658e-05,
+      "loss": 4.222,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9740211411849075e-05,
+      "loss": 4.2847,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9738718838682417e-05,
+      "loss": 4.1573,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.973722201264319e-05,
+      "loss": 4.2265,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9735720933988726e-05,
+      "loss": 4.3118,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973421560297707e-05,
+      "loss": 4.1674,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9732706019867004e-05,
+      "loss": 4.1092,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9731192184918054e-05,
+      "loss": 4.1914,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9729674098390454e-05,
+      "loss": 4.2227,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972815176054519e-05,
+      "loss": 4.1736,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972662517164396e-05,
+      "loss": 4.0886,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972509433194921e-05,
+      "loss": 4.1576,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972355924172411e-05,
+      "loss": 4.1097,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9722019901232554e-05,
+      "loss": 4.2072,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9720476310739174e-05,
+      "loss": 4.3282,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971892847050933e-05,
+      "loss": 4.1893,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9717376380809113e-05,
+      "loss": 4.1579,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971582004190536e-05,
+      "loss": 4.2249,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97142594540656e-05,
+      "loss": 4.2394,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971269461755813e-05,
+      "loss": 4.2081,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971112553265196e-05,
+      "loss": 4.1296,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970955219961684e-05,
+      "loss": 4.3146,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970797461872323e-05,
+      "loss": 4.3542,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970639279024235e-05,
+      "loss": 4.1974,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970480671444613e-05,
+      "loss": 4.309,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970321639160723e-05,
+      "loss": 4.2974,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970162182199905e-05,
+      "loss": 4.1022,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970002300589571e-05,
+      "loss": 4.2424,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9698419943572074e-05,
+      "loss": 4.2356,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9696812635303714e-05,
+      "loss": 4.1358,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9695201081366955e-05,
+      "loss": 4.2532,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969358528203884e-05,
+      "loss": 4.1192,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9691965237597135e-05,
+      "loss": 4.2299,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9690340948320365e-05,
+      "loss": 4.1164,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968871241448774e-05,
+      "loss": 4.2414,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968707963637924e-05,
+      "loss": 4.1248,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9685442614275556e-05,
+      "loss": 4.2167,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96838013484581e-05,
+      "loss": 4.3087,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968215583920904e-05,
+      "loss": 4.1225,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968050608681125e-05,
+      "loss": 4.1481,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9678852091548336e-05,
+      "loss": 4.3202,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967719385370465e-05,
+      "loss": 4.1675,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967553137356527e-05,
+      "loss": 4.3814,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967386465141597e-05,
+      "loss": 4.195,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967219368754329e-05,
+      "loss": 4.1678,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96705184822345e-05,
+      "loss": 4.1264,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966883903577757e-05,
+      "loss": 4.1024,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9667155348461224e-05,
+      "loss": 4.1376,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9665467420574915e-05,
+      "loss": 4.1933,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966377525240881e-05,
+      "loss": 4.2306,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96620788442538e-05,
+      "loss": 4.0838,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966037819640153e-05,
+      "loss": 4.23,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965867330914438e-05,
+      "loss": 4.2637,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96569641827754e-05,
+      "loss": 4.2939,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965525081758844e-05,
+      "loss": 4.3396,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9653533213878033e-05,
+      "loss": 4.2186,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965181137193946e-05,
+      "loss": 4.2139,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9650085292068724e-05,
+      "loss": 4.0142,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964835497456255e-05,
+      "loss": 4.1496,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964662041971841e-05,
+      "loss": 4.1949,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96448816278345e-05,
+      "loss": 4.1278,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9643138599209724e-05,
+      "loss": 4.1939,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964139133414374e-05,
+      "loss": 4.2208,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9639639832936915e-05,
+      "loss": 4.2312,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9637884095890354e-05,
+      "loss": 4.2451,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963612412330589e-05,
+      "loss": 4.1073,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9634359915486086e-05,
+      "loss": 4.1212,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963259147273422e-05,
+      "loss": 4.3205,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963081879535432e-05,
+      "loss": 4.1191,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9629041883651124e-05,
+      "loss": 4.2793,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9627260737930105e-05,
+      "loss": 4.065,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962547535849745e-05,
+      "loss": 4.1027,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962368574566011e-05,
+      "loss": 4.1578,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9621891899725724e-05,
+      "loss": 4.185,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962009382100269e-05,
+      "loss": 4.2076,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961829150980009e-05,
+      "loss": 4.3376,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961648496642779e-05,
+      "loss": 4.2291,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961467419119634e-05,
+      "loss": 4.1834,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9612859184417045e-05,
+      "loss": 4.2018,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9611039946401916e-05,
+      "loss": 4.1176,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960921647746371e-05,
+      "loss": 4.0888,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960738877791589e-05,
+      "loss": 4.1551,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960555684807267e-05,
+      "loss": 4.1929,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9603720688248966e-05,
+      "loss": 4.1629,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960188029876045e-05,
+      "loss": 4.1513,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9600035679923506e-05,
+      "loss": 4.1615,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959818683205524e-05,
+      "loss": 4.3276,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9596333755473485e-05,
+      "loss": 4.314,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959447645049682e-05,
+      "loss": 4.2641,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959261491744452e-05,
+      "loss": 4.1148,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959074915663662e-05,
+      "loss": 4.0546,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9588879168393855e-05,
+      "loss": 4.295,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958700495303771e-05,
+      "loss": 4.3454,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958512651089037e-05,
+      "loss": 4.2474,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958324384227477e-05,
+      "loss": 4.1316,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9581356947514554e-05,
+      "loss": 4.087,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9579465826934115e-05,
+      "loss": 4.17,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9577570480858553e-05,
+      "loss": 4.2692,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957567090961369e-05,
+      "loss": 4.1645,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95737671135261e-05,
+      "loss": 4.2784,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957185909292306e-05,
+      "loss": 4.1428,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956994684813257e-05,
+      "loss": 4.1358,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9568030379483384e-05,
+      "loss": 4.2023,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9566109687304954e-05,
+      "loss": 4.3108,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956418477192748e-05,
+      "loss": 4.1218,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956225563368187e-05,
+      "loss": 4.2256,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956032227289976e-05,
+      "loss": 4.1919,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955838468991353e-05,
+      "loss": 4.2209,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955644288505627e-05,
+      "loss": 4.2133,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955449685866179e-05,
+      "loss": 4.1663,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955254661106464e-05,
+      "loss": 4.209,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9550592142600095e-05,
+      "loss": 4.0639,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954863345360414e-05,
+      "loss": 4.0238,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95466705444135e-05,
+      "loss": 4.1231,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954470341536563e-05,
+      "loss": 4.1493,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9542732066798695e-05,
+      "loss": 4.1705,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95407564990516e-05,
+      "loss": 4.0887,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953877671246395e-05,
+      "loss": 4.108,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9536792707376115e-05,
+      "loss": 4.3472,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953480448412915e-05,
+      "loss": 4.1376,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953281204306487e-05,
+      "loss": 4.1261,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953081538452578e-05,
+      "loss": 3.9352,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952881450885515e-05,
+      "loss": 4.2585,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952680941639694e-05,
+      "loss": 4.1487,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952480010749585e-05,
+      "loss": 4.181,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952278658249731e-05,
+      "loss": 4.1394,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952076884174747e-05,
+      "loss": 4.1511,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9518746885593184e-05,
+      "loss": 4.1347,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951672071438207e-05,
+      "loss": 4.2612,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951469032846244e-05,
+      "loss": 4.2781,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9512655728183343e-05,
+      "loss": 4.1289,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9510616913894554e-05,
+      "loss": 4.0215,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9508573885946564e-05,
+      "loss": 4.1904,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9506526644690595e-05,
+      "loss": 4.1705,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950447519047859e-05,
+      "loss": 4.2242,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9502419523663226e-05,
+      "loss": 4.2008,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9500359644597886e-05,
+      "loss": 4.0204,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.949829555363669e-05,
+      "loss": 4.184,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9496227251134484e-05,
+      "loss": 4.0342,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9494154737446825e-05,
+      "loss": 4.0916,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9492078012930005e-05,
+      "loss": 4.1207,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948999707794104e-05,
+      "loss": 4.1424,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9487911932837664e-05,
+      "loss": 4.1851,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948582257797834e-05,
+      "loss": 4.2056,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948372901372225e-05,
+      "loss": 4.1026,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.94816312404293e-05,
+      "loss": 4.1997,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947952925846013e-05,
+      "loss": 4.0944,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947742306817608e-05,
+      "loss": 4.1099,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9475312669939246e-05,
+      "loss": 4.1459,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9473198064112426e-05,
+      "loss": 4.1264,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9471079251059126e-05,
+      "loss": 4.182,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9468956231143626e-05,
+      "loss": 4.1316,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9466829004730874e-05,
+      "loss": 4.1716,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946469757218657e-05,
+      "loss": 4.2168,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946256193387714e-05,
+      "loss": 4.1873,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946042209016971e-05,
+      "loss": 4.0694,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945827804143216e-05,
+      "loss": 4.252,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9456129788033066e-05,
+      "loss": 3.9956,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9453977330341747e-05,
+      "loss": 4.181,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945182066872823e-05,
+      "loss": 4.1622,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944965980356327e-05,
+      "loss": 4.0775,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9447494735218344e-05,
+      "loss": 4.1395,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.944532546406565e-05,
+      "loss": 4.1781,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.944315199047811e-05,
+      "loss": 4.1255,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9440974314829385e-05,
+      "loss": 4.1317,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943879243749383e-05,
+      "loss": 4.2029,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943660635884653e-05,
+      "loss": 4.0465,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94344160792633e-05,
+      "loss": 4.1777,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.943222159912069e-05,
+      "loss": 4.0933,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9430022918795936e-05,
+      "loss": 4.2005,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942782003866703e-05,
+      "loss": 4.194,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942561295911267e-05,
+      "loss": 4.1286,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942340168051226e-05,
+      "loss": 4.2283,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942118620324597e-05,
+      "loss": 4.0956,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9418966527694664e-05,
+      "loss": 4.1956,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941674265423992e-05,
+      "loss": 4.1088,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9414514583264046e-05,
+      "loss": 4.2966,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9412282315150086e-05,
+      "loss": 3.9655,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941004585028177e-05,
+      "loss": 4.1929,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94078051890436e-05,
+      "loss": 4.1037,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9405560331820757e-05,
+      "loss": 4.1405,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940331127899915e-05,
+      "loss": 4.2694,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940105803096544e-05,
+      "loss": 4.0368,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939880058810696e-05,
+      "loss": 4.0665,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9396538950811815e-05,
+      "loss": 4.1276,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9394273119468795e-05,
+      "loss": 3.9915,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9392003094467415e-05,
+      "loss": 4.1201,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938972887619793e-05,
+      "loss": 4.1583,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.93874504650513e-05,
+      "loss": 4.1486,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938516786141921e-05,
+      "loss": 4.2081,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938288106569408e-05,
+      "loss": 4.2208,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9380590078269014e-05,
+      "loss": 4.0712,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937829489953787e-05,
+      "loss": 4.2284,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9375995529895215e-05,
+      "loss": 4.118,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9373691969736335e-05,
+      "loss": 4.1875,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937138421945725e-05,
+      "loss": 4.1014,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936907227945467e-05,
+      "loss": 4.1751,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9366756150126056e-05,
+      "loss": 4.1583,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936443583186958e-05,
+      "loss": 4.1414,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936211132508413e-05,
+      "loss": 4.1166,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9359782630169316e-05,
+      "loss": 4.0819,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935744974752545e-05,
+      "loss": 4.1291,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935511267755361e-05,
+      "loss": 4.0749,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9352771420655556e-05,
+      "loss": 3.9779,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935042597723376e-05,
+      "loss": 4.173,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934807634769145e-05,
+      "loss": 4.2245,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934572253243255e-05,
+      "loss": 4.1134,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9343364531861706e-05,
+      "loss": 4.2134,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9341002346384284e-05,
+      "loss": 4.1643,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.933863597640638e-05,
+      "loss": 4.1745,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.933626542233478e-05,
+      "loss": 4.1799,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.933389068457704e-05,
+      "loss": 4.0732,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.933151176354138e-05,
+      "loss": 4.1325,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932912865963677e-05,
+      "loss": 4.1223,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932674137327291e-05,
+      "loss": 4.1262,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932434990486017e-05,
+      "loss": 4.0392,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9321954254809697e-05,
+      "loss": 4.0059,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9319554423533335e-05,
+      "loss": 4.2616,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931715041144361e-05,
+      "loss": 4.2677,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9314742218953835e-05,
+      "loss": 4.0284,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931232984647799e-05,
+      "loss": 4.1033,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930991329443079e-05,
+      "loss": 4.3229,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930749256322767e-05,
+      "loss": 4.2261,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930506765328478e-05,
+      "loss": 4.185,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930263856501899e-05,
+      "loss": 4.0654,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9300205298847894e-05,
+      "loss": 4.2408,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92977678551898e-05,
+      "loss": 4.1955,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929532623446372e-05,
+      "loss": 3.9983,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9292880437089406e-05,
+      "loss": 4.0561,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929043046348732e-05,
+      "loss": 4.0388,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928797631407863e-05,
+      "loss": 4.1409,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9285517989285254e-05,
+      "loss": 4.1378,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928305548952978e-05,
+      "loss": 4.0858,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928058881523557e-05,
+      "loss": 4.2176,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9278117966826654e-05,
+      "loss": 4.2305,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92756429447278e-05,
+      "loss": 4.0491,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927316374936449e-05,
+      "loss": 4.1978,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927068038116294e-05,
+      "loss": 4.1337,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9268192840550064e-05,
+      "loss": 4.1664,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926570112795349e-05,
+      "loss": 4.1932,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926320524380159e-05,
+      "loss": 4.1454,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926070518852341e-05,
+      "loss": 4.1215,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9258200962548775e-05,
+      "loss": 3.9694,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925569256630815e-05,
+      "loss": 4.1647,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925318000023279e-05,
+      "loss": 4.1234,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925066326475461e-05,
+      "loss": 4.0777,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9248142360306296e-05,
+      "loss": 4.0284,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.924561728732118e-05,
+      "loss": 4.1312,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9243088046233386e-05,
+      "loss": 4.1282,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.92405546374777e-05,
+      "loss": 4.1657,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.923801706148966e-05,
+      "loss": 4.1477,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.923547531870548e-05,
+      "loss": 4.1242,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.923292940956215e-05,
+      "loss": 4.075,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9230379334497314e-05,
+      "loss": 4.2274,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922782509394936e-05,
+      "loss": 4.1323,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922526668835741e-05,
+      "loss": 4.1324,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922270411816127e-05,
+      "loss": 4.1306,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922013738380148e-05,
+      "loss": 4.1161,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.921756648571928e-05,
+      "loss": 4.0503,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.921499142435666e-05,
+      "loss": 4.0721,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.921241220015628e-05,
+      "loss": 4.0669,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9209828813561554e-05,
+      "loss": 4.0061,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920724126501659e-05,
+      "loss": 4.1541,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9204649554966216e-05,
+      "loss": 4.0748,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920205368385598e-05,
+      "loss": 3.9827,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9199453652132144e-05,
+      "loss": 4.0876,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.919684946024168e-05,
+      "loss": 4.248,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9194241108632277e-05,
+      "loss": 4.0109,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9191628597752345e-05,
+      "loss": 4.2597,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9189011928051e-05,
+      "loss": 4.1619,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918639109997809e-05,
+      "loss": 4.2123,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9183766113984155e-05,
+      "loss": 4.1452,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918113697052046e-05,
+      "loss": 4.122,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917850367003898e-05,
+      "loss": 4.0915,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917586621299243e-05,
+      "loss": 4.2213,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9173224599834206e-05,
+      "loss": 4.0787,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9170578831018434e-05,
+      "loss": 4.1241,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.916792890699994e-05,
+      "loss": 4.1118,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9165274828234295e-05,
+      "loss": 4.248,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.916261659517777e-05,
+      "loss": 4.1569,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9159954208287324e-05,
+      "loss": 3.9937,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915728766802066e-05,
+      "loss": 4.2111,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91546169748362e-05,
+      "loss": 4.1005,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915194212919305e-05,
+      "loss": 4.247,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914926313155106e-05,
+      "loss": 4.0321,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914657998237078e-05,
+      "loss": 4.0916,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914389268211346e-05,
+      "loss": 4.1515,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9141201231241085e-05,
+      "loss": 4.1944,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9138505630216356e-05,
+      "loss": 4.2414,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913580587950267e-05,
+      "loss": 4.0902,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9133101979564156e-05,
+      "loss": 4.0939,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913039393086563e-05,
+      "loss": 4.1302,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912768173387264e-05,
+      "loss": 3.9594,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912496538905145e-05,
+      "loss": 4.1878,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912224489686903e-05,
+      "loss": 3.9571,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911952025779306e-05,
+      "loss": 4.2673,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9116791472291943e-05,
+      "loss": 4.259,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911405854083479e-05,
+      "loss": 4.1172,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911132146389141e-05,
+      "loss": 4.0728,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910858024193236e-05,
+      "loss": 3.9453,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9105834875428866e-05,
+      "loss": 4.1767,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910308536485291e-05,
+      "loss": 3.9747,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9100331710677136e-05,
+      "loss": 4.1929,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.909757391337496e-05,
+      "loss": 4.1314,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.909481197342046e-05,
+      "loss": 4.0642,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9092045891288454e-05,
+      "loss": 4.1907,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908927566745446e-05,
+      "loss": 4.141,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908650130239471e-05,
+      "loss": 4.1238,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908372279658614e-05,
+      "loss": 4.0385,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9080940150506434e-05,
+      "loss": 4.021,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907815336463394e-05,
+      "loss": 4.0759,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9075362439447744e-05,
+      "loss": 4.2298,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907256737542764e-05,
+      "loss": 4.0183,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.906976817305413e-05,
+      "loss": 4.076,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9066964832808436e-05,
+      "loss": 4.0693,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9064157355172474e-05,
+      "loss": 4.1802,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9061345740628886e-05,
+      "loss": 3.9912,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.905852998966103e-05,
+      "loss": 4.0447,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.905571010275295e-05,
+      "loss": 4.0459,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9052886080389425e-05,
+      "loss": 4.1637,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9050057923055945e-05,
+      "loss": 4.1132,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904722563123869e-05,
+      "loss": 4.0846,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9044389205424576e-05,
+      "loss": 4.0735,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904154864610121e-05,
+      "loss": 4.052,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.903870395375691e-05,
+      "loss": 4.0428,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9035855128880724e-05,
+      "loss": 4.0771,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9033002171962395e-05,
+      "loss": 4.1504,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9030145083492376e-05,
+      "loss": 4.1837,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902728386396184e-05,
+      "loss": 4.163,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9024418513862655e-05,
+      "loss": 4.0956,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9021549033687414e-05,
+      "loss": 4.0906,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9018675423929416e-05,
+      "loss": 4.03,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901579768508266e-05,
+      "loss": 4.1886,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901291581764187e-05,
+      "loss": 4.1486,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901002982210246e-05,
+      "loss": 4.1345,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900713969896059e-05,
+      "loss": 4.1206,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900424544871308e-05,
+      "loss": 4.1559,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900134707185751e-05,
+      "loss": 4.1397,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899844456889211e-05,
+      "loss": 4.3261,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899553794031589e-05,
+      "loss": 4.1242,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8992627186628514e-05,
+      "loss": 4.271,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8989712308330376e-05,
+      "loss": 4.0689,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.898679330592258e-05,
+      "loss": 4.1348,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.898387017990694e-05,
+      "loss": 4.0242,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8980942930785965e-05,
+      "loss": 4.1783,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8978011559062895e-05,
+      "loss": 3.9852,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.897507606524167e-05,
+      "loss": 4.0912,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.897213644982691e-05,
+      "loss": 4.0139,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896919271332399e-05,
+      "loss": 3.9977,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896624485623896e-05,
+      "loss": 4.0334,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896329287907861e-05,
+      "loss": 4.0652,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896033678235041e-05,
+      "loss": 4.2077,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.895737656656254e-05,
+      "loss": 4.0802,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89544122322239e-05,
+      "loss": 4.1537,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8951443779844096e-05,
+      "loss": 4.0923,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894847120993343e-05,
+      "loss": 3.9454,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894549452300294e-05,
+      "loss": 4.0154,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894251371956433e-05,
+      "loss": 4.1409,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893952880013006e-05,
+      "loss": 4.0892,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893653976521325e-05,
+      "loss": 4.0768,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8933546615327756e-05,
+      "loss": 4.1166,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893054935098814e-05,
+      "loss": 4.2203,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892754797270966e-05,
+      "loss": 4.1225,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8924542481008303e-05,
+      "loss": 4.2079,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892153287640073e-05,
+      "loss": 4.1073,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891851915940433e-05,
+      "loss": 4.0412,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89155013305372e-05,
+      "loss": 4.1034,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891247939031815e-05,
+      "loss": 4.0123,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8909453339266655e-05,
+      "loss": 4.0339,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.890642317790296e-05,
+      "loss": 4.1359,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8903388906747974e-05,
+      "loss": 4.2815,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.890035052632332e-05,
+      "loss": 4.0929,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889730803715134e-05,
+      "loss": 4.2052,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889426143975506e-05,
+      "loss": 4.1422,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889121073465824e-05,
+      "loss": 4.0958,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888815592238532e-05,
+      "loss": 3.9774,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888509700346146e-05,
+      "loss": 4.0617,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888203397841253e-05,
+      "loss": 4.0697,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8878966847765086e-05,
+      "loss": 3.99,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887589561204642e-05,
+      "loss": 4.0487,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8872820271784506e-05,
+      "loss": 3.9241,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8869740827508025e-05,
+      "loss": 4.0043,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886665727974637e-05,
+      "loss": 4.0497,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886356962902966e-05,
+      "loss": 4.193,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8860477875888664e-05,
+      "loss": 4.1071,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8857382020854915e-05,
+      "loss": 3.9764,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885428206446061e-05,
+      "loss": 4.0804,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8851178007238687e-05,
+      "loss": 4.256,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8848069849722744e-05,
+      "loss": 4.1705,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884495759244713e-05,
+      "loss": 4.1775,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884184123594687e-05,
+      "loss": 4.128,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8838720780757707e-05,
+      "loss": 4.0583,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.883559622741608e-05,
+      "loss": 4.1512,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8832467576459135e-05,
+      "loss": 4.182,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.882933482842472e-05,
+      "loss": 4.0232,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8826197983851396e-05,
+      "loss": 3.9886,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882305704327843e-05,
+      "loss": 4.1768,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8819912007245767e-05,
+      "loss": 4.1639,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8816762876294086e-05,
+      "loss": 4.0604,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8813609650964763e-05,
+      "loss": 4.0705,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881045233179987e-05,
+      "loss": 3.9886,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880729091934219e-05,
+      "loss": 4.1731,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8804125414135195e-05,
+      "loss": 4.0753,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880095581672308e-05,
+      "loss": 4.0277,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879778212765075e-05,
+      "loss": 3.9934,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879460434746377e-05,
+      "loss": 4.0971,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879142247670846e-05,
+      "loss": 4.092,
+      "step": 988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878823651593181e-05,
+      "loss": 4.1269,
+      "step": 989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878504646568152e-05,
+      "loss": 4.0241,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8781852326506004e-05,
+      "loss": 4.0841,
+      "step": 991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.877865409895438e-05,
+      "loss": 4.2149,
+      "step": 992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8775451783576444e-05,
+      "loss": 4.1261,
+      "step": 993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8772245380922714e-05,
+      "loss": 4.0717,
+      "step": 994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876903489154442e-05,
+      "loss": 4.0944,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876582031599347e-05,
+      "loss": 4.0503,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8762601654822485e-05,
+      "loss": 4.0675,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8759378908584805e-05,
+      "loss": 3.9608,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875615207783444e-05,
+      "loss": 4.2557,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875292116312613e-05,
+      "loss": 4.1267,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8749686165015307e-05,
+      "loss": 3.9184,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8746447084058106e-05,
+      "loss": 4.0259,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874320392081135e-05,
+      "loss": 4.1482,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873995667583259e-05,
+      "loss": 4.0443,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873670534968006e-05,
+      "loss": 4.1278,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8733449942912704e-05,
+      "loss": 4.1286,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873019045609016e-05,
+      "loss": 4.1073,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.872692688977277e-05,
+      "loss": 4.0457,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.872365924452158e-05,
+      "loss": 4.0325,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.872038752089834e-05,
+      "loss": 3.9659,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8717111719465494e-05,
+      "loss": 4.0665,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8713831840786194e-05,
+      "loss": 4.059,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8710547885424284e-05,
+      "loss": 4.174,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870725985394432e-05,
+      "loss": 3.9159,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870396774691154e-05,
+      "loss": 4.2476,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870067156489191e-05,
+      "loss": 4.0844,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8697371308452075e-05,
+      "loss": 4.0433,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.869406697815939e-05,
+      "loss": 4.1878,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.869075857458191e-05,
+      "loss": 3.9087,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868744609828838e-05,
+      "loss": 4.009,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868412954984825e-05,
+      "loss": 4.0164,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868080892983169e-05,
+      "loss": 4.1793,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.867748423880954e-05,
+      "loss": 4.009,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.867415547735336e-05,
+      "loss": 3.9984,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.86708226460354e-05,
+      "loss": 3.9915,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.86674857454286e-05,
+      "loss": 4.0481,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.866414477610664e-05,
+      "loss": 4.074,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.866079973864384e-05,
+      "loss": 4.0303,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.865745063361527e-05,
+      "loss": 4.0546,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8654097461596673e-05,
+      "loss": 4.1398,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8650740223164505e-05,
+      "loss": 3.9433,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.864737891889591e-05,
+      "loss": 4.1847,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.864401354936873e-05,
+      "loss": 3.8992,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8640644115161526e-05,
+      "loss": 4.1399,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.863727061685353e-05,
+      "loss": 3.9768,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.863389305502468e-05,
+      "loss": 4.1617,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8630511430255635e-05,
+      "loss": 4.0577,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8627125743127716e-05,
+      "loss": 3.9987,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.862373599422298e-05,
+      "loss": 4.15,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8620342184124154e-05,
+      "loss": 4.2085,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.861694431341468e-05,
+      "loss": 4.0463,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8613542382678684e-05,
+      "loss": 4.1683,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8610136392501e-05,
+      "loss": 4.1377,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8606726343467165e-05,
+      "loss": 4.1401,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.860331223616339e-05,
+      "loss": 4.1081,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.859989407117661e-05,
+      "loss": 4.2102,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.859647184909445e-05,
+      "loss": 4.0457,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.859304557050522e-05,
+      "loss": 3.9643,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8589615235997956e-05,
+      "loss": 4.029,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8586180846162344e-05,
+      "loss": 4.1022,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8582742401588817e-05,
+      "loss": 4.0092,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8579299902868466e-05,
+      "loss": 4.1759,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.857585335059312e-05,
+      "loss": 3.8937,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.857240274535526e-05,
+      "loss": 4.0576,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.85689480877481e-05,
+      "loss": 4.0753,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.856548937836552e-05,
+      "loss": 4.0771,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.856202661780214e-05,
+      "loss": 4.0966,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8558559806653214e-05,
+      "loss": 4.1272,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.855508894551474e-05,
+      "loss": 4.0647,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.855161403498341e-05,
+      "loss": 4.1448,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8548135075656596e-05,
+      "loss": 4.1615,
+      "step": 1061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.854465206813236e-05,
+      "loss": 4.173,
+      "step": 1062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.854116501300949e-05,
+      "loss": 4.0451,
+      "step": 1063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8537673910887435e-05,
+      "loss": 4.1146,
+      "step": 1064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8534178762366366e-05,
+      "loss": 4.0938,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.853067956804713e-05,
+      "loss": 4.0255,
+      "step": 1066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8527176328531296e-05,
+      "loss": 4.0395,
+      "step": 1067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.852366904442109e-05,
+      "loss": 4.1455,
+      "step": 1068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.852015771631947e-05,
+      "loss": 3.9825,
+      "step": 1069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.851664234483008e-05,
+      "loss": 4.0978,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.851312293055723e-05,
+      "loss": 4.0221,
+      "step": 1071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.850959947410596e-05,
+      "loss": 4.1255,
+      "step": 1072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8506071976081995e-05,
+      "loss": 4.0463,
+      "step": 1073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.850254043709176e-05,
+      "loss": 4.0898,
+      "step": 1074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8499004857742346e-05,
+      "loss": 4.0285,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.849546523864158e-05,
+      "loss": 4.0924,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.849192158039796e-05,
+      "loss": 4.1449,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.848837388362067e-05,
+      "loss": 4.0529,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8484822148919614e-05,
+      "loss": 4.0908,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.848126637690537e-05,
+      "loss": 4.1221,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.847770656818922e-05,
+      "loss": 4.0801,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.847414272338313e-05,
+      "loss": 4.1797,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.847057484309977e-05,
+      "loss": 3.9983,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.846700292795251e-05,
+      "loss": 3.9292,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8463426978555384e-05,
+      "loss": 4.0237,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.845984699552315e-05,
+      "loss": 4.1477,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.845626297947125e-05,
+      "loss": 4.0147,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.845267493101582e-05,
+      "loss": 4.0377,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8449082850773676e-05,
+      "loss": 3.9273,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8445486739362345e-05,
+      "loss": 3.9379,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.844188659740004e-05,
+      "loss": 3.9778,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8438282425505664e-05,
+      "loss": 4.0232,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8434674224298825e-05,
+      "loss": 3.9904,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.843106199439981e-05,
+      "loss": 4.2537,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.84274457364296e-05,
+      "loss": 4.1274,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.842382545100987e-05,
+      "loss": 4.0304,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8420201138763e-05,
+      "loss": 4.1596,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.841657280031204e-05,
+      "loss": 3.9021,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8412940436280747e-05,
+      "loss": 4.0505,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.840930404729357e-05,
+      "loss": 4.0402,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.840566363397564e-05,
+      "loss": 4.097,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8402019196952786e-05,
+      "loss": 3.9973,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.839837073685154e-05,
+      "loss": 4.1045,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.83947182542991e-05,
+      "loss": 4.0319,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.839106174992338e-05,
+      "loss": 3.8926,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.838740122435298e-05,
+      "loss": 3.9728,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8383736678217174e-05,
+      "loss": 4.0306,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8380068112145936e-05,
+      "loss": 4.1003,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.837639552676996e-05,
+      "loss": 4.0983,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.837271892272058e-05,
+      "loss": 4.1468,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.836903830062986e-05,
+      "loss": 4.0801,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.836535366113054e-05,
+      "loss": 3.917,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.836166500485605e-05,
+      "loss": 4.027,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.835797233244052e-05,
+      "loss": 4.0595,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.835427564451875e-05,
+      "loss": 4.0668,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.835057494172626e-05,
+      "loss": 3.9321,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.834687022469923e-05,
+      "loss": 4.0001,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.834316149407454e-05,
+      "loss": 3.9096,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8339448750489777e-05,
+      "loss": 4.0542,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.83357319945832e-05,
+      "loss": 4.1732,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8332011226993765e-05,
+      "loss": 4.1136,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8328286448361106e-05,
+      "loss": 4.0715,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.832455765932557e-05,
+      "loss": 4.1302,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8320824860528166e-05,
+      "loss": 4.0662,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.831708805261062e-05,
+      "loss": 4.1053,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.831334723621531e-05,
+      "loss": 4.0245,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.830960241198534e-05,
+      "loss": 4.0148,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8305853580564486e-05,
+      "loss": 4.1434,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.830210074259722e-05,
+      "loss": 4.0176,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.82983438987287e-05,
+      "loss": 4.1127,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.829458304960476e-05,
+      "loss": 4.1091,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.829081819587194e-05,
+      "loss": 4.025,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8287049338177465e-05,
+      "loss": 4.0188,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.828327647716924e-05,
+      "loss": 4.0746,
+      "step": 1134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.827949961349587e-05,
+      "loss": 4.1975,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8275718747806634e-05,
+      "loss": 4.0472,
+      "step": 1136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.827193388075151e-05,
+      "loss": 4.0095,
+      "step": 1137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8268145012981155e-05,
+      "loss": 4.0434,
+      "step": 1138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.826435214514693e-05,
+      "loss": 3.9671,
+      "step": 1139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.826055527790087e-05,
+      "loss": 4.0737,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.82567544118957e-05,
+      "loss": 3.9762,
+      "step": 1141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.825294954778482e-05,
+      "loss": 4.2451,
+      "step": 1142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.824914068622235e-05,
+      "loss": 3.9805,
+      "step": 1143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.824532782786307e-05,
+      "loss": 4.1891,
+      "step": 1144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8241510973362446e-05,
+      "loss": 4.1127,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.823769012337666e-05,
+      "loss": 4.1046,
+      "step": 1146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8233865278562536e-05,
+      "loss": 3.9583,
+      "step": 1147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.823003643957762e-05,
+      "loss": 4.048,
+      "step": 1148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8226203607080145e-05,
+      "loss": 4.1471,
+      "step": 1149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8222366781729e-05,
+      "loss": 4.2513,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.821852596418379e-05,
+      "loss": 4.0903,
+      "step": 1151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.821468115510479e-05,
+      "loss": 4.1935,
+      "step": 1152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8210832355152977e-05,
+      "loss": 4.0171,
+      "step": 1153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.820697956499e-05,
+      "loss": 4.1151,
+      "step": 1154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.82031227852782e-05,
+      "loss": 4.0403,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.819926201668059e-05,
+      "loss": 4.085,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.81953972598609e-05,
+      "loss": 4.0434,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.819152851548351e-05,
+      "loss": 4.135,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.818765578421352e-05,
+      "loss": 4.0254,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.818377906671667e-05,
+      "loss": 4.0996,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.817989836365944e-05,
+      "loss": 4.072,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8176013675708956e-05,
+      "loss": 4.0051,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8172125003533036e-05,
+      "loss": 4.0849,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.81682323478002e-05,
+      "loss": 3.9345,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8164335709179634e-05,
+      "loss": 3.9998,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.81604350883412e-05,
+      "loss": 3.9974,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8156530485955495e-05,
+      "loss": 3.9742,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.815262190269374e-05,
+      "loss": 3.9967,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.814870933922787e-05,
+      "loss": 4.0597,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8144792796230505e-05,
+      "loss": 3.9901,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.814087227437494e-05,
+      "loss": 4.0201,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.813694777433516e-05,
+      "loss": 4.0457,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.813301929678582e-05,
+      "loss": 4.2631,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8129086842402294e-05,
+      "loss": 4.0268,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8125150411860604e-05,
+      "loss": 4.0646,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.812121000583746e-05,
+      "loss": 4.0948,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.811726562501028e-05,
+      "loss": 3.9047,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.811331727005714e-05,
+      "loss": 4.1844,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.810936494165681e-05,
+      "loss": 4.1508,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.810540864048874e-05,
+      "loss": 3.9145,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.810144836723306e-05,
+      "loss": 3.97,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.809748412257059e-05,
+      "loss": 3.952,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.809351590718283e-05,
+      "loss": 4.1461,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.808954372175196e-05,
+      "loss": 4.0001,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8085567566960855e-05,
+      "loss": 4.1903,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.808158744349305e-05,
+      "loss": 4.07,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.807760335203277e-05,
+      "loss": 3.9429,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.807361529326495e-05,
+      "loss": 4.023,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.806962326787517e-05,
+      "loss": 4.0205,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8065627276549694e-05,
+      "loss": 3.9289,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.806162731997549e-05,
+      "loss": 4.0192,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.80576233988402e-05,
+      "loss": 4.1088,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.805361551383214e-05,
+      "loss": 3.9649,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8049603665640316e-05,
+      "loss": 3.9297,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8045587854954403e-05,
+      "loss": 4.0632,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.804156808246477e-05,
+      "loss": 3.9831,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8037544348862475e-05,
+      "loss": 3.8748,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.803351665483922e-05,
+      "loss": 4.0411,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8029485001087425e-05,
+      "loss": 3.9022,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.802544938830018e-05,
+      "loss": 4.1237,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8021409817171247e-05,
+      "loss": 4.1939,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.801736628839509e-05,
+      "loss": 4.0812,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.801331880266682e-05,
+      "loss": 3.985,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.800926736068225e-05,
+      "loss": 4.1056,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8005211963137884e-05,
+      "loss": 3.8845,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.800115261073088e-05,
+      "loss": 4.0056,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.799708930415909e-05,
+      "loss": 4.0684,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.799302204412104e-05,
+      "loss": 4.0636,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.798895083131595e-05,
+      "loss": 4.0176,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7984875666443694e-05,
+      "loss": 4.1736,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.798079655020484e-05,
+      "loss": 3.9407,
+      "step": 1211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.797671348330066e-05,
+      "loss": 4.1174,
+      "step": 1212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.797262646643305e-05,
+      "loss": 4.1744,
+      "step": 1213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7968535500304634e-05,
+      "loss": 3.9782,
+      "step": 1214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.79644405856187e-05,
+      "loss": 4.0275,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.796034172307919e-05,
+      "loss": 4.0006,
+      "step": 1216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.795623891339076e-05,
+      "loss": 4.0236,
+      "step": 1217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.795213215725873e-05,
+      "loss": 4.0362,
+      "step": 1218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7948021455389103e-05,
+      "loss": 4.0044,
+      "step": 1219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.794390680848855e-05,
+      "loss": 3.9515,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7939788217264424e-05,
+      "loss": 4.1241,
+      "step": 1221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7935665682424766e-05,
+      "loss": 4.063,
+      "step": 1222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.793153920467829e-05,
+      "loss": 3.9998,
+      "step": 1223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7927408784734365e-05,
+      "loss": 4.1704,
+      "step": 1224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.792327442330309e-05,
+      "loss": 3.8759,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7919136121095176e-05,
+      "loss": 3.9604,
+      "step": 1226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7914993878822065e-05,
+      "loss": 4.1501,
+      "step": 1227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.791084769719586e-05,
+      "loss": 3.9086,
+      "step": 1228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7906697576929324e-05,
+      "loss": 4.0739,
+      "step": 1229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.790254351873592e-05,
+      "loss": 4.0799,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789838552332978e-05,
+      "loss": 4.0938,
+      "step": 1231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.78942235914257e-05,
+      "loss": 3.9572,
+      "step": 1232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789005772373918e-05,
+      "loss": 4.0316,
+      "step": 1233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.788588792098637e-05,
+      "loss": 4.0154,
+      "step": 1234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.788171418388411e-05,
+      "loss": 4.1961,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.787753651314991e-05,
+      "loss": 4.0985,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.787335490950197e-05,
+      "loss": 3.9143,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7869169373659146e-05,
+      "loss": 3.901,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.786497990634099e-05,
+      "loss": 4.0508,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7860786508267706e-05,
+      "loss": 4.1498,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.78565891801602e-05,
+      "loss": 3.9707,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.785238792274004e-05,
+      "loss": 3.9363,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.784818273672947e-05,
+      "loss": 4.0024,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7843973622851404e-05,
+      "loss": 4.028,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7839760581829445e-05,
+      "loss": 4.0089,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.783554361438786e-05,
+      "loss": 4.175,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7831322721251595e-05,
+      "loss": 4.0708,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7827097903146264e-05,
+      "loss": 4.027,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.782286916079818e-05,
+      "loss": 4.0708,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.78186364949343e-05,
+      "loss": 4.0963,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.781439990628227e-05,
+      "loss": 4.0871,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.781015939557041e-05,
+      "loss": 4.0646,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.780591496352771e-05,
+      "loss": 3.9342,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7801666610883835e-05,
+      "loss": 3.7833,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7797414338369126e-05,
+      "loss": 4.094,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.779315814671461e-05,
+      "loss": 4.0519,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.778889803665197e-05,
+      "loss": 4.124,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.778463400891357e-05,
+      "loss": 4.1393,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.778036606423243e-05,
+      "loss": 3.9425,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.777609420334227e-05,
+      "loss": 4.0019,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7771818426977475e-05,
+      "loss": 4.0004,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.776753873587311e-05,
+      "loss": 4.0753,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7763255130764886e-05,
+      "loss": 4.0872,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7758967612389206e-05,
+      "loss": 4.0457,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.775467618148315e-05,
+      "loss": 4.2278,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.775038083878447e-05,
+      "loss": 3.9987,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7746081585031564e-05,
+      "loss": 4.1098,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7741778420963554e-05,
+      "loss": 4.0789,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7737471347320184e-05,
+      "loss": 4.268,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.773316036484189e-05,
+      "loss": 4.1135,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.772884547426979e-05,
+      "loss": 4.1592,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.772452667634565e-05,
+      "loss": 4.013,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.772020397181194e-05,
+      "loss": 4.1412,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.771587736141176e-05,
+      "loss": 4.1883,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.771154684588892e-05,
+      "loss": 4.1356,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.770721242598789e-05,
+      "loss": 4.0821,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.770287410245379e-05,
+      "loss": 3.9996,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.769853187603244e-05,
+      "loss": 4.1068,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7694185747470323e-05,
+      "loss": 4.0772,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.768983571751458e-05,
+      "loss": 4.0451,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.768548178691302e-05,
+      "loss": 4.1594,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7681123956414164e-05,
+      "loss": 4.1137,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.767676222676715e-05,
+      "loss": 4.2058,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.767239659872183e-05,
+      "loss": 4.1634,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.766802707302868e-05,
+      "loss": 4.1254,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.766365365043889e-05,
+      "loss": 4.078,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7659276331704296e-05,
+      "loss": 3.898,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.765489511757741e-05,
+      "loss": 4.0729,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.765051000881142e-05,
+      "loss": 4.188,
+      "step": 1289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7646121006160174e-05,
+      "loss": 4.1592,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.764172811037818e-05,
+      "loss": 4.108,
+      "step": 1291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7637331322220654e-05,
+      "loss": 3.9873,
+      "step": 1292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.763293064244343e-05,
+      "loss": 4.1707,
+      "step": 1293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.762852607180305e-05,
+      "loss": 4.1948,
+      "step": 1294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7624117611056694e-05,
+      "loss": 4.0804,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7619705260962246e-05,
+      "loss": 4.1807,
+      "step": 1296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.761528902227824e-05,
+      "loss": 4.0972,
+      "step": 1297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7610868895763865e-05,
+      "loss": 4.0126,
+      "step": 1298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7606444882179e-05,
+      "loss": 4.1935,
+      "step": 1299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.760201698228419e-05,
+      "loss": 3.9808,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.759758519684063e-05,
+      "loss": 4.1673,
+      "step": 1301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7593149526610206e-05,
+      "loss": 4.0998,
+      "step": 1302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.758870997235545e-05,
+      "loss": 4.1624,
+      "step": 1303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.758426653483958e-05,
+      "loss": 4.1307,
+      "step": 1304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.757981921482647e-05,
+      "loss": 3.9985,
+      "step": 1305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.757536801308067e-05,
+      "loss": 3.9653,
+      "step": 1306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.75709129303674e-05,
+      "loss": 4.0669,
+      "step": 1307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.756645396745253e-05,
+      "loss": 3.9985,
+      "step": 1308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.756199112510261e-05,
+      "loss": 3.9888,
+      "step": 1309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.755752440408485e-05,
+      "loss": 4.0017,
+      "step": 1310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.755305380516713e-05,
+      "loss": 4.0017,
+      "step": 1311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.754857932911802e-05,
+      "loss": 4.0483,
+      "step": 1312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.754410097670671e-05,
+      "loss": 4.0578,
+      "step": 1313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.753961874870309e-05,
+      "loss": 3.9595,
+      "step": 1314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.75351326458777e-05,
+      "loss": 3.9612,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.753064266900177e-05,
+      "loss": 3.9424,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7526148818847164e-05,
+      "loss": 3.9841,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.752165109618642e-05,
+      "loss": 4.1655,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.751714950179277e-05,
+      "loss": 4.0922,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7512644036440066e-05,
+      "loss": 4.0278,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.750813470090287e-05,
+      "loss": 4.1075,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.750362149595639e-05,
+      "loss": 3.9163,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.749910442237648e-05,
+      "loss": 3.9145,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.74945834809397e-05,
+      "loss": 4.1307,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.749005867242323e-05,
+      "loss": 4.0874,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.748552999760495e-05,
+      "loss": 4.0252,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.748099745726339e-05,
+      "loss": 4.0603,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.747646105217774e-05,
+      "loss": 4.0421,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.747192078312788e-05,
+      "loss": 4.1571,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.74673766508943e-05,
+      "loss": 4.172,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.746282865625822e-05,
+      "loss": 4.2277,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.745827680000149e-05,
+      "loss": 4.1589,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.745372108290661e-05,
+      "loss": 4.1409,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7449161505756766e-05,
+      "loss": 4.0248,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.744459806933581e-05,
+      "loss": 3.9145,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.744003077442825e-05,
+      "loss": 4.0391,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.743545962181926e-05,
+      "loss": 4.1184,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.743088461229465e-05,
+      "loss": 4.0451,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.742630574664094e-05,
+      "loss": 3.9427,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.742172302564528e-05,
+      "loss": 4.0403,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.741713645009551e-05,
+      "loss": 4.1325,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.741254602078009e-05,
+      "loss": 4.0115,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7407951738488186e-05,
+      "loss": 4.0774,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7403353604009605e-05,
+      "loss": 3.9336,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.739875161813481e-05,
+      "loss": 4.1267,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.739414578165495e-05,
+      "loss": 4.1113,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.738953609536181e-05,
+      "loss": 4.2074,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.738492256004786e-05,
+      "loss": 4.06,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.738030517650621e-05,
+      "loss": 3.9866,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737568394553064e-05,
+      "loss": 4.1827,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7371058867915606e-05,
+      "loss": 3.9805,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7366429944456204e-05,
+      "loss": 4.2215,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7361797175948194e-05,
+      "loss": 4.0552,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735716056318802e-05,
+      "loss": 3.9833,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735252010697275e-05,
+      "loss": 4.1341,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.734787580810015e-05,
+      "loss": 4.1506,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7343227667368604e-05,
+      "loss": 4.1443,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733857568557721e-05,
+      "loss": 4.2332,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7333919863525685e-05,
+      "loss": 4.0756,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7329260202014415e-05,
+      "loss": 4.1884,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732459670184445e-05,
+      "loss": 4.1602,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7319929363817503e-05,
+      "loss": 4.1283,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731525818873596e-05,
+      "loss": 4.1386,
+      "step": 1363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731058317740281e-05,
+      "loss": 4.0244,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730590433062178e-05,
+      "loss": 3.9934,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7301221649197206e-05,
+      "loss": 4.233,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729653513393408e-05,
+      "loss": 4.0487,
+      "step": 1367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729184478563808e-05,
+      "loss": 4.3329,
+      "step": 1368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.728715060511555e-05,
+      "loss": 4.0642,
+      "step": 1369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.728245259317344e-05,
+      "loss": 4.0395,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7277750750619415e-05,
+      "loss": 3.9785,
+      "step": 1371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.727304507826177e-05,
+      "loss": 4.1384,
+      "step": 1372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.726833557690946e-05,
+      "loss": 4.1361,
+      "step": 1373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7263622247372116e-05,
+      "loss": 4.1892,
+      "step": 1374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.725890509046e-05,
+      "loss": 4.1114,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.725418410698405e-05,
+      "loss": 4.091,
+      "step": 1376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.724945929775586e-05,
+      "loss": 4.1434,
+      "step": 1377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.724473066358768e-05,
+      "loss": 4.1029,
+      "step": 1378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7239998205292424e-05,
+      "loss": 4.0656,
+      "step": 1379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.723526192368364e-05,
+      "loss": 4.0707,
+      "step": 1380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.723052181957556e-05,
+      "loss": 4.1477,
+      "step": 1381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7225777893783055e-05,
+      "loss": 4.1218,
+      "step": 1382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.722103014712167e-05,
+      "loss": 4.2013,
+      "step": 1383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.72162785804076e-05,
+      "loss": 4.0602,
+      "step": 1384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.721152319445769e-05,
+      "loss": 4.1538,
+      "step": 1385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720676399008943e-05,
+      "loss": 4.1008,
+      "step": 1386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720200096812101e-05,
+      "loss": 4.0971,
+      "step": 1387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7197234129371234e-05,
+      "loss": 4.0409,
+      "step": 1388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7192463474659576e-05,
+      "loss": 4.1783,
+      "step": 1389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718768900480617e-05,
+      "loss": 4.1469,
+      "step": 1390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7182910720631805e-05,
+      "loss": 4.1618,
+      "step": 1391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717812862295792e-05,
+      "loss": 4.0541,
+      "step": 1392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7173342712606616e-05,
+      "loss": 4.2008,
+      "step": 1393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7168552990400636e-05,
+      "loss": 4.1765,
+      "step": 1394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7163759457163406e-05,
+      "loss": 4.1098,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7158962113718974e-05,
+      "loss": 4.0349,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715416096089207e-05,
+      "loss": 4.0103,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714935599950806e-05,
+      "loss": 4.2225,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7144547230392986e-05,
+      "loss": 4.1706,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713973465437351e-05,
+      "loss": 4.0819,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713491827227699e-05,
+      "loss": 4.0742,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.713009808493141e-05,
+      "loss": 4.204,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712527409316541e-05,
+      "loss": 4.2561,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712044629780829e-05,
+      "loss": 4.1643,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711561469969001e-05,
+      "loss": 4.0572,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7110779299641184e-05,
+      "loss": 4.161,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710594009849306e-05,
+      "loss": 4.1215,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7101097097077566e-05,
+      "loss": 4.1664,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7096250296227264e-05,
+      "loss": 4.0005,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7091399696775366e-05,
+      "loss": 4.0548,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7086545299555764e-05,
+      "loss": 4.2081,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708168710540297e-05,
+      "loss": 4.1,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707682511515217e-05,
+      "loss": 4.0876,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7071959329639196e-05,
+      "loss": 3.9732,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706708974970053e-05,
+      "loss": 4.006,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706221637617332e-05,
+      "loss": 4.1814,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705733920989534e-05,
+      "loss": 4.0528,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7052458251705045e-05,
+      "loss": 4.1468,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704757350244152e-05,
+      "loss": 4.1594,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704268496294451e-05,
+      "loss": 4.1621,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703779263405442e-05,
+      "loss": 4.1157,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703289651661229e-05,
+      "loss": 4.1452,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702799661145981e-05,
+      "loss": 4.1602,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702309291943936e-05,
+      "loss": 4.2888,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701818544139391e-05,
+      "loss": 4.0846,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701327417816714e-05,
+      "loss": 4.0975,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700835913060333e-05,
+      "loss": 4.1991,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700344029954744e-05,
+      "loss": 4.2078,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6998517685845085e-05,
+      "loss": 4.0256,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699359129034251e-05,
+      "loss": 4.0318,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6988661113886615e-05,
+      "loss": 4.0789,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698372715732496e-05,
+      "loss": 4.0656,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6978789421505754e-05,
+      "loss": 3.9765,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697384790727784e-05,
+      "loss": 4.0346,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6968902615490734e-05,
+      "loss": 4.1072,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696395354699458e-05,
+      "loss": 3.9705,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6959000702640185e-05,
+      "loss": 4.072,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6954044083279004e-05,
+      "loss": 3.9793,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694908368976312e-05,
+      "loss": 4.1039,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694411952294531e-05,
+      "loss": 4.0229,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6939151583678944e-05,
+      "loss": 3.9395,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.693417987281807e-05,
+      "loss": 4.0769,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6929204391217415e-05,
+      "loss": 4.0788,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.692422513973229e-05,
+      "loss": 4.0913,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69192421192187e-05,
+      "loss": 3.88,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691425533053327e-05,
+      "loss": 4.2431,
+      "step": 1446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690926477453331e-05,
+      "loss": 3.9683,
+      "step": 1447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690427045207673e-05,
+      "loss": 4.0424,
+      "step": 1448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689927236402213e-05,
+      "loss": 3.9912,
+      "step": 1449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6894270511228735e-05,
+      "loss": 4.1657,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688926489455641e-05,
+      "loss": 4.0662,
+      "step": 1451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6884255514865695e-05,
+      "loss": 3.982,
+      "step": 1452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687924237301774e-05,
+      "loss": 4.2006,
+      "step": 1453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687422546987439e-05,
+      "loss": 4.2075,
+      "step": 1454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6869204806298095e-05,
+      "loss": 4.0238,
+      "step": 1455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6864180383151955e-05,
+      "loss": 4.1238,
+      "step": 1456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685915220129974e-05,
+      "loss": 4.1263,
+      "step": 1457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6854120261605845e-05,
+      "loss": 4.1117,
+      "step": 1458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684908456493533e-05,
+      "loss": 4.146,
+      "step": 1459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684404511215387e-05,
+      "loss": 4.0825,
+      "step": 1460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683900190412782e-05,
+      "loss": 4.0694,
+      "step": 1461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683395494172417e-05,
+      "loss": 4.1039,
+      "step": 1462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6828904225810546e-05,
+      "loss": 4.1458,
+      "step": 1463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6823849757255225e-05,
+      "loss": 4.0182,
+      "step": 1464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681879153692711e-05,
+      "loss": 4.0921,
+      "step": 1465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6813729565695793e-05,
+      "loss": 4.2416,
+      "step": 1466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680866384443149e-05,
+      "loss": 4.1776,
+      "step": 1467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680359437400503e-05,
+      "loss": 3.9611,
+      "step": 1468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679852115528793e-05,
+      "loss": 4.0491,
+      "step": 1469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679344418915234e-05,
+      "loss": 4.1304,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678836347647104e-05,
+      "loss": 4.1664,
+      "step": 1471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6783279018117465e-05,
+      "loss": 4.1671,
+      "step": 1472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6778190814965694e-05,
+      "loss": 4.0608,
+      "step": 1473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.677309886789044e-05,
+      "loss": 4.1944,
+      "step": 1474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.676800317776708e-05,
+      "loss": 4.101,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.676290374547162e-05,
+      "loss": 4.0039,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.675780057188071e-05,
+      "loss": 4.082,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.675269365787164e-05,
+      "loss": 4.1711,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.674758300432236e-05,
+      "loss": 4.1924,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.674246861211144e-05,
+      "loss": 4.1617,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.673735048211809e-05,
+      "loss": 4.1601,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6732228615222204e-05,
+      "loss": 4.164,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6727103012304275e-05,
+      "loss": 4.0148,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.672197367424545e-05,
+      "loss": 4.1076,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6716840601927534e-05,
+      "loss": 4.2015,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6711703796232956e-05,
+      "loss": 4.1018,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.670656325804479e-05,
+      "loss": 3.979,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.670141898824676e-05,
+      "loss": 4.0323,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.669627098772321e-05,
+      "loss": 4.118,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.669111925735916e-05,
+      "loss": 4.0458,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.668596379804025e-05,
+      "loss": 3.9391,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.668080461065275e-05,
+      "loss": 4.0875,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6675641696083595e-05,
+      "loss": 4.1382,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6670475055220346e-05,
+      "loss": 4.029,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.666530468895121e-05,
+      "loss": 4.1011,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.666013059816503e-05,
+      "loss": 4.093,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.66549527837513e-05,
+      "loss": 4.0204,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.664977124660014e-05,
+      "loss": 3.9531,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.664458598760231e-05,
+      "loss": 3.988,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.663939700764923e-05,
+      "loss": 4.076,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.663420430763293e-05,
+      "loss": 3.9654,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.662900788844611e-05,
+      "loss": 4.0277,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.662380775098209e-05,
+      "loss": 4.0615,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.661860389613484e-05,
+      "loss": 4.2347,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.661339632479895e-05,
+      "loss": 3.9596,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6608185037869655e-05,
+      "loss": 4.0182,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6602970036242865e-05,
+      "loss": 4.22,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.659775132081507e-05,
+      "loss": 4.0399,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6592528892483456e-05,
+      "loss": 4.0533,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6587302752145795e-05,
+      "loss": 4.0506,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.658207290070053e-05,
+      "loss": 4.0279,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.657683933904673e-05,
+      "loss": 4.1642,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.657160206808411e-05,
+      "loss": 3.9905,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6566361088713015e-05,
+      "loss": 4.0986,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.656111640183442e-05,
+      "loss": 3.9812,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.655586800834998e-05,
+      "loss": 3.973,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.655061590916191e-05,
+      "loss": 4.0113,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.654536010517314e-05,
+      "loss": 4.1104,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.654010059728719e-05,
+      "loss": 4.1229,
+      "step": 1519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.653483738640824e-05,
+      "loss": 3.9238,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6529570473441084e-05,
+      "loss": 4.1604,
+      "step": 1521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.652429985929116e-05,
+      "loss": 4.1172,
+      "step": 1522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6519025544864583e-05,
+      "loss": 4.044,
+      "step": 1523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.651374753106803e-05,
+      "loss": 4.1416,
+      "step": 1524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.650846581880887e-05,
+      "loss": 4.0507,
+      "step": 1525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.650318040899509e-05,
+      "loss": 4.1027,
+      "step": 1526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.64978913025353e-05,
+      "loss": 4.1506,
+      "step": 1527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.649259850033879e-05,
+      "loss": 4.0529,
+      "step": 1528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.648730200331542e-05,
+      "loss": 3.9824,
+      "step": 1529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.648200181237574e-05,
+      "loss": 3.9314,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.64766979284309e-05,
+      "loss": 4.1088,
+      "step": 1531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.647139035239272e-05,
+      "loss": 4.1316,
+      "step": 1532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.646607908517361e-05,
+      "loss": 4.0843,
+      "step": 1533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.646076412768665e-05,
+      "loss": 4.2609,
+      "step": 1534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6455445480845546e-05,
+      "loss": 4.1475,
+      "step": 1535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6450123145564625e-05,
+      "loss": 4.1474,
+      "step": 1536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.644479712275887e-05,
+      "loss": 4.0762,
+      "step": 1537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.643946741334387e-05,
+      "loss": 4.0216,
+      "step": 1538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6434134018235884e-05,
+      "loss": 4.2339,
+      "step": 1539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.642879693835178e-05,
+      "loss": 4.0802,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6423456174609045e-05,
+      "loss": 4.1569,
+      "step": 1541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6418111727925836e-05,
+      "loss": 4.0052,
+      "step": 1542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6412763599220925e-05,
+      "loss": 4.2713,
+      "step": 1543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.640741178941371e-05,
+      "loss": 4.1469,
+      "step": 1544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6402056299424236e-05,
+      "loss": 4.0176,
+      "step": 1545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.639669713017317e-05,
+      "loss": 4.1005,
+      "step": 1546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.639133428258181e-05,
+      "loss": 4.1991,
+      "step": 1547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.63859677575721e-05,
+      "loss": 4.026,
+      "step": 1548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6380597556066606e-05,
+      "loss": 4.0512,
+      "step": 1549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.637522367898852e-05,
+      "loss": 4.1964,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6369846127261696e-05,
+      "loss": 4.0825,
+      "step": 1551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.636446490181057e-05,
+      "loss": 4.1068,
+      "step": 1552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.635908000356025e-05,
+      "loss": 4.191,
+      "step": 1553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.635369143343646e-05,
+      "loss": 4.1064,
+      "step": 1554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.634829919236557e-05,
+      "loss": 4.1517,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.634290328127455e-05,
+      "loss": 4.0486,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.633750370109103e-05,
+      "loss": 4.1388,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6332100452743254e-05,
+      "loss": 4.181,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.632669353716012e-05,
+      "loss": 3.9751,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.632128295527111e-05,
+      "loss": 4.06,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6315868708006395e-05,
+      "loss": 3.9339,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6310450796296724e-05,
+      "loss": 4.0366,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6305029221073516e-05,
+      "loss": 4.0394,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.629960398326879e-05,
+      "loss": 4.1682,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.629417508381522e-05,
+      "loss": 4.1011,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6288742523646086e-05,
+      "loss": 4.0724,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6283306303695316e-05,
+      "loss": 4.1323,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6277866424897456e-05,
+      "loss": 4.0457,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.627242288818767e-05,
+      "loss": 4.0061,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.626697569450179e-05,
+      "loss": 4.0059,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.626152484477624e-05,
+      "loss": 3.9361,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6256070339948085e-05,
+      "loss": 4.2066,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.625061218095501e-05,
+      "loss": 4.049,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.624515036873535e-05,
+      "loss": 4.1609,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.623968490422804e-05,
+      "loss": 4.0815,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.623421578837267e-05,
+      "loss": 3.9758,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6228743022109435e-05,
+      "loss": 4.1253,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6223266606379166e-05,
+      "loss": 4.1665,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.621778654212333e-05,
+      "loss": 4.2103,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.621230283028401e-05,
+      "loss": 4.1224,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.620681547180392e-05,
+      "loss": 4.1161,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6201324467626405e-05,
+      "loss": 4.1774,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.619582981869542e-05,
+      "loss": 4.1665,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.619033152595556e-05,
+      "loss": 4.1107,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6184829590352065e-05,
+      "loss": 4.1465,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.617932401283076e-05,
+      "loss": 4.0342,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.617381479433813e-05,
+      "loss": 4.0188,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.616830193582127e-05,
+      "loss": 4.072,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6162785438227895e-05,
+      "loss": 4.0001,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.615726530250637e-05,
+      "loss": 4.1073,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6151741529605656e-05,
+      "loss": 4.2312,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.614621412047537e-05,
+      "loss": 4.2016,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6140683076065724e-05,
+      "loss": 4.0194,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.613514839732757e-05,
+      "loss": 4.1641,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6129610085212396e-05,
+      "loss": 4.0328,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6124068140672284e-05,
+      "loss": 4.0104,
+      "step": 1596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6118522564659975e-05,
+      "loss": 4.1181,
+      "step": 1597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.61129733581288e-05,
+      "loss": 3.9928,
+      "step": 1598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6107420522032755e-05,
+      "loss": 4.1439,
+      "step": 1599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.610186405732641e-05,
+      "loss": 4.0312,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6096303964965e-05,
+      "loss": 3.9958,
+      "step": 1601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.609074024590438e-05,
+      "loss": 3.9792,
+      "step": 1602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.608517290110101e-05,
+      "loss": 3.9987,
+      "step": 1603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.607960193151197e-05,
+      "loss": 4.0431,
+      "step": 1604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.607402733809498e-05,
+      "loss": 3.9871,
+      "step": 1605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.606844912180839e-05,
+      "loss": 4.1946,
+      "step": 1606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.606286728361113e-05,
+      "loss": 4.1454,
+      "step": 1607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6057281824462824e-05,
+      "loss": 3.9258,
+      "step": 1608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6051692745323646e-05,
+      "loss": 3.9606,
+      "step": 1609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6046100047154425e-05,
+      "loss": 4.1145,
+      "step": 1610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.604050373091663e-05,
+      "loss": 3.9436,
+      "step": 1611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.603490379757232e-05,
+      "loss": 4.0716,
+      "step": 1612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.602930024808418e-05,
+      "loss": 3.9762,
+      "step": 1613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.602369308341555e-05,
+      "loss": 4.0302,
+      "step": 1614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6018082304530344e-05,
+      "loss": 4.0459,
+      "step": 1615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.601246791239312e-05,
+      "loss": 3.9801,
+      "step": 1616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6006849907969076e-05,
+      "loss": 4.0329,
+      "step": 1617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6001228292223994e-05,
+      "loss": 4.0549,
+      "step": 1618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.59956030661243e-05,
+      "loss": 3.8949,
+      "step": 1619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.598997423063704e-05,
+      "loss": 4.0448,
+      "step": 1620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.598434178672988e-05,
+      "loss": 4.306,
+      "step": 1621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.597870573537109e-05,
+      "loss": 4.1613,
+      "step": 1622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.597306607752958e-05,
+      "loss": 4.0405,
+      "step": 1623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.596742281417486e-05,
+      "loss": 4.085,
+      "step": 1624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.596177594627709e-05,
+      "loss": 4.0838,
+      "step": 1625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.595612547480703e-05,
+      "loss": 3.9452,
+      "step": 1626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.595047140073605e-05,
+      "loss": 4.0773,
+      "step": 1627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5944813725036164e-05,
+      "loss": 4.1476,
+      "step": 1628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.593915244867998e-05,
+      "loss": 4.0069,
+      "step": 1629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5933487572640745e-05,
+      "loss": 4.1492,
+      "step": 1630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.592781909789231e-05,
+      "loss": 4.1132,
+      "step": 1631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.592214702540916e-05,
+      "loss": 4.1055,
+      "step": 1632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5916471356166384e-05,
+      "loss": 4.1009,
+      "step": 1633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.59107920911397e-05,
+      "loss": 4.089,
+      "step": 1634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.590510923130544e-05,
+      "loss": 3.8787,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.589942277764054e-05,
+      "loss": 4.0291,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5893732731122585e-05,
+      "loss": 4.0107,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.588803909272975e-05,
+      "loss": 3.9806,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.588234186344084e-05,
+      "loss": 4.0329,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.587664104423528e-05,
+      "loss": 4.0527,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.58709366360931e-05,
+      "loss": 4.0582,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.586522863999495e-05,
+      "loss": 4.1687,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.585951705692211e-05,
+      "loss": 4.1337,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.585380188785646e-05,
+      "loss": 3.939,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.584808313378051e-05,
+      "loss": 4.092,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.584236079567738e-05,
+      "loss": 4.1756,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.58366348745308e-05,
+      "loss": 4.0356,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5830905371325125e-05,
+      "loss": 4.1135,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.582517228704533e-05,
+      "loss": 3.9985,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5819435622676985e-05,
+      "loss": 4.1369,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5813695379206314e-05,
+      "loss": 4.0222,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.58079515576201e-05,
+      "loss": 4.1846,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5802204158905784e-05,
+      "loss": 4.0385,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5796453184051416e-05,
+      "loss": 4.1363,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.579069863404566e-05,
+      "loss": 4.0693,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.578494050987777e-05,
+      "loss": 4.128,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.577917881253766e-05,
+      "loss": 4.1169,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.577341354301581e-05,
+      "loss": 4.0964,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.576764470230335e-05,
+      "loss": 4.0973,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5761872291392e-05,
+      "loss": 3.9717,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.575609631127412e-05,
+      "loss": 4.0854,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.575031676294266e-05,
+      "loss": 4.01,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.574453364739118e-05,
+      "loss": 3.9116,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5738746965613876e-05,
+      "loss": 4.1264,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.573295671860555e-05,
+      "loss": 4.1075,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.572716290736161e-05,
+      "loss": 4.0107,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5721365532878066e-05,
+      "loss": 3.9673,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5715564596151575e-05,
+      "loss": 4.0965,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.570976009817936e-05,
+      "loss": 4.1024,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5703952039959305e-05,
+      "loss": 4.1388,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5698140422489875e-05,
+      "loss": 4.0694,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.569232524677016e-05,
+      "loss": 3.9985,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.568650651379984e-05,
+      "loss": 3.9839,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5680684224579226e-05,
+      "loss": 4.1255,
+      "step": 1674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.567485838010925e-05,
+      "loss": 4.0946,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.566902898139144e-05,
+      "loss": 4.0694,
+      "step": 1676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5663196029427926e-05,
+      "loss": 3.9834,
+      "step": 1677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.565735952522146e-05,
+      "loss": 4.0298,
+      "step": 1678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5651519469775416e-05,
+      "loss": 4.0244,
+      "step": 1679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.564567586409377e-05,
+      "loss": 4.1596,
+      "step": 1680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5639828709181086e-05,
+      "loss": 4.107,
+      "step": 1681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5633978006042574e-05,
+      "loss": 4.1342,
+      "step": 1682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5628123755684036e-05,
+      "loss": 4.062,
+      "step": 1683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.562226595911188e-05,
+      "loss": 4.1051,
+      "step": 1684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.561640461733313e-05,
+      "loss": 4.0136,
+      "step": 1685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5610539731355426e-05,
+      "loss": 4.0438,
+      "step": 1686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5604671302187e-05,
+      "loss": 4.102,
+      "step": 1687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.559879933083671e-05,
+      "loss": 3.9856,
+      "step": 1688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5592923818314015e-05,
+      "loss": 4.0115,
+      "step": 1689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.558704476562897e-05,
+      "loss": 3.9635,
+      "step": 1690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.558116217379228e-05,
+      "loss": 4.0701,
+      "step": 1691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.557527604381521e-05,
+      "loss": 4.0705,
+      "step": 1692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5569386376709656e-05,
+      "loss": 4.0559,
+      "step": 1693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.556349317348812e-05,
+      "loss": 3.8675,
+      "step": 1694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.555759643516372e-05,
+      "loss": 4.1144,
+      "step": 1695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5551696162750166e-05,
+      "loss": 4.1112,
+      "step": 1696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5545792357261785e-05,
+      "loss": 4.036,
+      "step": 1697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.553988501971351e-05,
+      "loss": 4.0898,
+      "step": 1698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.553397415112089e-05,
+      "loss": 4.0035,
+      "step": 1699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5528059752500055e-05,
+      "loss": 4.125,
+      "step": 1700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.552214182486777e-05,
+      "loss": 3.8995,
+      "step": 1701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.551622036924139e-05,
+      "loss": 4.2235,
+      "step": 1702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.551029538663889e-05,
+      "loss": 4.1744,
+      "step": 1703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.550436687807883e-05,
+      "loss": 3.8676,
+      "step": 1704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5498434844580405e-05,
+      "loss": 3.9368,
+      "step": 1705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.549249928716338e-05,
+      "loss": 3.9872,
+      "step": 1706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.548656020684817e-05,
+      "loss": 3.9562,
+      "step": 1707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.548061760465575e-05,
+      "loss": 3.959,
+      "step": 1708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.547467148160774e-05,
+      "loss": 4.1083,
+      "step": 1709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.546872183872634e-05,
+      "loss": 4.0228,
+      "step": 1710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5462768677034357e-05,
+      "loss": 4.0739,
+      "step": 1711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.545681199755521e-05,
+      "loss": 4.0309,
+      "step": 1712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.545085180131293e-05,
+      "loss": 4.0453,
+      "step": 1713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5444888089332136e-05,
+      "loss": 4.0971,
+      "step": 1714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.543892086263807e-05,
+      "loss": 4.0594,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5432950122256555e-05,
+      "loss": 3.9623,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.542697586921403e-05,
+      "loss": 4.1893,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5420998104537546e-05,
+      "loss": 4.1163,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.541501682925475e-05,
+      "loss": 4.0641,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.540903204439389e-05,
+      "loss": 4.1075,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.540304375098382e-05,
+      "loss": 4.1429,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.539705195005399e-05,
+      "loss": 4.0287,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.539105664263448e-05,
+      "loss": 3.9141,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.538505782975593e-05,
+      "loss": 3.9312,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.537905551244961e-05,
+      "loss": 4.0869,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.53730496917474e-05,
+      "loss": 4.129,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.536704036868177e-05,
+      "loss": 4.0634,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.536102754428577e-05,
+      "loss": 4.0789,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.53550112195931e-05,
+      "loss": 4.0607,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.534899139563803e-05,
+      "loss": 4.1821,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5342968073455424e-05,
+      "loss": 4.132,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5336941254080775e-05,
+      "loss": 3.9951,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.533091093855016e-05,
+      "loss": 4.1447,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5324877127900255e-05,
+      "loss": 3.9438,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.531883982316835e-05,
+      "loss": 4.0321,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5312799025392315e-05,
+      "loss": 4.0494,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.530675473561065e-05,
+      "loss": 3.9915,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5300706954862425e-05,
+      "loss": 3.8937,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.529465568418733e-05,
+      "loss": 4.1818,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.528860092462564e-05,
+      "loss": 4.1371,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.528254267721825e-05,
+      "loss": 4.0174,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.527648094300665e-05,
+      "loss": 4.1361,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.52704157230329e-05,
+      "loss": 3.9199,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5264347018339695e-05,
+      "loss": 4.1271,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5258274829970304e-05,
+      "loss": 4.074,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.525219915896863e-05,
+      "loss": 4.0914,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.524612000637913e-05,
+      "loss": 4.1079,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5240037373246886e-05,
+      "loss": 4.1572,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.523395126061758e-05,
+      "loss": 4.0368,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.522786166953747e-05,
+      "loss": 4.1629,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.522176860105345e-05,
+      "loss": 4.0793,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.521567205621297e-05,
+      "loss": 4.0355,
+      "step": 1752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5209572036064107e-05,
+      "loss": 4.1254,
+      "step": 1753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5203468541655514e-05,
+      "loss": 3.9507,
+      "step": 1754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5197361574036466e-05,
+      "loss": 3.9756,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.519125113425681e-05,
+      "loss": 4.1597,
+      "step": 1756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5185137223367016e-05,
+      "loss": 3.9262,
+      "step": 1757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.517901984241812e-05,
+      "loss": 4.1564,
+      "step": 1758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.517289899246178e-05,
+      "loss": 4.0556,
+      "step": 1759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.516677467455024e-05,
+      "loss": 4.0879,
+      "step": 1760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.516064688973634e-05,
+      "loss": 4.0469,
+      "step": 1761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5154515639073515e-05,
+      "loss": 4.0681,
+      "step": 1762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5148380923615806e-05,
+      "loss": 4.033,
+      "step": 1763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.514224274441783e-05,
+      "loss": 4.0336,
+      "step": 1764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.513610110253482e-05,
+      "loss": 3.9931,
+      "step": 1765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.512995599902259e-05,
+      "loss": 3.9625,
+      "step": 1766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5123807434937557e-05,
+      "loss": 4.0143,
+      "step": 1767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5117655411336736e-05,
+      "loss": 3.9372,
+      "step": 1768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5111499929277726e-05,
+      "loss": 4.0688,
+      "step": 1769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.510534098981872e-05,
+      "loss": 3.943,
+      "step": 1770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5099178594018524e-05,
+      "loss": 3.9505,
+      "step": 1771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.509301274293651e-05,
+      "loss": 4.0862,
+      "step": 1772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.508684343763268e-05,
+      "loss": 4.0342,
+      "step": 1773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.508067067916758e-05,
+      "loss": 3.9773,
+      "step": 1774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5074494468602405e-05,
+      "loss": 4.103,
+      "step": 1775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.50683148069989e-05,
+      "loss": 4.1489,
+      "step": 1776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.506213169541944e-05,
+      "loss": 4.0359,
+      "step": 1777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.505594513492695e-05,
+      "loss": 4.0657,
+      "step": 1778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.504975512658499e-05,
+      "loss": 4.0665,
+      "step": 1779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.504356167145768e-05,
+      "loss": 4.0917,
+      "step": 1780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.503736477060976e-05,
+      "loss": 4.1513,
+      "step": 1781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.503116442510654e-05,
+      "loss": 4.125,
+      "step": 1782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5024960636013934e-05,
+      "loss": 4.0105,
+      "step": 1783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.501875340439845e-05,
+      "loss": 4.0932,
+      "step": 1784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.501254273132717e-05,
+      "loss": 4.0721,
+      "step": 1785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.50063286178678e-05,
+      "loss": 4.0424,
+      "step": 1786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.500011106508859e-05,
+      "loss": 4.0022,
+      "step": 1787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4993890074058444e-05,
+      "loss": 4.0112,
+      "step": 1788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.49876656458468e-05,
+      "loss": 3.9874,
+      "step": 1789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4981437781523714e-05,
+      "loss": 4.0063,
+      "step": 1790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.497520648215982e-05,
+      "loss": 4.0584,
+      "step": 1791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.496897174882637e-05,
+      "loss": 3.9498,
+      "step": 1792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4962733582595176e-05,
+      "loss": 4.0289,
+      "step": 1793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4956491984538654e-05,
+      "loss": 4.0497,
+      "step": 1794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.49502469557298e-05,
+      "loss": 4.1827,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4943998497242204e-05,
+      "loss": 3.9756,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.493774661015007e-05,
+      "loss": 3.9145,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4931491295528146e-05,
+      "loss": 4.0736,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4925232554451805e-05,
+      "loss": 3.9447,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4918970387996987e-05,
+      "loss": 4.136,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.491270479724024e-05,
+      "loss": 3.9528,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.49064357832587e-05,
+      "loss": 4.143,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.490016334713007e-05,
+      "loss": 4.1073,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4893887489932664e-05,
+      "loss": 4.0089,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4887608212745364e-05,
+      "loss": 3.9815,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.488132551664765e-05,
+      "loss": 4.1614,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4875039402719606e-05,
+      "loss": 4.0529,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.486874987204187e-05,
+      "loss": 4.197,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.486245692569569e-05,
+      "loss": 4.0654,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.485616056476291e-05,
+      "loss": 4.0994,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4849860790325924e-05,
+      "loss": 4.1399,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.484355760346776e-05,
+      "loss": 3.9748,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4837251005271985e-05,
+      "loss": 4.0264,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.48309409968228e-05,
+      "loss": 4.1965,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4824627579204956e-05,
+      "loss": 4.0234,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.48183107535038e-05,
+      "loss": 3.9168,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.481199052080527e-05,
+      "loss": 4.0113,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.48056668821959e-05,
+      "loss": 4.0596,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.479933983876278e-05,
+      "loss": 4.0727,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.47930093915936e-05,
+      "loss": 4.0862,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.478667554177666e-05,
+      "loss": 4.039,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4780338290400793e-05,
+      "loss": 3.9848,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.477399763855547e-05,
+      "loss": 3.9712,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.476765358733071e-05,
+      "loss": 3.9447,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.476130613781713e-05,
+      "loss": 4.0621,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4754955291105937e-05,
+      "loss": 4.1461,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4748601048288904e-05,
+      "loss": 4.0313,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4742243410458415e-05,
+      "loss": 4.0409,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4735882378707405e-05,
+      "loss": 4.0615,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.472951795412942e-05,
+      "loss": 3.967,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.472315013781857e-05,
+      "loss": 4.2178,
+      "step": 1831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.471677893086956e-05,
+      "loss": 4.0308,
+      "step": 1832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.471040433437769e-05,
+      "loss": 3.9147,
+      "step": 1833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4704026349438796e-05,
+      "loss": 4.196,
+      "step": 1834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.469764497714934e-05,
+      "loss": 4.0984,
+      "step": 1835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.469126021860637e-05,
+      "loss": 3.9856,
+      "step": 1836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4684872074907494e-05,
+      "loss": 4.0394,
+      "step": 1837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.467848054715089e-05,
+      "loss": 3.985,
+      "step": 1838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.467208563643534e-05,
+      "loss": 4.0331,
+      "step": 1839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.466568734386023e-05,
+      "loss": 3.9984,
+      "step": 1840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4659285670525464e-05,
+      "loss": 3.9853,
+      "step": 1841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.465288061753159e-05,
+      "loss": 3.9773,
+      "step": 1842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.464647218597971e-05,
+      "loss": 4.0568,
+      "step": 1843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.464006037697148e-05,
+      "loss": 3.8515,
+      "step": 1844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4633645191609195e-05,
+      "loss": 3.9487,
+      "step": 1845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4627226630995686e-05,
+      "loss": 3.9608,
+      "step": 1846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4620804696234387e-05,
+      "loss": 3.9468,
+      "step": 1847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.461437938842929e-05,
+      "loss": 4.1175,
+      "step": 1848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4607950708684996e-05,
+      "loss": 3.9478,
+      "step": 1849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.460151865810666e-05,
+      "loss": 3.99,
+      "step": 1850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.459508323780001e-05,
+      "loss": 4.1642,
+      "step": 1851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.45886444488714e-05,
+      "loss": 3.9962,
+      "step": 1852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.458220229242771e-05,
+      "loss": 4.0841,
+      "step": 1853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.457575676957644e-05,
+      "loss": 4.0834,
+      "step": 1854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.456930788142563e-05,
+      "loss": 3.9308,
+      "step": 1855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.456285562908393e-05,
+      "loss": 4.0401,
+      "step": 1856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.455640001366056e-05,
+      "loss": 4.0871,
+      "step": 1857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4549941036265306e-05,
+      "loss": 4.0579,
+      "step": 1858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.454347869800854e-05,
+      "loss": 4.0019,
+      "step": 1859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.453701300000124e-05,
+      "loss": 4.0836,
+      "step": 1860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.45305439433549e-05,
+      "loss": 4.063,
+      "step": 1861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4524071529181635e-05,
+      "loss": 4.0726,
+      "step": 1862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4517595758594124e-05,
+      "loss": 3.9224,
+      "step": 1863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.451111663270565e-05,
+      "loss": 3.9699,
+      "step": 1864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4504634152630025e-05,
+      "loss": 4.1364,
+      "step": 1865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.449814831948168e-05,
+      "loss": 4.1338,
+      "step": 1866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.449165913437559e-05,
+      "loss": 4.1586,
+      "step": 1867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.448516659842733e-05,
+      "loss": 4.0757,
+      "step": 1868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.447867071275305e-05,
+      "loss": 3.8878,
+      "step": 1869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.447217147846944e-05,
+      "loss": 4.0648,
+      "step": 1870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.446566889669382e-05,
+      "loss": 4.0788,
+      "step": 1871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.445916296854406e-05,
+      "loss": 4.0718,
+      "step": 1872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.445265369513859e-05,
+      "loss": 4.0398,
+      "step": 1873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4446141077596424e-05,
+      "loss": 4.2148,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.443962511703719e-05,
+      "loss": 3.9534,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4433105814581015e-05,
+      "loss": 4.0657,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4426583171348666e-05,
+      "loss": 4.0345,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.442005718846147e-05,
+      "loss": 4.1051,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.441352786704129e-05,
+      "loss": 4.0849,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.440699520821062e-05,
+      "loss": 3.9415,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.440045921309249e-05,
+      "loss": 4.0601,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.439391988281051e-05,
+      "loss": 4.1194,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.438737721848886e-05,
+      "loss": 3.9127,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4380831221252316e-05,
+      "loss": 3.954,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.437428189222621e-05,
+      "loss": 4.0556,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.436772923253643e-05,
+      "loss": 4.1722,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.436117324330946e-05,
+      "loss": 3.9959,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.435461392567236e-05,
+      "loss": 4.2194,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.434805128075276e-05,
+      "loss": 3.9694,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.434148530967883e-05,
+      "loss": 3.8755,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.433491601357935e-05,
+      "loss": 4.083,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.432834339358367e-05,
+      "loss": 3.926,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4321767450821675e-05,
+      "loss": 4.0227,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.431518818642386e-05,
+      "loss": 3.9319,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.430860560152128e-05,
+      "loss": 4.0281,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.430201969724555e-05,
+      "loss": 4.0981,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.429543047472886e-05,
+      "loss": 4.1391,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.428883793510399e-05,
+      "loss": 4.0585,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4282242079504257e-05,
+      "loss": 4.0815,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.427564290906357e-05,
+      "loss": 4.058,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.426904042491641e-05,
+      "loss": 4.1091,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.426243462819781e-05,
+      "loss": 4.1111,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.425582552004339e-05,
+      "loss": 3.9484,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.424921310158933e-05,
+      "loss": 3.9769,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4242597373972386e-05,
+      "loss": 4.0913,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.423597833832986e-05,
+      "loss": 3.9946,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.422935599579967e-05,
+      "loss": 4.0619,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.422273034752026e-05,
+      "loss": 4.1054,
+      "step": 1908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4216101394630645e-05,
+      "loss": 3.921,
+      "step": 1909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.420946913827043e-05,
+      "loss": 4.0875,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.420283357957978e-05,
+      "loss": 4.0162,
+      "step": 1911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.419619471969941e-05,
+      "loss": 3.9199,
+      "step": 1912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.418955255977064e-05,
+      "loss": 4.1886,
+      "step": 1913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.418290710093531e-05,
+      "loss": 4.1413,
+      "step": 1914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4176258344335865e-05,
+      "loss": 4.1845,
+      "step": 1915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.41696062911153e-05,
+      "loss": 3.908,
+      "step": 1916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.416295094241718e-05,
+      "loss": 3.9286,
+      "step": 1917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.415629229938564e-05,
+      "loss": 4.1038,
+      "step": 1918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.414963036316537e-05,
+      "loss": 4.07,
+      "step": 1919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.414296513490164e-05,
+      "loss": 4.1696,
+      "step": 1920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.413629661574028e-05,
+      "loss": 3.8335,
+      "step": 1921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4129624806827686e-05,
+      "loss": 4.058,
+      "step": 1922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4122949709310816e-05,
+      "loss": 4.1074,
+      "step": 1923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.41162713243372e-05,
+      "loss": 3.9925,
+      "step": 1924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.410958965305493e-05,
+      "loss": 4.0679,
+      "step": 1925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4102904696612654e-05,
+      "loss": 4.0936,
+      "step": 1926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.409621645615961e-05,
+      "loss": 3.938,
+      "step": 1927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4089524932845575e-05,
+      "loss": 4.0786,
+      "step": 1928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4082830127820897e-05,
+      "loss": 4.0132,
+      "step": 1929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.407613204223648e-05,
+      "loss": 4.0671,
+      "step": 1930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4069430677243835e-05,
+      "loss": 4.0094,
+      "step": 1931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4062726033994975e-05,
+      "loss": 4.0467,
+      "step": 1932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4056018113642516e-05,
+      "loss": 3.9997,
+      "step": 1933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4049306917339624e-05,
+      "loss": 3.9806,
+      "step": 1934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4042592446240046e-05,
+      "loss": 4.0221,
+      "step": 1935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4035874701498055e-05,
+      "loss": 3.9876,
+      "step": 1936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4029153684268526e-05,
+      "loss": 3.8967,
+      "step": 1937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.402242939570687e-05,
+      "loss": 3.8928,
+      "step": 1938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.401570183696907e-05,
+      "loss": 4.0167,
+      "step": 1939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4008971009211686e-05,
+      "loss": 4.0245,
+      "step": 1940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.400223691359181e-05,
+      "loss": 3.9689,
+      "step": 1941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3995499551267114e-05,
+      "loss": 4.0042,
+      "step": 1942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3988758923395834e-05,
+      "loss": 3.992,
+      "step": 1943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3982015031136756e-05,
+      "loss": 4.1142,
+      "step": 1944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.397526787564923e-05,
+      "loss": 4.2171,
+      "step": 1945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.396851745809318e-05,
+      "loss": 4.0149,
+      "step": 1946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.396176377962908e-05,
+      "loss": 4.1179,
+      "step": 1947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3955006841417964e-05,
+      "loss": 3.8889,
+      "step": 1948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.394824664462143e-05,
+      "loss": 3.96,
+      "step": 1949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.394148319040162e-05,
+      "loss": 3.896,
+      "step": 1950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.393471647992127e-05,
+      "loss": 3.9781,
+      "step": 1951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3927946514343636e-05,
+      "loss": 4.0048,
+      "step": 1952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3921173294832574e-05,
+      "loss": 3.8941,
+      "step": 1953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3914396822552465e-05,
+      "loss": 3.9531,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.390761709866827e-05,
+      "loss": 4.0999,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3900834124345505e-05,
+      "loss": 4.0412,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3894047900750224e-05,
+      "loss": 4.0553,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3887258429049084e-05,
+      "loss": 4.0454,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.388046571040925e-05,
+      "loss": 4.0734,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.387366974599848e-05,
+      "loss": 3.9981,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3866870536985085e-05,
+      "loss": 3.9625,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.386006808453792e-05,
+      "loss": 3.9911,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.38532623898264e-05,
+      "loss": 3.9827,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3846453454020514e-05,
+      "loss": 4.024,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.383964127829079e-05,
+      "loss": 4.0617,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3832825863808326e-05,
+      "loss": 4.1159,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.382600721174477e-05,
+      "loss": 3.9807,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3819185323272314e-05,
+      "loss": 4.0341,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.381236019956374e-05,
+      "loss": 3.9521,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3805531841792345e-05,
+      "loss": 4.0402,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.379870025113203e-05,
+      "loss": 4.0797,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3791865428757196e-05,
+      "loss": 3.8791,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3785027375842846e-05,
+      "loss": 4.0127,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.377818609356451e-05,
+      "loss": 4.0614,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.37713415830983e-05,
+      "loss": 4.0181,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.376449384562085e-05,
+      "loss": 4.1744,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.375764288230936e-05,
+      "loss": 4.1009,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.375078869434161e-05,
+      "loss": 3.966,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.374393128289591e-05,
+      "loss": 4.0573,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3737070649151126e-05,
+      "loss": 3.9233,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3730206794286674e-05,
+      "loss": 4.1367,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3723339719482534e-05,
+      "loss": 4.1717,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.371646942591924e-05,
+      "loss": 4.0031,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.370959591477787e-05,
+      "loss": 3.9305,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.370271918724006e-05,
+      "loss": 4.1022,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3695839244488e-05,
+      "loss": 4.0073,
+      "step": 1986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3688956087704435e-05,
+      "loss": 3.9984,
+      "step": 1987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.368206971807266e-05,
+      "loss": 3.9015,
+      "step": 1988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.367518013677652e-05,
+      "loss": 4.0287,
+      "step": 1989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.36682873450004e-05,
+      "loss": 4.0351,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.366139134392928e-05,
+      "loss": 3.9796,
+      "step": 1991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3654492134748636e-05,
+      "loss": 4.0607,
+      "step": 1992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.364758971864455e-05,
+      "loss": 4.0175,
+      "step": 1993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.364068409680359e-05,
+      "loss": 3.9283,
+      "step": 1994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.363377527041294e-05,
+      "loss": 4.0371,
+      "step": 1995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.36268632406603e-05,
+      "loss": 4.1458,
+      "step": 1996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.361994800873392e-05,
+      "loss": 3.9455,
+      "step": 1997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.361302957582263e-05,
+      "loss": 4.06,
+      "step": 1998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.360610794311577e-05,
+      "loss": 4.0235,
+      "step": 1999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3599183111803254e-05,
+      "loss": 4.1721,
+      "step": 2000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.359225508307554e-05,
+      "loss": 4.0375,
+      "step": 2001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.358532385812364e-05,
+      "loss": 4.1049,
+      "step": 2002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.357838943813911e-05,
+      "loss": 4.1051,
+      "step": 2003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.357145182431405e-05,
+      "loss": 4.0099,
+      "step": 2004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3564511017841126e-05,
+      "loss": 3.8946,
+      "step": 2005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3557567019913535e-05,
+      "loss": 4.0031,
+      "step": 2006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.355061983172504e-05,
+      "loss": 3.993,
+      "step": 2007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.354366945446994e-05,
+      "loss": 3.9084,
+      "step": 2008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3536715889343084e-05,
+      "loss": 3.9994,
+      "step": 2009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3529759137539874e-05,
+      "loss": 4.1271,
+      "step": 2010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.352279920025624e-05,
+      "loss": 3.969,
+      "step": 2011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3515836078688696e-05,
+      "loss": 3.9395,
+      "step": 2012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.350886977403428e-05,
+      "loss": 4.0894,
+      "step": 2013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.350190028749057e-05,
+      "loss": 3.975,
+      "step": 2014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3494927620255715e-05,
+      "loss": 3.9683,
+      "step": 2015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.348795177352839e-05,
+      "loss": 4.0504,
+      "step": 2016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.348097274850782e-05,
+      "loss": 4.0688,
+      "step": 2017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3473990546393785e-05,
+      "loss": 4.0691,
+      "step": 2018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.346700516838661e-05,
+      "loss": 3.9823,
+      "step": 2019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.346001661568715e-05,
+      "loss": 4.0268,
+      "step": 2020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.345302488949684e-05,
+      "loss": 3.9442,
+      "step": 2021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3446029991017615e-05,
+      "loss": 4.0733,
+      "step": 2022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3439031921451995e-05,
+      "loss": 3.7964,
+      "step": 2023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.343203068200302e-05,
+      "loss": 3.9276,
+      "step": 2024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.342502627387429e-05,
+      "loss": 4.0096,
+      "step": 2025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3418018698269946e-05,
+      "loss": 4.0271,
+      "step": 2026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3411007956394666e-05,
+      "loss": 3.981,
+      "step": 2027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3403994049453677e-05,
+      "loss": 4.0865,
+      "step": 2028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.339697697865276e-05,
+      "loss": 3.9942,
+      "step": 2029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.338995674519822e-05,
+      "loss": 4.1338,
+      "step": 2030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.338293335029692e-05,
+      "loss": 4.0603,
+      "step": 2031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.337590679515626e-05,
+      "loss": 4.0424,
+      "step": 2032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.336887708098419e-05,
+      "loss": 4.0412,
+      "step": 2033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3361844208989214e-05,
+      "loss": 4.092,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.335480818038034e-05,
+      "loss": 3.9915,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3347768996367144e-05,
+      "loss": 4.1343,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3340726658159766e-05,
+      "loss": 4.0217,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3333681166968845e-05,
+      "loss": 3.9686,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.332663252400558e-05,
+      "loss": 4.0361,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.331958073048174e-05,
+      "loss": 4.1315,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3312525787609585e-05,
+      "loss": 4.2054,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.330546769660195e-05,
+      "loss": 4.0022,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3298406458672205e-05,
+      "loss": 4.0962,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3291342075034256e-05,
+      "loss": 3.9426,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.328427454690256e-05,
+      "loss": 4.0087,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.327720387549209e-05,
+      "loss": 4.0616,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.327013006201839e-05,
+      "loss": 4.1092,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.326305310769753e-05,
+      "loss": 4.1953,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.325597301374611e-05,
+      "loss": 4.0784,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.324888978138129e-05,
+      "loss": 3.9711,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3241803411820756e-05,
+      "loss": 3.8589,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.323471390628274e-05,
+      "loss": 3.8316,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.322762126598601e-05,
+      "loss": 4.1133,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.322052549214987e-05,
+      "loss": 4.1127,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3213426585994164e-05,
+      "loss": 4.1253,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.320632454873929e-05,
+      "loss": 3.7984,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3199219381606154e-05,
+      "loss": 4.1099,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.319211108581622e-05,
+      "loss": 3.978,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3184999662591494e-05,
+      "loss": 4.1255,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3177885113154506e-05,
+      "loss": 4.029,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.317076743872833e-05,
+      "loss": 3.9407,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.316364664053658e-05,
+      "loss": 3.9095,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.31565227198034e-05,
+      "loss": 4.0605,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.314939567775347e-05,
+      "loss": 4.092,
+      "step": 2064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.314226551561202e-05,
+      "loss": 4.0556,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.313513223460482e-05,
+      "loss": 4.0691,
+      "step": 2066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.312799583595814e-05,
+      "loss": 4.1829,
+      "step": 2067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.312085632089881e-05,
+      "loss": 4.0514,
+      "step": 2068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.31137136906542e-05,
+      "loss": 4.0121,
+      "step": 2069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.310656794645223e-05,
+      "loss": 4.1858,
+      "step": 2070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.309941908952131e-05,
+      "loss": 4.0041,
+      "step": 2071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3092267121090425e-05,
+      "loss": 4.0322,
+      "step": 2072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.308511204238908e-05,
+      "loss": 4.1168,
+      "step": 2073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.30779538546473e-05,
+      "loss": 4.1027,
+      "step": 2074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.307079255909569e-05,
+      "loss": 4.0641,
+      "step": 2075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.306362815696534e-05,
+      "loss": 4.091,
+      "step": 2076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3056460649487904e-05,
+      "loss": 3.9452,
+      "step": 2077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.304929003789555e-05,
+      "loss": 3.908,
+      "step": 2078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.304211632342099e-05,
+      "loss": 3.8844,
+      "step": 2079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3034939507297475e-05,
+      "loss": 3.899,
+      "step": 2080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3027759590758775e-05,
+      "loss": 3.9599,
+      "step": 2081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3020576575039216e-05,
+      "loss": 4.0276,
+      "step": 2082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3013390461373625e-05,
+      "loss": 4.011,
+      "step": 2083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.300620125099738e-05,
+      "loss": 4.0031,
+      "step": 2084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.29990089451464e-05,
+      "loss": 4.1096,
+      "step": 2085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.299181354505712e-05,
+      "loss": 4.0275,
+      "step": 2086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.298461505196652e-05,
+      "loss": 4.1175,
+      "step": 2087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.297741346711208e-05,
+      "loss": 4.1343,
+      "step": 2088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.297020879173186e-05,
+      "loss": 3.9313,
+      "step": 2089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.296300102706442e-05,
+      "loss": 4.038,
+      "step": 2090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.295579017434886e-05,
+      "loss": 4.0804,
+      "step": 2091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.294857623482481e-05,
+      "loss": 4.0495,
+      "step": 2092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.294135920973242e-05,
+      "loss": 4.0432,
+      "step": 2093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.29341391003124e-05,
+      "loss": 4.1513,
+      "step": 2094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2926915907805954e-05,
+      "loss": 4.1041,
+      "step": 2095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.291968963345483e-05,
+      "loss": 4.0108,
+      "step": 2096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.291246027850132e-05,
+      "loss": 4.1166,
+      "step": 2097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.290522784418823e-05,
+      "loss": 4.0002,
+      "step": 2098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.28979923317589e-05,
+      "loss": 4.0793,
+      "step": 2099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.289075374245719e-05,
+      "loss": 3.9334,
+      "step": 2100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2883512077527506e-05,
+      "loss": 4.1003,
+      "step": 2101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.287626733821477e-05,
+      "loss": 4.0463,
+      "step": 2102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.286901952576445e-05,
+      "loss": 4.1291,
+      "step": 2103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2861768641422505e-05,
+      "loss": 4.1,
+      "step": 2104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2854514686435466e-05,
+      "loss": 4.1294,
+      "step": 2105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2847257662050365e-05,
+      "loss": 4.033,
+      "step": 2106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.283999756951477e-05,
+      "loss": 4.0206,
+      "step": 2107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.283273441007677e-05,
+      "loss": 3.8403,
+      "step": 2108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2825468184984993e-05,
+      "loss": 3.9361,
+      "step": 2109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.281819889548858e-05,
+      "loss": 3.9798,
+      "step": 2110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.281092654283722e-05,
+      "loss": 4.146,
+      "step": 2111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.28036511282811e-05,
+      "loss": 3.9921,
+      "step": 2112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2796372653070946e-05,
+      "loss": 4.0653,
+      "step": 2113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2789091118458034e-05,
+      "loss": 4.0058,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2781806525694126e-05,
+      "loss": 4.053,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2774518876031524e-05,
+      "loss": 3.9907,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.276722817072307e-05,
+      "loss": 3.9595,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.275993441102212e-05,
+      "loss": 4.0539,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.275263759818256e-05,
+      "loss": 4.0142,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.274533773345878e-05,
+      "loss": 3.9754,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2738034818105725e-05,
+      "loss": 4.0198,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.273072885337885e-05,
+      "loss": 4.1375,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.272341984053413e-05,
+      "loss": 3.9831,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2716107780828075e-05,
+      "loss": 4.0849,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2708792675517705e-05,
+      "loss": 4.0194,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2701474525860585e-05,
+      "loss": 4.054,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2694153333114775e-05,
+      "loss": 3.8597,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.268682909853888e-05,
+      "loss": 3.9686,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.267950182339201e-05,
+      "loss": 4.1025,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.267217150893383e-05,
+      "loss": 3.9725,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.266483815642449e-05,
+      "loss": 3.9156,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2657501767124685e-05,
+      "loss": 3.9112,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.265016234229563e-05,
+      "loss": 3.9675,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2642819883199035e-05,
+      "loss": 4.0983,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.263547439109717e-05,
+      "loss": 3.9585,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2628125867252824e-05,
+      "loss": 4.1069,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.262077431292927e-05,
+      "loss": 4.1051,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2613419729390337e-05,
+      "loss": 3.9107,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2606062117900354e-05,
+      "loss": 3.969,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.259870147972419e-05,
+      "loss": 4.0215,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.259133781612722e-05,
+      "loss": 4.0979,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.258397112837534e-05,
+      "loss": 4.0836,
+      "step": 2142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.257660141773498e-05,
+      "loss": 4.016,
+      "step": 2143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.256922868547306e-05,
+      "loss": 3.9522,
+      "step": 2144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.256185293285705e-05,
+      "loss": 3.9865,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.255447416115493e-05,
+      "loss": 4.034,
+      "step": 2146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.25470923716352e-05,
+      "loss": 3.9878,
+      "step": 2147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2539707565566854e-05,
+      "loss": 3.9623,
+      "step": 2148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.253231974421945e-05,
+      "loss": 3.9918,
+      "step": 2149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2524928908863025e-05,
+      "loss": 3.9539,
+      "step": 2150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.251753506076816e-05,
+      "loss": 3.98,
+      "step": 2151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2510138201205936e-05,
+      "loss": 4.0754,
+      "step": 2152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2502738331447956e-05,
+      "loss": 4.0438,
+      "step": 2153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2495335452766346e-05,
+      "loss": 4.0738,
+      "step": 2154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.248792956643376e-05,
+      "loss": 4.0622,
+      "step": 2155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2480520673723334e-05,
+      "loss": 4.1532,
+      "step": 2156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2473108775908756e-05,
+      "loss": 3.8815,
+      "step": 2157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2465693874264206e-05,
+      "loss": 3.966,
+      "step": 2158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2458275970064406e-05,
+      "loss": 3.918,
+      "step": 2159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.245085506458457e-05,
+      "loss": 4.0554,
+      "step": 2160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.244343115910045e-05,
+      "loss": 4.0482,
+      "step": 2161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.243600425488828e-05,
+      "loss": 4.0173,
+      "step": 2162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2428574353224846e-05,
+      "loss": 3.857,
+      "step": 2163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2421141455387435e-05,
+      "loss": 3.9567,
+      "step": 2164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.241370556265385e-05,
+      "loss": 3.9692,
+      "step": 2165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.240626667630239e-05,
+      "loss": 3.9563,
+      "step": 2166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2398824797611906e-05,
+      "loss": 3.9554,
+      "step": 2167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.239137992786173e-05,
+      "loss": 3.9022,
+      "step": 2168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.238393206833173e-05,
+      "loss": 4.041,
+      "step": 2169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.237648122030227e-05,
+      "loss": 3.8663,
+      "step": 2170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.236902738505425e-05,
+      "loss": 4.0125,
+      "step": 2171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2361570563869056e-05,
+      "loss": 4.1445,
+      "step": 2172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.235411075802861e-05,
+      "loss": 3.9701,
+      "step": 2173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.234664796881535e-05,
+      "loss": 3.9252,
+      "step": 2174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.233918219751219e-05,
+      "loss": 4.0578,
+      "step": 2175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.23317134454026e-05,
+      "loss": 4.0768,
+      "step": 2176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.232424171377055e-05,
+      "loss": 3.8815,
+      "step": 2177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.23167670039005e-05,
+      "loss": 4.0529,
+      "step": 2178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.230928931707746e-05,
+      "loss": 4.0541,
+      "step": 2179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.230180865458692e-05,
+      "loss": 4.0411,
+      "step": 2180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2294325017714875e-05,
+      "loss": 4.0509,
+      "step": 2181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.228683840774788e-05,
+      "loss": 4.0885,
+      "step": 2182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.227934882597296e-05,
+      "loss": 3.9628,
+      "step": 2183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.227185627367765e-05,
+      "loss": 3.8319,
+      "step": 2184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.226436075215001e-05,
+      "loss": 4.0177,
+      "step": 2185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.225686226267862e-05,
+      "loss": 4.0622,
+      "step": 2186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.224936080655255e-05,
+      "loss": 3.9829,
+      "step": 2187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.224185638506138e-05,
+      "loss": 3.9371,
+      "step": 2188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.223434899949521e-05,
+      "loss": 3.956,
+      "step": 2189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.222683865114465e-05,
+      "loss": 4.1096,
+      "step": 2190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.221932534130082e-05,
+      "loss": 3.9535,
+      "step": 2191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2211809071255345e-05,
+      "loss": 3.9436,
+      "step": 2192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2204289842300345e-05,
+      "loss": 3.9775,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2196767655728474e-05,
+      "loss": 4.0859,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.218924251283288e-05,
+      "loss": 4.1018,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2181714414907225e-05,
+      "loss": 4.0309,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2174183363245675e-05,
+      "loss": 4.0591,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.216664935914291e-05,
+      "loss": 4.0515,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.215911240389409e-05,
+      "loss": 3.8485,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2151572498794936e-05,
+      "loss": 3.9616,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.214402964514163e-05,
+      "loss": 3.9021,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.213648384423088e-05,
+      "loss": 4.0773,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.212893509735989e-05,
+      "loss": 3.9817,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.212138340582637e-05,
+      "loss": 4.0663,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2113828770928575e-05,
+      "loss": 3.9969,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.21062711939652e-05,
+      "loss": 3.9085,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.20987106762355e-05,
+      "loss": 4.0802,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2091147219039206e-05,
+      "loss": 3.8817,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.208358082367657e-05,
+      "loss": 4.0125,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2076011491448345e-05,
+      "loss": 4.078,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2068439223655775e-05,
+      "loss": 4.0689,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.206086402160063e-05,
+      "loss": 3.9264,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.205328588658517e-05,
+      "loss": 4.0856,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.204570481991217e-05,
+      "loss": 4.0034,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2038120822884907e-05,
+      "loss": 4.0312,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2030533896807145e-05,
+      "loss": 4.0487,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.202294404298317e-05,
+      "loss": 3.9967,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2015351262717775e-05,
+      "loss": 4.0282,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.200775555731623e-05,
+      "loss": 3.939,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.200015692808434e-05,
+      "loss": 4.0204,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.199255537632838e-05,
+      "loss": 3.9853,
+      "step": 2221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.198495090335517e-05,
+      "loss": 3.8835,
+      "step": 2222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.197734351047199e-05,
+      "loss": 3.9481,
+      "step": 2223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1969733198986636e-05,
+      "loss": 3.9945,
+      "step": 2224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.196211997020742e-05,
+      "loss": 4.1035,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.195450382544315e-05,
+      "loss": 3.983,
+      "step": 2226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1946884766003105e-05,
+      "loss": 3.9502,
+      "step": 2227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.19392627931971e-05,
+      "loss": 3.954,
+      "step": 2228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.193163790833545e-05,
+      "loss": 4.0654,
+      "step": 2229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.192401011272896e-05,
+      "loss": 3.8879,
+      "step": 2230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.191637940768893e-05,
+      "loss": 4.0518,
+      "step": 2231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.190874579452716e-05,
+      "loss": 3.9771,
+      "step": 2232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1901109274555974e-05,
+      "loss": 4.003,
+      "step": 2233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1893469849088164e-05,
+      "loss": 3.9303,
+      "step": 2234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.188582751943705e-05,
+      "loss": 3.9796,
+      "step": 2235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.187818228691641e-05,
+      "loss": 4.0327,
+      "step": 2236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.187053415284059e-05,
+      "loss": 4.0101,
+      "step": 2237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1862883118524353e-05,
+      "loss": 3.9504,
+      "step": 2238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.185522918528302e-05,
+      "loss": 4.0059,
+      "step": 2239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1847572354432384e-05,
+      "loss": 3.9427,
+      "step": 2240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1839912627288754e-05,
+      "loss": 3.9586,
+      "step": 2241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.183225000516891e-05,
+      "loss": 3.9503,
+      "step": 2242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.182458448939016e-05,
+      "loss": 4.0631,
+      "step": 2243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1816916081270284e-05,
+      "loss": 4.0995,
+      "step": 2244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.180924478212758e-05,
+      "loss": 3.9011,
+      "step": 2245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1801570593280824e-05,
+      "loss": 3.9268,
+      "step": 2246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.179389351604931e-05,
+      "loss": 4.0089,
+      "step": 2247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.178621355175281e-05,
+      "loss": 4.0303,
+      "step": 2248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1778530701711586e-05,
+      "loss": 3.9889,
+      "step": 2249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.177084496724643e-05,
+      "loss": 4.032,
+      "step": 2250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.176315634967859e-05,
+      "loss": 4.1312,
+      "step": 2251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.175546485032985e-05,
+      "loss": 4.0486,
+      "step": 2252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.174777047052245e-05,
+      "loss": 4.0812,
+      "step": 2253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.174007321157916e-05,
+      "loss": 3.9607,
+      "step": 2254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1732373074823204e-05,
+      "loss": 4.1672,
+      "step": 2255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.172467006157834e-05,
+      "loss": 4.0142,
+      "step": 2256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.171696417316881e-05,
+      "loss": 3.8674,
+      "step": 2257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.170925541091933e-05,
+      "loss": 4.0723,
+      "step": 2258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.170154377615514e-05,
+      "loss": 4.062,
+      "step": 2259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1693829270201955e-05,
+      "loss": 3.9658,
+      "step": 2260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.168611189438598e-05,
+      "loss": 4.0334,
+      "step": 2261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.167839165003394e-05,
+      "loss": 3.9308,
+      "step": 2262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.167066853847302e-05,
+      "loss": 4.1951,
+      "step": 2263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.16629425610309e-05,
+      "loss": 4.2178,
+      "step": 2264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.16552137190358e-05,
+      "loss": 4.1046,
+      "step": 2265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.164748201381637e-05,
+      "loss": 4.0002,
+      "step": 2266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1639747446701794e-05,
+      "loss": 3.9217,
+      "step": 2267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1632010019021715e-05,
+      "loss": 3.9223,
+      "step": 2268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.16242697321063e-05,
+      "loss": 4.0223,
+      "step": 2269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.16165265872862e-05,
+      "loss": 3.9447,
+      "step": 2270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1608780585892546e-05,
+      "loss": 3.9905,
+      "step": 2271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.160103172925696e-05,
+      "loss": 3.8536,
+      "step": 2272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.159328001871157e-05,
+      "loss": 4.1097,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.158552545558897e-05,
+      "loss": 3.9602,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.157776804122227e-05,
+      "loss": 3.9832,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1570007776945066e-05,
+      "loss": 3.895,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1562244664091415e-05,
+      "loss": 3.8866,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.15544787039959e-05,
+      "loss": 4.1498,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.15467098979936e-05,
+      "loss": 3.9643,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1538938247420025e-05,
+      "loss": 4.0207,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1531163753611236e-05,
+      "loss": 4.025,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.152338641790375e-05,
+      "loss": 4.0504,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.151560624163458e-05,
+      "loss": 3.982,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.150782322614124e-05,
+      "loss": 3.984,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.150003737276171e-05,
+      "loss": 4.0411,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1492248682834465e-05,
+      "loss": 4.0297,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1484457157698475e-05,
+      "loss": 4.2011,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.147666279869321e-05,
+      "loss": 4.0183,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.146886560715858e-05,
+      "loss": 4.1541,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.146106558443504e-05,
+      "loss": 4.0384,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1453262731863484e-05,
+      "loss": 3.9972,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.144545705078533e-05,
+      "loss": 4.1119,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1437648542542454e-05,
+      "loss": 3.8767,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1429837208477226e-05,
+      "loss": 4.0242,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1422023049932525e-05,
+      "loss": 3.9978,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.141420606825166e-05,
+      "loss": 4.0475,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1406386264778493e-05,
+      "loss": 4.0763,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1398563640857334e-05,
+      "loss": 4.0333,
+      "step": 2298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.139073819783298e-05,
+      "loss": 4.061,
+      "step": 2299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1382909937050694e-05,
+      "loss": 4.0244,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.137507885985628e-05,
+      "loss": 3.9193,
+      "step": 2301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1367244967595964e-05,
+      "loss": 4.0249,
+      "step": 2302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.13594082616165e-05,
+      "loss": 4.002,
+      "step": 2303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.135156874326509e-05,
+      "loss": 4.0097,
+      "step": 2304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.134372641388946e-05,
+      "loss": 4.0061,
+      "step": 2305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1335881274837776e-05,
+      "loss": 4.1068,
+      "step": 2306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.132803332745873e-05,
+      "loss": 3.9589,
+      "step": 2307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1320182573101464e-05,
+      "loss": 3.8755,
+      "step": 2308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.131232901311561e-05,
+      "loss": 4.0458,
+      "step": 2309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.130447264885128e-05,
+      "loss": 3.8971,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1296613481659095e-05,
+      "loss": 3.9809,
+      "step": 2311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1288751512890135e-05,
+      "loss": 4.0457,
+      "step": 2312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.128088674389594e-05,
+      "loss": 3.9844,
+      "step": 2313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.127301917602857e-05,
+      "loss": 4.0417,
+      "step": 2314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.126514881064055e-05,
+      "loss": 4.0835,
+      "step": 2315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.12572756490849e-05,
+      "loss": 4.1202,
+      "step": 2316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.124939969271509e-05,
+      "loss": 3.8493,
+      "step": 2317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.124152094288509e-05,
+      "loss": 3.9877,
+      "step": 2318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1233639400949345e-05,
+      "loss": 4.1268,
+      "step": 2319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.12257550682628e-05,
+      "loss": 4.1182,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.121786794618086e-05,
+      "loss": 4.0297,
+      "step": 2321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.120997803605939e-05,
+      "loss": 3.9609,
+      "step": 2322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.120208533925479e-05,
+      "loss": 4.1254,
+      "step": 2323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.119418985712388e-05,
+      "loss": 4.0165,
+      "step": 2324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.118629159102399e-05,
+      "loss": 4.14,
+      "step": 2325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.117839054231294e-05,
+      "loss": 3.9489,
+      "step": 2326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.117048671234899e-05,
+      "loss": 3.9607,
+      "step": 2327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1162580102490915e-05,
+      "loss": 3.8879,
+      "step": 2328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.115467071409794e-05,
+      "loss": 3.987,
+      "step": 2329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1146758548529804e-05,
+      "loss": 3.9462,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.113884360714667e-05,
+      "loss": 3.9069,
+      "step": 2331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.113092589130923e-05,
+      "loss": 3.9834,
+      "step": 2332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.112300540237862e-05,
+      "loss": 3.9918,
+      "step": 2333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1115082141716465e-05,
+      "loss": 4.1226,
+      "step": 2334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1107156110684876e-05,
+      "loss": 4.0328,
+      "step": 2335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1099227310646415e-05,
+      "loss": 4.018,
+      "step": 2336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1091295742964145e-05,
+      "loss": 3.9185,
+      "step": 2337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.108336140900158e-05,
+      "loss": 4.0029,
+      "step": 2338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1075424310122745e-05,
+      "loss": 3.886,
+      "step": 2339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1067484447692104e-05,
+      "loss": 4.0685,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.105954182307462e-05,
+      "loss": 4.0766,
+      "step": 2341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.105159643763572e-05,
+      "loss": 3.9781,
+      "step": 2342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.10436482927413e-05,
+      "loss": 4.02,
+      "step": 2343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.103569738975774e-05,
+      "loss": 3.9495,
+      "step": 2344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.102774373005191e-05,
+      "loss": 3.945,
+      "step": 2345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1019787314991124e-05,
+      "loss": 4.0933,
+      "step": 2346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.101182814594316e-05,
+      "loss": 3.9914,
+      "step": 2347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1003866224276336e-05,
+      "loss": 4.0366,
+      "step": 2348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.099590155135937e-05,
+      "loss": 4.0949,
+      "step": 2349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0987934128561476e-05,
+      "loss": 3.9852,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0979963957252374e-05,
+      "loss": 3.8914,
+      "step": 2351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0971991038802195e-05,
+      "loss": 3.9251,
+      "step": 2352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.09640153745816e-05,
+      "loss": 4.0728,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0956036965961697e-05,
+      "loss": 4.0954,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0948055814314055e-05,
+      "loss": 4.0337,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.094007192101074e-05,
+      "loss": 4.0274,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.093208528742426e-05,
+      "loss": 3.9875,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0924095914927615e-05,
+      "loss": 3.984,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.091610380489429e-05,
+      "loss": 4.0125,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.09081089586982e-05,
+      "loss": 4.1112,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0900111377713745e-05,
+      "loss": 4.1206,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.089211106331583e-05,
+      "loss": 3.9609,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.088410801687977e-05,
+      "loss": 4.0813,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.087610223978141e-05,
+      "loss": 3.9321,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.086809373339703e-05,
+      "loss": 4.0519,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.086008249910336e-05,
+      "loss": 4.1293,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0852068538277655e-05,
+      "loss": 4.0333,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.084405185229759e-05,
+      "loss": 4.0344,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0836032442541336e-05,
+      "loss": 4.0879,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.082801031038751e-05,
+      "loss": 4.0343,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.081998545721522e-05,
+      "loss": 3.9399,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.081195788440404e-05,
+      "loss": 4.1394,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.080392759333398e-05,
+      "loss": 3.8671,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.079589458538556e-05,
+      "loss": 4.0657,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.078785886193973e-05,
+      "loss": 3.9488,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0779820424377933e-05,
+      "loss": 3.9929,
+      "step": 2376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.077177927408208e-05,
+      "loss": 4.1683,
+      "step": 2377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0763735412434525e-05,
+      "loss": 4.1171,
+      "step": 2378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.07556888408181e-05,
+      "loss": 3.9525,
+      "step": 2379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0747639560616104e-05,
+      "loss": 4.0331,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.073958757321231e-05,
+      "loss": 4.003,
+      "step": 2381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.073153287999094e-05,
+      "loss": 4.1316,
+      "step": 2382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.072347548233671e-05,
+      "loss": 4.0582,
+      "step": 2383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0715415381634734e-05,
+      "loss": 3.9931,
+      "step": 2384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.070735257927068e-05,
+      "loss": 3.8935,
+      "step": 2385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.069928707663062e-05,
+      "loss": 3.9822,
+      "step": 2386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.069121887510111e-05,
+      "loss": 3.9506,
+      "step": 2387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0683147976069166e-05,
+      "loss": 3.8885,
+      "step": 2388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.067507438092227e-05,
+      "loss": 3.9936,
+      "step": 2389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0666998091048374e-05,
+      "loss": 4.0269,
+      "step": 2390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0658919107835876e-05,
+      "loss": 3.9875,
+      "step": 2391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.065083743267365e-05,
+      "loss": 4.0329,
+      "step": 2392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0642753066951024e-05,
+      "loss": 4.041,
+      "step": 2393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.06346660120578e-05,
+      "loss": 4.0444,
+      "step": 2394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.062657626938424e-05,
+      "loss": 3.9424,
+      "step": 2395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0618483840321056e-05,
+      "loss": 3.8889,
+      "step": 2396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.061038872625944e-05,
+      "loss": 3.9966,
+      "step": 2397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.060229092859103e-05,
+      "loss": 3.9716,
+      "step": 2398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.059419044870792e-05,
+      "loss": 3.8737,
+      "step": 2399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.058608728800271e-05,
+      "loss": 3.9689,
+      "step": 2400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.05779814478684e-05,
+      "loss": 4.0282,
+      "step": 2401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.056987292969847e-05,
+      "loss": 3.9443,
+      "step": 2402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.056176173488689e-05,
+      "loss": 3.9992,
+      "step": 2403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.055364786482806e-05,
+      "loss": 3.9279,
+      "step": 2404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.054553132091685e-05,
+      "loss": 4.0451,
+      "step": 2405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.053741210454859e-05,
+      "loss": 3.8433,
+      "step": 2406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.052929021711905e-05,
+      "loss": 4.1182,
+      "step": 2407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.052116566002451e-05,
+      "loss": 4.1806,
+      "step": 2408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0513038434661646e-05,
+      "loss": 3.9977,
+      "step": 2409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0504908542427635e-05,
+      "loss": 3.8967,
+      "step": 2410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.049677598472009e-05,
+      "loss": 3.9901,
+      "step": 2411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0488640762937104e-05,
+      "loss": 4.0892,
+      "step": 2412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0480502878477214e-05,
+      "loss": 3.9716,
+      "step": 2413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.047236233273941e-05,
+      "loss": 4.1297,
+      "step": 2414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.046421912712315e-05,
+      "loss": 4.0965,
+      "step": 2415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.045607326302835e-05,
+      "loss": 4.0088,
+      "step": 2416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.044792474185537e-05,
+      "loss": 4.0705,
+      "step": 2417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.043977356500503e-05,
+      "loss": 3.9461,
+      "step": 2418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.043161973387864e-05,
+      "loss": 3.9393,
+      "step": 2419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.042346324987791e-05,
+      "loss": 3.8314,
+      "step": 2420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.041530411440504e-05,
+      "loss": 4.0657,
+      "step": 2421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.040714232886269e-05,
+      "loss": 4.044,
+      "step": 2422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0398977894653954e-05,
+      "loss": 3.9514,
+      "step": 2423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.03908108131824e-05,
+      "loss": 4.1012,
+      "step": 2424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.038264108585204e-05,
+      "loss": 4.0,
+      "step": 2425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0374468714067345e-05,
+      "loss": 3.9835,
+      "step": 2426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.036629369923325e-05,
+      "loss": 3.9907,
+      "step": 2427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.035811604275512e-05,
+      "loss": 3.9199,
+      "step": 2428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.034993574603879e-05,
+      "loss": 4.0577,
+      "step": 2429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0341752810490566e-05,
+      "loss": 3.9226,
+      "step": 2430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0333567237517165e-05,
+      "loss": 3.9824,
+      "step": 2431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.03253790285258e-05,
+      "loss": 4.0151,
+      "step": 2432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.031718818492411e-05,
+      "loss": 3.9519,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.030899470812021e-05,
+      "loss": 3.9568,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.030079859952263e-05,
+      "loss": 3.9211,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.029259986054039e-05,
+      "loss": 4.0459,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0284398492582955e-05,
+      "loss": 4.0176,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.027619449706023e-05,
+      "loss": 4.0325,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.026798787538256e-05,
+      "loss": 3.9146,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0259778628960775e-05,
+      "loss": 3.8974,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.025156675920614e-05,
+      "loss": 4.0489,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.024335226753037e-05,
+      "loss": 3.9787,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0235135155345625e-05,
+      "loss": 3.8674,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.022691542406453e-05,
+      "loss": 4.044,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0218693075100146e-05,
+      "loss": 3.9713,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0210468109866e-05,
+      "loss": 3.9982,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.020224052977604e-05,
+      "loss": 4.1073,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.01940103362447e-05,
+      "loss": 3.9844,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.018577753068683e-05,
+      "loss": 3.9327,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0177542114517765e-05,
+      "loss": 4.1052,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.016930408915325e-05,
+      "loss": 4.0251,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.016106345600951e-05,
+      "loss": 3.8723,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0152820216503194e-05,
+      "loss": 4.0796,
+      "step": 2453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.014457437205142e-05,
+      "loss": 4.05,
+      "step": 2454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.013632592407174e-05,
+      "loss": 3.8259,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0128074873982156e-05,
+      "loss": 3.9702,
+      "step": 2456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.011982122320113e-05,
+      "loss": 3.8826,
+      "step": 2457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.011156497314754e-05,
+      "loss": 3.8481,
+      "step": 2458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.010330612524075e-05,
+      "loss": 4.1546,
+      "step": 2459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.009504468090054e-05,
+      "loss": 4.0407,
+      "step": 2460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0086780641547155e-05,
+      "loss": 4.0721,
+      "step": 2461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.007851400860128e-05,
+      "loss": 3.9862,
+      "step": 2462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.007024478348403e-05,
+      "loss": 3.9038,
+      "step": 2463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.006197296761701e-05,
+      "loss": 3.934,
+      "step": 2464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.005369856242221e-05,
+      "loss": 4.0438,
+      "step": 2465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0045421569322115e-05,
+      "loss": 3.9938,
+      "step": 2466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.003714198973963e-05,
+      "loss": 4.0774,
+      "step": 2467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.002885982509811e-05,
+      "loss": 3.9224,
+      "step": 2468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.002057507682136e-05,
+      "loss": 4.1385,
+      "step": 2469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.001228774633362e-05,
+      "loss": 4.0157,
+      "step": 2470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.000399783505958e-05,
+      "loss": 4.011,
+      "step": 2471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999570534442436e-05,
+      "loss": 3.9913,
+      "step": 2472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.998741027585356e-05,
+      "loss": 4.0256,
+      "step": 2473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.997911263077318e-05,
+      "loss": 3.9869,
+      "step": 2474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9970812410609683e-05,
+      "loss": 4.0654,
+      "step": 2475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9962509616789985e-05,
+      "loss": 3.833,
+      "step": 2476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.995420425074141e-05,
+      "loss": 3.9215,
+      "step": 2477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.994589631389177e-05,
+      "loss": 3.8649,
+      "step": 2478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.993758580766929e-05,
+      "loss": 3.9458,
+      "step": 2479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.992927273350263e-05,
+      "loss": 4.0445,
+      "step": 2480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.992095709282091e-05,
+      "loss": 3.9512,
+      "step": 2481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.991263888705369e-05,
+      "loss": 4.0518,
+      "step": 2482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.990431811763097e-05,
+      "loss": 4.1869,
+      "step": 2483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.989599478598317e-05,
+      "loss": 3.9221,
+      "step": 2484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.988766889354118e-05,
+      "loss": 3.999,
+      "step": 2485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9879340441736314e-05,
+      "loss": 4.0878,
+      "step": 2486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9871009432000325e-05,
+      "loss": 4.137,
+      "step": 2487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.986267586576543e-05,
+      "loss": 4.058,
+      "step": 2488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.985433974446424e-05,
+      "loss": 4.0028,
+      "step": 2489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9846001069529845e-05,
+      "loss": 3.8963,
+      "step": 2490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.983765984239576e-05,
+      "loss": 4.0695,
+      "step": 2491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9829316064495927e-05,
+      "loss": 4.0957,
+      "step": 2492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.982096973726475e-05,
+      "loss": 4.0015,
+      "step": 2493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9812620862137056e-05,
+      "loss": 3.8944,
+      "step": 2494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9804269440548114e-05,
+      "loss": 3.8501,
+      "step": 2495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.979591547393363e-05,
+      "loss": 4.0651,
+      "step": 2496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9787558963729744e-05,
+      "loss": 3.9094,
+      "step": 2497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.977919991137304e-05,
+      "loss": 4.0215,
+      "step": 2498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.977083831830054e-05,
+      "loss": 3.9588,
+      "step": 2499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.976247418594969e-05,
+      "loss": 3.9731,
+      "step": 2500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.975410751575839e-05,
+      "loss": 4.2002,
+      "step": 2501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.974573830916496e-05,
+      "loss": 4.0774,
+      "step": 2502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.973736656760817e-05,
+      "loss": 4.1509,
+      "step": 2503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.972899229252721e-05,
+      "loss": 3.8998,
+      "step": 2504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.972061548536173e-05,
+      "loss": 4.1663,
+      "step": 2505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.97122361475518e-05,
+      "loss": 3.8962,
+      "step": 2506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.97038542805379e-05,
+      "loss": 4.0651,
+      "step": 2507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9695469885761e-05,
+      "loss": 3.9931,
+      "step": 2508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.968708296466245e-05,
+      "loss": 3.987,
+      "step": 2509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9678693518684085e-05,
+      "loss": 4.0049,
+      "step": 2510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.967030154926813e-05,
+      "loss": 3.9862,
+      "step": 2511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.966190705785726e-05,
+      "loss": 3.948,
+      "step": 2512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.965351004589459e-05,
+      "loss": 4.0564,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9645110514823676e-05,
+      "loss": 3.9995,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.963670846608848e-05,
+      "loss": 4.1633,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.962830390113341e-05,
+      "loss": 4.0183,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.961989682140332e-05,
+      "loss": 3.9584,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.961148722834347e-05,
+      "loss": 3.9625,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.960307512339958e-05,
+      "loss": 4.0165,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.959466050801778e-05,
+      "loss": 4.1305,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.958624338364464e-05,
+      "loss": 4.0612,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.957782375172718e-05,
+      "loss": 4.2473,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.95694016137128e-05,
+      "loss": 4.073,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.956097697104938e-05,
+      "loss": 3.9635,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.955254982518523e-05,
+      "loss": 3.9942,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.954412017756905e-05,
+      "loss": 3.9731,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9535688029649984e-05,
+      "loss": 3.9699,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.952725338287765e-05,
+      "loss": 4.0129,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.951881623870204e-05,
+      "loss": 4.0386,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.951037659857361e-05,
+      "loss": 4.0834,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.950193446394321e-05,
+      "loss": 4.1521,
+      "step": 2531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9493489836262165e-05,
+      "loss": 4.027,
+      "step": 2532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.948504271698219e-05,
+      "loss": 4.0898,
+      "step": 2533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9476593107555446e-05,
+      "loss": 4.0042,
+      "step": 2534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.946814100943453e-05,
+      "loss": 4.0894,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.945968642407244e-05,
+      "loss": 3.9374,
+      "step": 2536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.945122935292261e-05,
+      "loss": 3.9464,
+      "step": 2537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.944276979743894e-05,
+      "loss": 4.1095,
+      "step": 2538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.943430775907569e-05,
+      "loss": 3.9341,
+      "step": 2539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9425843239287616e-05,
+      "loss": 4.0002,
+      "step": 2540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9417376239529846e-05,
+      "loss": 4.0351,
+      "step": 2541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.940890676125796e-05,
+      "loss": 3.9905,
+      "step": 2542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.940043480592795e-05,
+      "loss": 4.0879,
+      "step": 2543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.939196037499626e-05,
+      "loss": 3.9574,
+      "step": 2544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.938348346991974e-05,
+      "loss": 4.0578,
+      "step": 2545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.937500409215565e-05,
+      "loss": 3.8804,
+      "step": 2546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9366522243161716e-05,
+      "loss": 3.9145,
+      "step": 2547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.935803792439606e-05,
+      "loss": 3.9426,
+      "step": 2548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9349551137317234e-05,
+      "loss": 4.071,
+      "step": 2549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.93410618833842e-05,
+      "loss": 4.1068,
+      "step": 2550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.933257016405638e-05,
+      "loss": 3.8902,
+      "step": 2551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.932407598079359e-05,
+      "loss": 3.9334,
+      "step": 2552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.931557933505608e-05,
+      "loss": 3.8769,
+      "step": 2553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9307080228304524e-05,
+      "loss": 4.1245,
+      "step": 2554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.929857866200002e-05,
+      "loss": 3.8949,
+      "step": 2555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9290074637604065e-05,
+      "loss": 4.0261,
+      "step": 2556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.928156815657863e-05,
+      "loss": 4.0112,
+      "step": 2557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.927305922038606e-05,
+      "loss": 4.1421,
+      "step": 2558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.926454783048914e-05,
+      "loss": 3.9642,
+      "step": 2559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.925603398835108e-05,
+      "loss": 3.8941,
+      "step": 2560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.924751769543551e-05,
+      "loss": 3.8678,
+      "step": 2561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.923899895320648e-05,
+      "loss": 3.9093,
+      "step": 2562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9230477763128446e-05,
+      "loss": 4.0013,
+      "step": 2563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9221954126666316e-05,
+      "loss": 4.0887,
+      "step": 2564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9213428045285386e-05,
+      "loss": 3.9528,
+      "step": 2565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.920489952045141e-05,
+      "loss": 3.9334,
+      "step": 2566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9196368553630535e-05,
+      "loss": 3.9206,
+      "step": 2567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.918783514628932e-05,
+      "loss": 4.0183,
+      "step": 2568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9179299299894756e-05,
+      "loss": 3.9048,
+      "step": 2569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.917076101591427e-05,
+      "loss": 4.1194,
+      "step": 2570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.916222029581569e-05,
+      "loss": 4.0218,
+      "step": 2571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9153677141067254e-05,
+      "loss": 3.88,
+      "step": 2572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.914513155313763e-05,
+      "loss": 4.0784,
+      "step": 2573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.913658353349591e-05,
+      "loss": 3.9388,
+      "step": 2574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.912803308361159e-05,
+      "loss": 4.0808,
+      "step": 2575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9119480204954606e-05,
+      "loss": 3.9813,
+      "step": 2576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9110924898995285e-05,
+      "loss": 4.031,
+      "step": 2577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.910236716720438e-05,
+      "loss": 3.9379,
+      "step": 2578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9093807011053076e-05,
+      "loss": 4.0559,
+      "step": 2579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9085244432012956e-05,
+      "loss": 3.8663,
+      "step": 2580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.907667943155603e-05,
+      "loss": 3.9931,
+      "step": 2581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.906811201115472e-05,
+      "loss": 4.0304,
+      "step": 2582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.905954217228186e-05,
+      "loss": 3.876,
+      "step": 2583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.905096991641071e-05,
+      "loss": 3.9445,
+      "step": 2584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9042395245014936e-05,
+      "loss": 4.0523,
+      "step": 2585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.903381815956862e-05,
+      "loss": 3.9492,
+      "step": 2586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.902523866154628e-05,
+      "loss": 3.9136,
+      "step": 2587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.901665675242281e-05,
+      "loss": 3.9616,
+      "step": 2588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.900807243367356e-05,
+      "loss": 3.8675,
+      "step": 2589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.899948570677426e-05,
+      "loss": 3.9144,
+      "step": 2590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.899089657320107e-05,
+      "loss": 3.9793,
+      "step": 2591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8982305034430556e-05,
+      "loss": 3.8605,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.897371109193972e-05,
+      "loss": 4.016,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.896511474720594e-05,
+      "loss": 3.951,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.895651600170705e-05,
+      "loss": 4.1466,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8947914856921256e-05,
+      "loss": 3.9791,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.893931131432719e-05,
+      "loss": 3.9273,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.893070537540393e-05,
+      "loss": 4.0022,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.89220970416309e-05,
+      "loss": 3.9857,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8913486314488e-05,
+      "loss": 3.9638,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.89048731954555e-05,
+      "loss": 4.0134,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.889625768601409e-05,
+      "loss": 3.9082,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88876397876449e-05,
+      "loss": 4.0254,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8879019501829423e-05,
+      "loss": 3.9904,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.887039683004959e-05,
+      "loss": 4.0127,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.886177177378775e-05,
+      "loss": 3.9881,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.885314433452664e-05,
+      "loss": 3.9437,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8844514513749425e-05,
+      "loss": 3.8951,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.883588231293966e-05,
+      "loss": 3.9882,
+      "step": 2609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.882724773358133e-05,
+      "loss": 3.8557,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88186107771588e-05,
+      "loss": 3.9958,
+      "step": 2611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88099714451569e-05,
+      "loss": 3.9852,
+      "step": 2612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88013297390608e-05,
+      "loss": 3.9906,
+      "step": 2613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.879268566035612e-05,
+      "loss": 3.9286,
+      "step": 2614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8784039210528874e-05,
+      "loss": 3.9916,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.877539039106549e-05,
+      "loss": 3.9688,
+      "step": 2616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8766739203452804e-05,
+      "loss": 4.01,
+      "step": 2617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.875808564917804e-05,
+      "loss": 4.0137,
+      "step": 2618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.874942972972887e-05,
+      "loss": 3.9916,
+      "step": 2619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8740771446593325e-05,
+      "loss": 4.0959,
+      "step": 2620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.873211080125987e-05,
+      "loss": 3.9526,
+      "step": 2621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.872344779521738e-05,
+      "loss": 4.0169,
+      "step": 2622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.871478242995511e-05,
+      "loss": 3.9258,
+      "step": 2623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.870611470696275e-05,
+      "loss": 3.8121,
+      "step": 2624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.869744462773037e-05,
+      "loss": 4.0563,
+      "step": 2625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.868877219374847e-05,
+      "loss": 3.9595,
+      "step": 2626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.868009740650794e-05,
+      "loss": 4.036,
+      "step": 2627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.867142026750007e-05,
+      "loss": 3.989,
+      "step": 2628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8662740778216556e-05,
+      "loss": 4.0898,
+      "step": 2629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8654058940149514e-05,
+      "loss": 3.9692,
+      "step": 2630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.864537475479145e-05,
+      "loss": 4.0561,
+      "step": 2631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8636688223635275e-05,
+      "loss": 4.0797,
+      "step": 2632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.86279993481743e-05,
+      "loss": 3.9211,
+      "step": 2633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.861930812990225e-05,
+      "loss": 4.0625,
+      "step": 2634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8610614570313244e-05,
+      "loss": 3.9245,
+      "step": 2635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.860191867090181e-05,
+      "loss": 4.0171,
+      "step": 2636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.859322043316287e-05,
+      "loss": 4.0063,
+      "step": 2637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.858451985859175e-05,
+      "loss": 4.0303,
+      "step": 2638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.857581694868417e-05,
+      "loss": 3.9769,
+      "step": 2639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.856711170493629e-05,
+      "loss": 3.8314,
+      "step": 2640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8558404128844605e-05,
+      "loss": 3.9529,
+      "step": 2641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.854969422190608e-05,
+      "loss": 4.0267,
+      "step": 2642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.854098198561804e-05,
+      "loss": 3.9314,
+      "step": 2643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.853226742147821e-05,
+      "loss": 4.0589,
+      "step": 2644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.852355053098473e-05,
+      "loss": 3.9559,
+      "step": 2645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.851483131563613e-05,
+      "loss": 3.9627,
+      "step": 2646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.850610977693137e-05,
+      "loss": 3.9351,
+      "step": 2647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8497385916369744e-05,
+      "loss": 4.0261,
+      "step": 2648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8488659735451004e-05,
+      "loss": 3.953,
+      "step": 2649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.847993123567529e-05,
+      "loss": 3.9981,
+      "step": 2650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.847120041854312e-05,
+      "loss": 4.01,
+      "step": 2651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.846246728555542e-05,
+      "loss": 4.0771,
+      "step": 2652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.845373183821354e-05,
+      "loss": 3.9237,
+      "step": 2653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.844499407801918e-05,
+      "loss": 3.9294,
+      "step": 2654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8436254006474467e-05,
+      "loss": 4.1266,
+      "step": 2655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8427511625081924e-05,
+      "loss": 4.0774,
+      "step": 2656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8418766935344466e-05,
+      "loss": 4.1343,
+      "step": 2657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8410019938765415e-05,
+      "loss": 4.023,
+      "step": 2658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8401270636848476e-05,
+      "loss": 4.0536,
+      "step": 2659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8392519031097745e-05,
+      "loss": 3.9264,
+      "step": 2660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.838376512301773e-05,
+      "loss": 3.893,
+      "step": 2661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.837500891411334e-05,
+      "loss": 4.1039,
+      "step": 2662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.836625040588986e-05,
+      "loss": 4.0231,
+      "step": 2663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.835748959985299e-05,
+      "loss": 3.9604,
+      "step": 2664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.834872649750879e-05,
+      "loss": 3.8627,
+      "step": 2665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.833996110036375e-05,
+      "loss": 3.8998,
+      "step": 2666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.833119340992476e-05,
+      "loss": 4.0249,
+      "step": 2667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8322423427699075e-05,
+      "loss": 4.0033,
+      "step": 2668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8313651155194344e-05,
+      "loss": 3.9575,
+      "step": 2669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8304876593918646e-05,
+      "loss": 3.9915,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8296099745380406e-05,
+      "loss": 4.0102,
+      "step": 2671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.828732061108849e-05,
+      "loss": 3.9588,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.827853919255211e-05,
+      "loss": 3.9009,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.826975549128092e-05,
+      "loss": 3.908,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8260969508784916e-05,
+      "loss": 3.9611,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.825218124657452e-05,
+      "loss": 3.8864,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.824339070616053e-05,
+      "loss": 4.0485,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.823459788905415e-05,
+      "loss": 4.0591,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8225802796766964e-05,
+      "loss": 4.1376,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.821700543081095e-05,
+      "loss": 4.0251,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.820820579269848e-05,
+      "loss": 3.885,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.819940388394232e-05,
+      "loss": 4.0545,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.81905997060556e-05,
+      "loss": 3.9743,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8181793260551876e-05,
+      "loss": 4.0571,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.817298454894507e-05,
+      "loss": 3.8955,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8164173572749505e-05,
+      "loss": 3.9323,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8155360333479895e-05,
+      "loss": 4.1319,
+      "step": 2687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8146544832651336e-05,
+      "loss": 3.8546,
+      "step": 2688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8137727071779306e-05,
+      "loss": 4.0987,
+      "step": 2689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.812890705237969e-05,
+      "loss": 4.0615,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.812008477596875e-05,
+      "loss": 3.9849,
+      "step": 2691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.811126024406314e-05,
+      "loss": 3.9156,
+      "step": 2692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.810243345817989e-05,
+      "loss": 4.0447,
+      "step": 2693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.809360441983643e-05,
+      "loss": 3.922,
+      "step": 2694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.808477313055058e-05,
+      "loss": 3.8884,
+      "step": 2695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8075939591840534e-05,
+      "loss": 4.0968,
+      "step": 2696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.806710380522488e-05,
+      "loss": 3.8697,
+      "step": 2697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.80582657722226e-05,
+      "loss": 4.0153,
+      "step": 2698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.804942549435305e-05,
+      "loss": 4.0174,
+      "step": 2699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8040582973135974e-05,
+      "loss": 4.0636,
+      "step": 2700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.803173821009151e-05,
+      "loss": 3.9859,
+      "step": 2701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.802289120674016e-05,
+      "loss": 4.0625,
+      "step": 2702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.801404196460284e-05,
+      "loss": 4.0585,
+      "step": 2703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.800519048520084e-05,
+      "loss": 4.0886,
+      "step": 2704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.799633677005583e-05,
+      "loss": 4.0398,
+      "step": 2705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.798748082068986e-05,
+      "loss": 4.0639,
+      "step": 2706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.797862263862537e-05,
+      "loss": 4.0683,
+      "step": 2707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7969762225385174e-05,
+      "loss": 4.1093,
+      "step": 2708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.796089958249252e-05,
+      "loss": 3.8829,
+      "step": 2709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.795203471147095e-05,
+      "loss": 3.9615,
+      "step": 2710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7943167613844475e-05,
+      "loss": 3.9917,
+      "step": 2711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.793429829113743e-05,
+      "loss": 3.9347,
+      "step": 2712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.792542674487456e-05,
+      "loss": 4.0635,
+      "step": 2713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7916552976580996e-05,
+      "loss": 3.9933,
+      "step": 2714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.790767698778223e-05,
+      "loss": 4.0056,
+      "step": 2715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7898798780004145e-05,
+      "loss": 3.9309,
+      "step": 2716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.788991835477302e-05,
+      "loss": 3.7407,
+      "step": 2717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.78810357136155e-05,
+      "loss": 3.889,
+      "step": 2718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.78721508580586e-05,
+      "loss": 3.7296,
+      "step": 2719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7863263789629754e-05,
+      "loss": 4.0151,
+      "step": 2720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7854374509856734e-05,
+      "loss": 4.0299,
+      "step": 2721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.784548302026771e-05,
+      "loss": 4.1769,
+      "step": 2722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7836589322391236e-05,
+      "loss": 3.8507,
+      "step": 2723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7827693417756254e-05,
+      "loss": 3.9446,
+      "step": 2724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7818795307892056e-05,
+      "loss": 3.9834,
+      "step": 2725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7809894994328333e-05,
+      "loss": 3.8612,
+      "step": 2726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.780099247859516e-05,
+      "loss": 3.9776,
+      "step": 2727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.779208776222298e-05,
+      "loss": 3.905,
+      "step": 2728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.778318084674262e-05,
+      "loss": 3.9243,
+      "step": 2729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7774271733685264e-05,
+      "loss": 3.9267,
+      "step": 2730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.77653604245825e-05,
+      "loss": 4.1373,
+      "step": 2731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.77564469209663e-05,
+      "loss": 3.9657,
+      "step": 2732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7747531224368994e-05,
+      "loss": 4.0002,
+      "step": 2733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.773861333632328e-05,
+      "loss": 4.0818,
+      "step": 2734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.772969325836227e-05,
+      "loss": 3.971,
+      "step": 2735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.772077099201939e-05,
+      "loss": 3.9838,
+      "step": 2736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.771184653882852e-05,
+      "loss": 4.1253,
+      "step": 2737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7702919900323856e-05,
+      "loss": 4.0261,
+      "step": 2738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.769399107804e-05,
+      "loss": 3.9241,
+      "step": 2739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7685060073511916e-05,
+      "loss": 3.9601,
+      "step": 2740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.767612688827494e-05,
+      "loss": 3.9972,
+      "step": 2741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.766719152386481e-05,
+      "loss": 3.9924,
+      "step": 2742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.76582539818176e-05,
+      "loss": 4.0014,
+      "step": 2743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7649314263669786e-05,
+      "loss": 3.9702,
+      "step": 2744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.76403723709582e-05,
+      "loss": 4.0517,
+      "step": 2745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.763142830522007e-05,
+      "loss": 3.8925,
+      "step": 2746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.762248206799297e-05,
+      "loss": 4.0077,
+      "step": 2747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.761353366081488e-05,
+      "loss": 4.0804,
+      "step": 2748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.760458308522413e-05,
+      "loss": 4.1819,
+      "step": 2749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.759563034275941e-05,
+      "loss": 4.0255,
+      "step": 2750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7586675434959816e-05,
+      "loss": 4.0721,
+      "step": 2751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.75777183633648e-05,
+      "loss": 4.1275,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.756875912951418e-05,
+      "loss": 3.8726,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.755979773494816e-05,
+      "loss": 3.9127,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.75508341812073e-05,
+      "loss": 4.0499,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7541868469832547e-05,
+      "loss": 3.9383,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7532900602365205e-05,
+      "loss": 4.0243,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.752393058034696e-05,
+      "loss": 4.0235,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.751495840531984e-05,
+      "loss": 4.037,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.750598407882631e-05,
+      "loss": 3.9714,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.749700760240912e-05,
+      "loss": 3.8719,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7488028977611446e-05,
+      "loss": 4.0138,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.747904820597682e-05,
+      "loss": 3.8573,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7470065289049136e-05,
+      "loss": 3.9891,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.746108022837266e-05,
+      "loss": 4.0435,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.745209302549204e-05,
+      "loss": 3.9541,
+      "step": 2766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7443103681952266e-05,
+      "loss": 4.0572,
+      "step": 2767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.743411219929872e-05,
+      "loss": 3.8687,
+      "step": 2768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7425118579077135e-05,
+      "loss": 3.9368,
+      "step": 2769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.741612282283362e-05,
+      "loss": 4.1018,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7407124932114664e-05,
+      "loss": 3.9843,
+      "step": 2771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.739812490846708e-05,
+      "loss": 3.9267,
+      "step": 2772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.73891227534381e-05,
+      "loss": 3.8229,
+      "step": 2773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.738011846857529e-05,
+      "loss": 3.8431,
+      "step": 2774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.73711120554266e-05,
+      "loss": 4.0889,
+      "step": 2775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.736210351554032e-05,
+      "loss": 4.0454,
+      "step": 2776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.735309285046513e-05,
+      "loss": 4.1075,
+      "step": 2777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.734408006175008e-05,
+      "loss": 3.8301,
+      "step": 2778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7335065150944556e-05,
+      "loss": 3.9726,
+      "step": 2779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.732604811959834e-05,
+      "loss": 4.0032,
+      "step": 2780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.731702896926155e-05,
+      "loss": 3.8888,
+      "step": 2781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.730800770148469e-05,
+      "loss": 3.9591,
+      "step": 2782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.729898431781861e-05,
+      "loss": 4.1195,
+      "step": 2783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7289958819814555e-05,
+      "loss": 4.0131,
+      "step": 2784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7280931209024105e-05,
+      "loss": 4.0368,
+      "step": 2785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.72719014869992e-05,
+      "loss": 3.871,
+      "step": 2786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.726286965529216e-05,
+      "loss": 3.9173,
+      "step": 2787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7253835715455664e-05,
+      "loss": 4.0202,
+      "step": 2788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.724479966904276e-05,
+      "loss": 4.0063,
+      "step": 2789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7235761517606825e-05,
+      "loss": 3.92,
+      "step": 2790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7226721262701635e-05,
+      "loss": 3.9151,
+      "step": 2791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.721767890588133e-05,
+      "loss": 4.0707,
+      "step": 2792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.720863444870038e-05,
+      "loss": 3.9134,
+      "step": 2793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.719958789271362e-05,
+      "loss": 3.9292,
+      "step": 2794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.719053923947628e-05,
+      "loss": 3.8777,
+      "step": 2795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7181488490543915e-05,
+      "loss": 3.9257,
+      "step": 2796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.717243564747247e-05,
+      "loss": 4.0193,
+      "step": 2797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.716338071181821e-05,
+      "loss": 4.0295,
+      "step": 2798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7154323685137806e-05,
+      "loss": 4.0761,
+      "step": 2799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.714526456898824e-05,
+      "loss": 3.9154,
+      "step": 2800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.71362033649269e-05,
+      "loss": 3.8972,
+      "step": 2801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.712714007451152e-05,
+      "loss": 4.0318,
+      "step": 2802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7118074699300164e-05,
+      "loss": 3.9615,
+      "step": 2803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7109007240851277e-05,
+      "loss": 3.9014,
+      "step": 2804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.709993770072366e-05,
+      "loss": 3.8651,
+      "step": 2805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.709086608047649e-05,
+      "loss": 3.9565,
+      "step": 2806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.708179238166928e-05,
+      "loss": 3.9122,
+      "step": 2807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7072716605861876e-05,
+      "loss": 3.893,
+      "step": 2808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.706363875461454e-05,
+      "loss": 3.9512,
+      "step": 2809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.705455882948784e-05,
+      "loss": 3.9704,
+      "step": 2810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.704547683204274e-05,
+      "loss": 3.8804,
+      "step": 2811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.703639276384052e-05,
+      "loss": 4.03,
+      "step": 2812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.702730662644285e-05,
+      "loss": 3.8868,
+      "step": 2813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.701821842141173e-05,
+      "loss": 3.9825,
+      "step": 2814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.700912815030955e-05,
+      "loss": 3.9622,
+      "step": 2815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.700003581469901e-05,
+      "loss": 4.0817,
+      "step": 2816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6990941416143205e-05,
+      "loss": 4.0,
+      "step": 2817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.698184495620555e-05,
+      "loss": 4.0224,
+      "step": 2818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6972746436449845e-05,
+      "loss": 3.9956,
+      "step": 2819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6963645858440234e-05,
+      "loss": 4.0856,
+      "step": 2820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.69545432237412e-05,
+      "loss": 3.9862,
+      "step": 2821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6945438533917594e-05,
+      "loss": 3.93,
+      "step": 2822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.693633179053462e-05,
+      "loss": 3.9512,
+      "step": 2823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.692722299515784e-05,
+      "loss": 3.9795,
+      "step": 2824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.691811214935315e-05,
+      "loss": 3.9,
+      "step": 2825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6908999254686814e-05,
+      "loss": 4.1841,
+      "step": 2826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.689988431272545e-05,
+      "loss": 3.9574,
+      "step": 2827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.689076732503601e-05,
+      "loss": 4.0384,
+      "step": 2828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.688164829318583e-05,
+      "loss": 3.9713,
+      "step": 2829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6872527218742556e-05,
+      "loss": 3.8727,
+      "step": 2830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6863404103274215e-05,
+      "loss": 3.9423,
+      "step": 2831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6854278948349184e-05,
+      "loss": 3.9749,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.684515175553616e-05,
+      "loss": 3.9431,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.683602252640424e-05,
+      "loss": 3.9504,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.682689126252283e-05,
+      "loss": 3.9472,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.681775796546171e-05,
+      "loss": 4.0644,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.680862263679097e-05,
+      "loss": 3.9818,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6799485278081125e-05,
+      "loss": 4.042,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.679034589090296e-05,
+      "loss": 3.9189,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6781204476827654e-05,
+      "loss": 3.7873,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.677206103742671e-05,
+      "loss": 4.0031,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.676291557427201e-05,
+      "loss": 3.9302,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6753768088935755e-05,
+      "loss": 4.1128,
+      "step": 2843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.67446185829905e-05,
+      "loss": 3.9094,
+      "step": 2844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.673546705800916e-05,
+      "loss": 4.0072,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6726313515564984e-05,
+      "loss": 3.8986,
+      "step": 2846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6717157957231565e-05,
+      "loss": 4.0623,
+      "step": 2847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.670800038458285e-05,
+      "loss": 3.8196,
+      "step": 2848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6698840799193154e-05,
+      "loss": 4.0664,
+      "step": 2849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.668967920263709e-05,
+      "loss": 4.0897,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.668051559648965e-05,
+      "loss": 4.0476,
+      "step": 2851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6671349982326176e-05,
+      "loss": 3.9673,
+      "step": 2852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.666218236172234e-05,
+      "loss": 3.864,
+      "step": 2853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.665301273625414e-05,
+      "loss": 4.1229,
+      "step": 2854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.664384110749797e-05,
+      "loss": 3.9465,
+      "step": 2855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.663466747703052e-05,
+      "loss": 3.892,
+      "step": 2856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.662549184642885e-05,
+      "loss": 3.925,
+      "step": 2857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.661631421727036e-05,
+      "loss": 3.8764,
+      "step": 2858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6607134591132785e-05,
+      "loss": 3.8689,
+      "step": 2859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6597952969594217e-05,
+      "loss": 4.042,
+      "step": 2860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6588769354233075e-05,
+      "loss": 3.924,
+      "step": 2861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6579583746628126e-05,
+      "loss": 3.9844,
+      "step": 2862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.65703961483585e-05,
+      "loss": 3.8353,
+      "step": 2863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.656120656100364e-05,
+      "loss": 4.0779,
+      "step": 2864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.655201498614333e-05,
+      "loss": 4.0331,
+      "step": 2865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.654282142535773e-05,
+      "loss": 3.9594,
+      "step": 2866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6533625880227315e-05,
+      "loss": 3.9086,
+      "step": 2867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6524428352332904e-05,
+      "loss": 3.8725,
+      "step": 2868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.651522884325565e-05,
+      "loss": 3.9298,
+      "step": 2869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.650602735457706e-05,
+      "loss": 3.8985,
+      "step": 2870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6496823887878986e-05,
+      "loss": 4.0294,
+      "step": 2871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6487618444743596e-05,
+      "loss": 3.9508,
+      "step": 2872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6478411026753424e-05,
+      "loss": 3.8588,
+      "step": 2873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.646920163549132e-05,
+      "loss": 3.953,
+      "step": 2874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.645999027254051e-05,
+      "loss": 3.9332,
+      "step": 2875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.64507769394845e-05,
+      "loss": 4.0018,
+      "step": 2876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.644156163790719e-05,
+      "loss": 3.9852,
+      "step": 2877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.64323443693928e-05,
+      "loss": 4.0011,
+      "step": 2878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.642312513552586e-05,
+      "loss": 3.9036,
+      "step": 2879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6413903937891294e-05,
+      "loss": 3.9077,
+      "step": 2880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.640468077807432e-05,
+      "loss": 3.8989,
+      "step": 2881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.63954556576605e-05,
+      "loss": 3.8936,
+      "step": 2882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6386228578235746e-05,
+      "loss": 4.0425,
+      "step": 2883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6376999541386304e-05,
+      "loss": 3.9697,
+      "step": 2884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.636776854869874e-05,
+      "loss": 3.9653,
+      "step": 2885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6358535601759985e-05,
+      "loss": 3.9498,
+      "step": 2886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.634930070215727e-05,
+      "loss": 4.0242,
+      "step": 2887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.63400638514782e-05,
+      "loss": 3.9089,
+      "step": 2888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.633082505131069e-05,
+      "loss": 4.0023,
+      "step": 2889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.632158430324299e-05,
+      "loss": 3.8864,
+      "step": 2890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6312341608863695e-05,
+      "loss": 3.8112,
+      "step": 2891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6303096969761744e-05,
+      "loss": 4.0285,
+      "step": 2892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.629385038752638e-05,
+      "loss": 4.0398,
+      "step": 2893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.62846018637472e-05,
+      "loss": 3.9029,
+      "step": 2894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.627535140001414e-05,
+      "loss": 3.8734,
+      "step": 2895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.626609899791748e-05,
+      "loss": 3.9635,
+      "step": 2896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.625684465904777e-05,
+      "loss": 3.962,
+      "step": 2897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.624758838499596e-05,
+      "loss": 4.1202,
+      "step": 2898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.623833017735333e-05,
+      "loss": 3.9036,
+      "step": 2899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.622907003771145e-05,
+      "loss": 3.8918,
+      "step": 2900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.621980796766225e-05,
+      "loss": 4.06,
+      "step": 2901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6210543968797985e-05,
+      "loss": 3.9516,
+      "step": 2902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.620127804271126e-05,
+      "loss": 4.0262,
+      "step": 2903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6192010190994975e-05,
+      "loss": 4.0872,
+      "step": 2904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.618274041524239e-05,
+      "loss": 4.0403,
+      "step": 2905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.617346871704709e-05,
+      "loss": 3.9649,
+      "step": 2906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6164195098002974e-05,
+      "loss": 4.0447,
+      "step": 2907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.61549195597043e-05,
+      "loss": 3.9179,
+      "step": 2908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6145642103745635e-05,
+      "loss": 3.9402,
+      "step": 2909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6136362731721884e-05,
+      "loss": 3.9462,
+      "step": 2910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6127081445228276e-05,
+      "loss": 4.0626,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.611779824586036e-05,
+      "loss": 3.8226,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.610851313521405e-05,
+      "loss": 4.0872,
+      "step": 2913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.609922611488555e-05,
+      "loss": 3.9957,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.60899371864714e-05,
+      "loss": 4.1189,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6080646351568486e-05,
+      "loss": 3.8005,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6071353611774005e-05,
+      "loss": 3.9326,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.60620589686855e-05,
+      "loss": 3.9159,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.60527624239008e-05,
+      "loss": 3.9512,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.604346397901811e-05,
+      "loss": 4.0132,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.603416363563593e-05,
+      "loss": 3.9283,
+      "step": 2921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.602486139535311e-05,
+      "loss": 4.0338,
+      "step": 2922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.60155572597688e-05,
+      "loss": 3.9738,
+      "step": 2923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6006251230482504e-05,
+      "loss": 3.9723,
+      "step": 2924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.599694330909401e-05,
+      "loss": 4.0645,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.598763349720348e-05,
+      "loss": 3.8083,
+      "step": 2926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5978321796411377e-05,
+      "loss": 3.9962,
+      "step": 2927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5969008208318486e-05,
+      "loss": 3.8348,
+      "step": 2928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.595969273452591e-05,
+      "loss": 3.8367,
+      "step": 2929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.59503753766351e-05,
+      "loss": 3.9649,
+      "step": 2930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5941056136247825e-05,
+      "loss": 3.9399,
+      "step": 2931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.593173501496616e-05,
+      "loss": 3.9926,
+      "step": 2932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5922412014392506e-05,
+      "loss": 4.0025,
+      "step": 2933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.591308713612961e-05,
+      "loss": 4.0642,
+      "step": 2934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.590376038178051e-05,
+      "loss": 3.9683,
+      "step": 2935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5894431752948605e-05,
+      "loss": 3.9672,
+      "step": 2936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.588510125123757e-05,
+      "loss": 4.1359,
+      "step": 2937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.587576887825145e-05,
+      "loss": 3.9243,
+      "step": 2938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.586643463559457e-05,
+      "loss": 4.0587,
+      "step": 2939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.58570985248716e-05,
+      "loss": 4.076,
+      "step": 2940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5847760547687536e-05,
+      "loss": 3.8778,
+      "step": 2941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.583842070564768e-05,
+      "loss": 4.0635,
+      "step": 2942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5829079000357646e-05,
+      "loss": 4.0346,
+      "step": 2943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5819735433423386e-05,
+      "loss": 3.9798,
+      "step": 2944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.581039000645118e-05,
+      "loss": 3.8866,
+      "step": 2945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5801042721047604e-05,
+      "loss": 4.0231,
+      "step": 2946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5791693578819564e-05,
+      "loss": 3.9042,
+      "step": 2947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.57823425813743e-05,
+      "loss": 3.9331,
+      "step": 2948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.577298973031934e-05,
+      "loss": 3.8926,
+      "step": 2949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5763635027262555e-05,
+      "loss": 4.0423,
+      "step": 2950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.575427847381212e-05,
+      "loss": 4.0322,
+      "step": 2951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5744920071576546e-05,
+      "loss": 3.854,
+      "step": 2952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.573555982216465e-05,
+      "loss": 3.9707,
+      "step": 2953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.572619772718556e-05,
+      "loss": 3.9334,
+      "step": 2954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5716833788248725e-05,
+      "loss": 3.8849,
+      "step": 2955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5707468006963935e-05,
+      "loss": 3.9605,
+      "step": 2956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5698100384941255e-05,
+      "loss": 3.8509,
+      "step": 2957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.568873092379109e-05,
+      "loss": 3.9647,
+      "step": 2958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5679359625124176e-05,
+      "loss": 3.955,
+      "step": 2959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.566998649055154e-05,
+      "loss": 4.0304,
+      "step": 2960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5660611521684515e-05,
+      "loss": 4.0561,
+      "step": 2961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.565123472013478e-05,
+      "loss": 3.8509,
+      "step": 2962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5641856087514334e-05,
+      "loss": 3.9168,
+      "step": 2963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.563247562543545e-05,
+      "loss": 3.8823,
+      "step": 2964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5623093335510735e-05,
+      "loss": 3.8374,
+      "step": 2965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.561370921935313e-05,
+      "loss": 3.9141,
+      "step": 2966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.560432327857586e-05,
+      "loss": 3.9599,
+      "step": 2967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.559493551479249e-05,
+      "loss": 4.0308,
+      "step": 2968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.558554592961687e-05,
+      "loss": 3.9822,
+      "step": 2969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.557615452466319e-05,
+      "loss": 4.0757,
+      "step": 2970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5566761301545945e-05,
+      "loss": 3.9663,
+      "step": 2971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5557366261879914e-05,
+      "loss": 3.9096,
+      "step": 2972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.554796940728024e-05,
+      "loss": 3.9242,
+      "step": 2973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.553857073936235e-05,
+      "loss": 4.1119,
+      "step": 2974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.552917025974197e-05,
+      "loss": 3.8865,
+      "step": 2975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.551976797003515e-05,
+      "loss": 3.9828,
+      "step": 2976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5510363871858274e-05,
+      "loss": 3.9404,
+      "step": 2977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5500957966828e-05,
+      "loss": 4.1628,
+      "step": 2978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.54915502565613e-05,
+      "loss": 4.0329,
+      "step": 2979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.54821407426755e-05,
+      "loss": 3.8918,
+      "step": 2980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.547272942678818e-05,
+      "loss": 3.9578,
+      "step": 2981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.546331631051726e-05,
+      "loss": 3.848,
+      "step": 2982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.545390139548096e-05,
+      "loss": 3.9508,
+      "step": 2983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.544448468329783e-05,
+      "loss": 3.9729,
+      "step": 2984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5435066175586694e-05,
+      "loss": 4.0147,
+      "step": 2985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.542564587396671e-05,
+      "loss": 3.8538,
+      "step": 2986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.541622378005733e-05,
+      "loss": 3.967,
+      "step": 2987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.540679989547833e-05,
+      "loss": 3.8237,
+      "step": 2988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.539737422184979e-05,
+      "loss": 4.0565,
+      "step": 2989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.538794676079207e-05,
+      "loss": 4.1739,
+      "step": 2990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5378517513925884e-05,
+      "loss": 3.8322,
+      "step": 2991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.536908648287222e-05,
+      "loss": 4.0767,
+      "step": 2992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.535965366925238e-05,
+      "loss": 4.0342,
+      "step": 2993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5350219074687975e-05,
+      "loss": 3.7871,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.534078270080091e-05,
+      "loss": 3.9454,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.533134454921344e-05,
+      "loss": 3.9415,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.532190462154805e-05,
+      "loss": 4.0162,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.53124629194276e-05,
+      "loss": 3.9287,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.530301944447523e-05,
+      "loss": 3.9958,
+      "step": 2999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5293574198314366e-05,
+      "loss": 3.9777,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.528412718256877e-05,
+      "loss": 3.8849,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5274678398862485e-05,
+      "loss": 3.8675,
+      "step": 3002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5265227848819865e-05,
+      "loss": 3.9754,
+      "step": 3003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.525577553406557e-05,
+      "loss": 3.916,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5246321456224575e-05,
+      "loss": 3.9026,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5236865616922135e-05,
+      "loss": 4.0423,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.522740801778381e-05,
+      "loss": 3.8932,
+      "step": 3007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.521794866043549e-05,
+      "loss": 4.0304,
+      "step": 3008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.520848754650333e-05,
+      "loss": 4.0639,
+      "step": 3009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.519902467761383e-05,
+      "loss": 4.0926,
+      "step": 3010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.518956005539374e-05,
+      "loss": 3.9792,
+      "step": 3011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.518009368147015e-05,
+      "loss": 4.0184,
+      "step": 3012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.517062555747043e-05,
+      "loss": 3.9683,
+      "step": 3013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5161155685022285e-05,
+      "loss": 3.9719,
+      "step": 3014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5151684065753664e-05,
+      "loss": 3.8939,
+      "step": 3015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5142210701292876e-05,
+      "loss": 4.0781,
+      "step": 3016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.513273559326848e-05,
+      "loss": 3.9043,
+      "step": 3017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.512325874330937e-05,
+      "loss": 3.9089,
+      "step": 3018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5113780153044726e-05,
+      "loss": 4.0102,
+      "step": 3019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.510429982410403e-05,
+      "loss": 3.9408,
+      "step": 3020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.509481775811705e-05,
+      "loss": 4.0278,
+      "step": 3021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.508533395671386e-05,
+      "loss": 4.0017,
+      "step": 3022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.507584842152485e-05,
+      "loss": 3.9163,
+      "step": 3023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.506636115418069e-05,
+      "loss": 4.0533,
+      "step": 3024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.505687215631235e-05,
+      "loss": 4.0444,
+      "step": 3025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.504738142955109e-05,
+      "loss": 3.9724,
+      "step": 3026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.503788897552848e-05,
+      "loss": 4.0484,
+      "step": 3027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5028394795876394e-05,
+      "loss": 4.1191,
+      "step": 3028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5018898892226985e-05,
+      "loss": 3.9668,
+      "step": 3029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5009401266212704e-05,
+      "loss": 3.9471,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4999901919466305e-05,
+      "loss": 3.9649,
+      "step": 3031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4990400853620834e-05,
+      "loss": 4.0602,
+      "step": 3032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.498089807030964e-05,
+      "loss": 3.9589,
+      "step": 3033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.497139357116637e-05,
+      "loss": 3.9944,
+      "step": 3034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4961887357824934e-05,
+      "loss": 4.0009,
+      "step": 3035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.495237943191957e-05,
+      "loss": 3.8791,
+      "step": 3036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.494286979508481e-05,
+      "loss": 4.0723,
+      "step": 3037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4933358448955474e-05,
+      "loss": 3.8888,
+      "step": 3038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4923845395166656e-05,
+      "loss": 4.112,
+      "step": 3039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.491433063535376e-05,
+      "loss": 3.9227,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.49048141711525e-05,
+      "loss": 4.02,
+      "step": 3041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.489529600419885e-05,
+      "loss": 3.935,
+      "step": 3042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.488577613612911e-05,
+      "loss": 3.9316,
+      "step": 3043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.487625456857984e-05,
+      "loss": 3.9016,
+      "step": 3044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4866731303187916e-05,
+      "loss": 4.0699,
+      "step": 3045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.48572063415905e-05,
+      "loss": 3.9556,
+      "step": 3046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.484767968542503e-05,
+      "loss": 4.0298,
+      "step": 3047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4838151336329266e-05,
+      "loss": 4.1392,
+      "step": 3048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4828621295941234e-05,
+      "loss": 3.9583,
+      "step": 3049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4819089565899246e-05,
+      "loss": 3.8591,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4809556147841936e-05,
+      "loss": 4.0348,
+      "step": 3051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.480002104340821e-05,
+      "loss": 4.0325,
+      "step": 3052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.479048425423725e-05,
+      "loss": 4.017,
+      "step": 3053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4780945781968546e-05,
+      "loss": 3.8673,
+      "step": 3054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.477140562824187e-05,
+      "loss": 4.03,
+      "step": 3055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4761863794697284e-05,
+      "loss": 4.005,
+      "step": 3056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.475232028297515e-05,
+      "loss": 3.9275,
+      "step": 3057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.47427750947161e-05,
+      "loss": 4.0581,
+      "step": 3058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.473322823156105e-05,
+      "loss": 3.8592,
+      "step": 3059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.472367969515124e-05,
+      "loss": 4.064,
+      "step": 3060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4714129487128166e-05,
+      "loss": 4.0221,
+      "step": 3061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.470457760913362e-05,
+      "loss": 4.013,
+      "step": 3062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4695024062809666e-05,
+      "loss": 4.013,
+      "step": 3063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4685468849798684e-05,
+      "loss": 4.0152,
+      "step": 3064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4675911971743324e-05,
+      "loss": 3.8647,
+      "step": 3065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.466635343028653e-05,
+      "loss": 3.9831,
+      "step": 3066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.465679322707152e-05,
+      "loss": 4.0793,
+      "step": 3067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.46472313637418e-05,
+      "loss": 3.8484,
+      "step": 3068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.463766784194116e-05,
+      "loss": 4.1333,
+      "step": 3069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.46281026633137e-05,
+      "loss": 4.0067,
+      "step": 3070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4618535829503766e-05,
+      "loss": 3.9293,
+      "step": 3071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4608967342156015e-05,
+      "loss": 3.8798,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.459939720291539e-05,
+      "loss": 3.8039,
+      "step": 3073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.458982541342709e-05,
+      "loss": 3.981,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.458025197533664e-05,
+      "loss": 3.9929,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4570676890289794e-05,
+      "loss": 4.0305,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.456110015993265e-05,
+      "loss": 3.9973,
+      "step": 3077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.455152178591154e-05,
+      "loss": 4.0697,
+      "step": 3078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.454194176987311e-05,
+      "loss": 3.9082,
+      "step": 3079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.453236011346427e-05,
+      "loss": 3.9705,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4522776818332214e-05,
+      "loss": 3.9747,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.451319188612443e-05,
+      "loss": 4.0139,
+      "step": 3082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4503605318488664e-05,
+      "loss": 4.0164,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4494017117072976e-05,
+      "loss": 3.9665,
+      "step": 3084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4484427283525684e-05,
+      "loss": 4.0115,
+      "step": 3085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4474835819495384e-05,
+      "loss": 4.04,
+      "step": 3086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4465242726630966e-05,
+      "loss": 4.1008,
+      "step": 3087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4455648006581596e-05,
+      "loss": 4.0732,
+      "step": 3088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.444605166099672e-05,
+      "loss": 4.1936,
+      "step": 3089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.443645369152604e-05,
+      "loss": 4.0855,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.442685409981958e-05,
+      "loss": 3.8749,
+      "step": 3091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4417252887527616e-05,
+      "loss": 3.8895,
+      "step": 3092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.440765005630071e-05,
+      "loss": 4.0375,
+      "step": 3093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4398045607789685e-05,
+      "loss": 3.9111,
+      "step": 3094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4388439543645686e-05,
+      "loss": 4.07,
+      "step": 3095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.437883186552008e-05,
+      "loss": 4.0305,
+      "step": 3096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.436922257506454e-05,
+      "loss": 3.84,
+      "step": 3097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.435961167393103e-05,
+      "loss": 3.9063,
+      "step": 3098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.434999916377177e-05,
+      "loss": 4.11,
+      "step": 3099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.434038504623926e-05,
+      "loss": 4.1114,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.433076932298627e-05,
+      "loss": 3.9966,
+      "step": 3101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.432115199566586e-05,
+      "loss": 4.067,
+      "step": 3102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.431153306593137e-05,
+      "loss": 4.0682,
+      "step": 3103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.43019125354364e-05,
+      "loss": 3.9562,
+      "step": 3104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.429229040583482e-05,
+      "loss": 4.0135,
+      "step": 3105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.428266667878079e-05,
+      "loss": 3.9895,
+      "step": 3106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.427304135592876e-05,
+      "loss": 4.0271,
+      "step": 3107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.426341443893341e-05,
+      "loss": 4.0794,
+      "step": 3108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.425378592944972e-05,
+      "loss": 4.0759,
+      "step": 3109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4244155829132956e-05,
+      "loss": 4.042,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.423452413963864e-05,
+      "loss": 3.9271,
+      "step": 3111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4224890862622566e-05,
+      "loss": 4.0682,
+      "step": 3112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4215255999740805e-05,
+      "loss": 3.9873,
+      "step": 3113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.420561955264972e-05,
+      "loss": 4.1215,
+      "step": 3114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.419598152300589e-05,
+      "loss": 3.952,
+      "step": 3115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.418634191246624e-05,
+      "loss": 3.9836,
+      "step": 3116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.417670072268792e-05,
+      "loss": 4.061,
+      "step": 3117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.416705795532836e-05,
+      "loss": 4.0543,
+      "step": 3118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.415741361204526e-05,
+      "loss": 4.0444,
+      "step": 3119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.41477676944966e-05,
+      "loss": 4.038,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4138120204340626e-05,
+      "loss": 4.0009,
+      "step": 3121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4128471143235844e-05,
+      "loss": 4.1129,
+      "step": 3122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4118820512841055e-05,
+      "loss": 4.0633,
+      "step": 3123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.41091683148153e-05,
+      "loss": 3.9973,
+      "step": 3124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.409951455081791e-05,
+      "loss": 3.9245,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4089859222508476e-05,
+      "loss": 4.1039,
+      "step": 3126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.408020233154686e-05,
+      "loss": 4.0323,
+      "step": 3127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.407054387959321e-05,
+      "loss": 3.8787,
+      "step": 3128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.40608838683079e-05,
+      "loss": 3.8693,
+      "step": 3129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.405122229935161e-05,
+      "loss": 3.9339,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4041559174385284e-05,
+      "loss": 4.0482,
+      "step": 3131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4031894495070116e-05,
+      "loss": 3.9523,
+      "step": 3132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.402222826306757e-05,
+      "loss": 3.9385,
+      "step": 3133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.401256048003939e-05,
+      "loss": 3.9638,
+      "step": 3134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4002891147647595e-05,
+      "loss": 4.0657,
+      "step": 3135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.399322026755443e-05,
+      "loss": 4.0567,
+      "step": 3136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3983547841422444e-05,
+      "loss": 4.1076,
+      "step": 3137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3973873870914434e-05,
+      "loss": 4.0444,
+      "step": 3138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.396419835769348e-05,
+      "loss": 4.0746,
+      "step": 3139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.39545213034229e-05,
+      "loss": 3.9478,
+      "step": 3140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.394484270976631e-05,
+      "loss": 3.9903,
+      "step": 3141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.393516257838755e-05,
+      "loss": 4.048,
+      "step": 3142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3925480910950756e-05,
+      "loss": 4.0394,
+      "step": 3143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.391579770912032e-05,
+      "loss": 4.1413,
+      "step": 3144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.39061129745609e-05,
+      "loss": 4.0354,
+      "step": 3145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.389642670893742e-05,
+      "loss": 4.0256,
+      "step": 3146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.388673891391504e-05,
+      "loss": 3.8341,
+      "step": 3147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.387704959115922e-05,
+      "loss": 4.0235,
+      "step": 3148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.386735874233566e-05,
+      "loss": 3.911,
+      "step": 3149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.385766636911035e-05,
+      "loss": 3.8874,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.384797247314949e-05,
+      "loss": 3.9922,
+      "step": 3151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.383827705611958e-05,
+      "loss": 3.8682,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.38285801196874e-05,
+      "loss": 4.0416,
+      "step": 3153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.381888166551994e-05,
+      "loss": 3.9378,
+      "step": 3154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3809181695284484e-05,
+      "loss": 3.9538,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.379948021064857e-05,
+      "loss": 3.8916,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3789777213279986e-05,
+      "loss": 3.9522,
+      "step": 3157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.378007270484681e-05,
+      "loss": 3.9896,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.377036668701735e-05,
+      "loss": 4.0061,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.376065916146018e-05,
+      "loss": 4.0081,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.375095012984413e-05,
+      "loss": 3.9647,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.374123959383831e-05,
+      "loss": 3.9629,
+      "step": 3162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.373152755511207e-05,
+      "loss": 3.9681,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3721814015335014e-05,
+      "loss": 3.8971,
+      "step": 3164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.371209897617701e-05,
+      "loss": 3.971,
+      "step": 3165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.37023824393082e-05,
+      "loss": 3.9193,
+      "step": 3166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3692664406398966e-05,
+      "loss": 3.953,
+      "step": 3167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.368294487911995e-05,
+      "loss": 3.9122,
+      "step": 3168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.367322385914203e-05,
+      "loss": 3.9719,
+      "step": 3169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.366350134813639e-05,
+      "loss": 3.9798,
+      "step": 3170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.365377734777444e-05,
+      "loss": 3.9449,
+      "step": 3171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3644051859727835e-05,
+      "loss": 4.049,
+      "step": 3172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.36343248856685e-05,
+      "loss": 3.9062,
+      "step": 3173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.362459642726863e-05,
+      "loss": 4.06,
+      "step": 3174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3614866486200635e-05,
+      "loss": 4.0455,
+      "step": 3175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.360513506413722e-05,
+      "loss": 3.9329,
+      "step": 3176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.359540216275134e-05,
+      "loss": 3.969,
+      "step": 3177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.358566778371618e-05,
+      "loss": 4.0549,
+      "step": 3178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.357593192870518e-05,
+      "loss": 4.0364,
+      "step": 3179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.356619459939208e-05,
+      "loss": 3.9665,
+      "step": 3180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.355645579745081e-05,
+      "loss": 4.04,
+      "step": 3181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3546715524555594e-05,
+      "loss": 3.9089,
+      "step": 3182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.35369737823809e-05,
+      "loss": 3.908,
+      "step": 3183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3527230572601444e-05,
+      "loss": 3.9871,
+      "step": 3184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.351748589689219e-05,
+      "loss": 3.9398,
+      "step": 3185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.350773975692837e-05,
+      "loss": 3.9282,
+      "step": 3186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.349799215438546e-05,
+      "loss": 4.0373,
+      "step": 3187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.348824309093918e-05,
+      "loss": 3.9208,
+      "step": 3188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3478492568265507e-05,
+      "loss": 3.9495,
+      "step": 3189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.346874058804067e-05,
+      "loss": 3.9343,
+      "step": 3190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.345898715194115e-05,
+      "loss": 3.9516,
+      "step": 3191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.344923226164368e-05,
+      "loss": 3.9131,
+      "step": 3192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.343947591882521e-05,
+      "loss": 3.8381,
+      "step": 3193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3429718125163015e-05,
+      "loss": 4.1283,
+      "step": 3194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3419958882334536e-05,
+      "loss": 3.987,
+      "step": 3195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.341019819201752e-05,
+      "loss": 3.8953,
+      "step": 3196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.340043605588993e-05,
+      "loss": 3.8537,
+      "step": 3197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3390672475629995e-05,
+      "loss": 3.9903,
+      "step": 3198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.338090745291619e-05,
+      "loss": 3.8459,
+      "step": 3199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3371140989427235e-05,
+      "loss": 4.0258,
+      "step": 3200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3361373086842095e-05,
+      "loss": 3.9321,
+      "step": 3201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.335160374683999e-05,
+      "loss": 3.9908,
+      "step": 3202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.334183297110038e-05,
+      "loss": 3.9652,
+      "step": 3203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.333206076130298e-05,
+      "loss": 4.0855,
+      "step": 3204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.332228711912773e-05,
+      "loss": 4.0262,
+      "step": 3205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.331251204625485e-05,
+      "loss": 3.9255,
+      "step": 3206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3302735544364797e-05,
+      "loss": 3.9251,
+      "step": 3207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.329295761513822e-05,
+      "loss": 3.9618,
+      "step": 3208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.328317826025611e-05,
+      "loss": 3.878,
+      "step": 3209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3273397481399626e-05,
+      "loss": 3.9248,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.326361528025021e-05,
+      "loss": 3.9507,
+      "step": 3211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3253831658489507e-05,
+      "loss": 3.9415,
+      "step": 3212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3244046617799473e-05,
+      "loss": 3.8668,
+      "step": 3213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.323426015986225e-05,
+      "loss": 3.8841,
+      "step": 3214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3224472286360234e-05,
+      "loss": 3.8417,
+      "step": 3215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.32146829989761e-05,
+      "loss": 3.9909,
+      "step": 3216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3204892299392715e-05,
+      "loss": 3.8904,
+      "step": 3217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.319510018929323e-05,
+      "loss": 3.9341,
+      "step": 3218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3185306670361016e-05,
+      "loss": 3.9409,
+      "step": 3219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.31755117442797e-05,
+      "loss": 3.932,
+      "step": 3220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.316571541273312e-05,
+      "loss": 3.9597,
+      "step": 3221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.315591767740541e-05,
+      "loss": 3.9256,
+      "step": 3222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.314611853998089e-05,
+      "loss": 3.9847,
+      "step": 3223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3136318002144165e-05,
+      "loss": 4.0298,
+      "step": 3224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.312651606558004e-05,
+      "loss": 3.9258,
+      "step": 3225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.31167127319736e-05,
+      "loss": 3.8805,
+      "step": 3226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.310690800301014e-05,
+      "loss": 4.043,
+      "step": 3227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3097101880375215e-05,
+      "loss": 3.8345,
+      "step": 3228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.308729436575461e-05,
+      "loss": 4.1037,
+      "step": 3229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.307748546083434e-05,
+      "loss": 4.0772,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3067675167300676e-05,
+      "loss": 3.8763,
+      "step": 3231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.305786348684013e-05,
+      "loss": 3.9972,
+      "step": 3232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.304805042113943e-05,
+      "loss": 3.93,
+      "step": 3233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.303823597188557e-05,
+      "loss": 3.9705,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.302842014076575e-05,
+      "loss": 4.014,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3018602929467435e-05,
+      "loss": 4.106,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.300878433967832e-05,
+      "loss": 3.9671,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2998964373086336e-05,
+      "loss": 3.9878,
+      "step": 3238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.298914303137964e-05,
+      "loss": 4.0429,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.297932031624663e-05,
+      "loss": 4.0489,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2969496229375966e-05,
+      "loss": 3.8748,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.29596707724565e-05,
+      "loss": 4.002,
+      "step": 3242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.294984394717736e-05,
+      "loss": 3.9928,
+      "step": 3243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2940015755227885e-05,
+      "loss": 3.9229,
+      "step": 3244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2930186198297644e-05,
+      "loss": 3.9587,
+      "step": 3245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2920355278076466e-05,
+      "loss": 3.8819,
+      "step": 3246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.291052299625439e-05,
+      "loss": 4.0663,
+      "step": 3247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.290068935452171e-05,
+      "loss": 4.0265,
+      "step": 3248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2890854354568935e-05,
+      "loss": 4.0633,
+      "step": 3249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.288101799808682e-05,
+      "loss": 3.9866,
+      "step": 3250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.287118028676634e-05,
+      "loss": 3.9995,
+      "step": 3251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2861341222298724e-05,
+      "loss": 3.9084,
+      "step": 3252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2851500806375415e-05,
+      "loss": 4.0255,
+      "step": 3253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.28416590406881e-05,
+      "loss": 3.9481,
+      "step": 3254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.283181592692868e-05,
+      "loss": 4.0281,
+      "step": 3255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2821971466789306e-05,
+      "loss": 3.9559,
+      "step": 3256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.281212566196236e-05,
+      "loss": 4.1112,
+      "step": 3257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.280227851414044e-05,
+      "loss": 4.0928,
+      "step": 3258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.279243002501641e-05,
+      "loss": 3.9763,
+      "step": 3259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.278258019628331e-05,
+      "loss": 4.0709,
+      "step": 3260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2772729029634455e-05,
+      "loss": 4.0819,
+      "step": 3261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.276287652676336e-05,
+      "loss": 3.9365,
+      "step": 3262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.27530226893638e-05,
+      "loss": 3.882,
+      "step": 3263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.274316751912976e-05,
+      "loss": 3.9934,
+      "step": 3264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.273331101775546e-05,
+      "loss": 4.0623,
+      "step": 3265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.272345318693534e-05,
+      "loss": 4.0107,
+      "step": 3266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.271359402836408e-05,
+      "loss": 3.9061,
+      "step": 3267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2703733543736585e-05,
+      "loss": 3.9342,
+      "step": 3268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2693871734747986e-05,
+      "loss": 3.9304,
+      "step": 3269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2684008603093634e-05,
+      "loss": 3.908,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2674144150469135e-05,
+      "loss": 3.969,
+      "step": 3271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2664278378570274e-05,
+      "loss": 3.8257,
+      "step": 3272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.265441128909311e-05,
+      "loss": 3.9947,
+      "step": 3273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.264454288373392e-05,
+      "loss": 4.1109,
+      "step": 3274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.263467316418918e-05,
+      "loss": 4.0165,
+      "step": 3275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.262480213215561e-05,
+      "loss": 3.9707,
+      "step": 3276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2614929789330154e-05,
+      "loss": 3.9668,
+      "step": 3277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.260505613741e-05,
+      "loss": 4.0724,
+      "step": 3278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.259518117809252e-05,
+      "loss": 3.9716,
+      "step": 3279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.258530491307535e-05,
+      "loss": 4.1494,
+      "step": 3280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.257542734405634e-05,
+      "loss": 4.0666,
+      "step": 3281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.256554847273353e-05,
+      "loss": 3.9298,
+      "step": 3282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2555668300805235e-05,
+      "loss": 3.9868,
+      "step": 3283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.254578682996997e-05,
+      "loss": 3.8321,
+      "step": 3284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2535904061926476e-05,
+      "loss": 4.0886,
+      "step": 3285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2526019998373705e-05,
+      "loss": 3.9169,
+      "step": 3286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2516134641010836e-05,
+      "loss": 3.9413,
+      "step": 3287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.25062479915373e-05,
+      "loss": 4.0114,
+      "step": 3288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.249636005165271e-05,
+      "loss": 4.0329,
+      "step": 3289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.248647082305693e-05,
+      "loss": 3.8731,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2476580307450014e-05,
+      "loss": 3.9339,
+      "step": 3291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2466688506532275e-05,
+      "loss": 4.0185,
+      "step": 3292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.245679542200422e-05,
+      "loss": 4.0681,
+      "step": 3293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.244690105556657e-05,
+      "loss": 4.0676,
+      "step": 3294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2437005408920315e-05,
+      "loss": 3.9295,
+      "step": 3295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.242710848376661e-05,
+      "loss": 3.9731,
+      "step": 3296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2417210281806844e-05,
+      "loss": 3.9681,
+      "step": 3297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.240731080474265e-05,
+      "loss": 3.8936,
+      "step": 3298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.239741005427586e-05,
+      "loss": 4.0357,
+      "step": 3299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2387508032108515e-05,
+      "loss": 4.0009,
+      "step": 3300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.23776047399429e-05,
+      "loss": 3.9981,
+      "step": 3301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.23677001794815e-05,
+      "loss": 4.0019,
+      "step": 3302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.235779435242703e-05,
+      "loss": 3.7997,
+      "step": 3303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.234788726048241e-05,
+      "loss": 3.8907,
+      "step": 3304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.233797890535079e-05,
+      "loss": 3.9046,
+      "step": 3305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2328069288735515e-05,
+      "loss": 3.7728,
+      "step": 3306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.231815841234019e-05,
+      "loss": 3.9364,
+      "step": 3307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.230824627786858e-05,
+      "loss": 4.0983,
+      "step": 3308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2298332887024716e-05,
+      "loss": 3.9454,
+      "step": 3309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.228841824151283e-05,
+      "loss": 3.785,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2278502343037344e-05,
+      "loss": 3.8934,
+      "step": 3311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.226858519330292e-05,
+      "loss": 4.0249,
+      "step": 3312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.225866679401444e-05,
+      "loss": 3.8401,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.224874714687699e-05,
+      "loss": 3.9602,
+      "step": 3314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.223882625359587e-05,
+      "loss": 3.9771,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2228904115876604e-05,
+      "loss": 3.9111,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.221898073542491e-05,
+      "loss": 4.0932,
+      "step": 3317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.220905611394674e-05,
+      "loss": 4.1247,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.219913025314824e-05,
+      "loss": 3.8526,
+      "step": 3319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2189203154735804e-05,
+      "loss": 3.9973,
+      "step": 3320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2179274820415995e-05,
+      "loss": 3.895,
+      "step": 3321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.216934525189561e-05,
+      "loss": 3.9762,
+      "step": 3322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.215941445088167e-05,
+      "loss": 4.0244,
+      "step": 3323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.214948241908138e-05,
+      "loss": 4.1189,
+      "step": 3324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.213954915820219e-05,
+      "loss": 4.1074,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.212961466995172e-05,
+      "loss": 4.0735,
+      "step": 3326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2119678956037844e-05,
+      "loss": 4.1732,
+      "step": 3327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.21097420181686e-05,
+      "loss": 3.8745,
+      "step": 3328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.20998038580523e-05,
+      "loss": 4.0284,
+      "step": 3329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.20898644773974e-05,
+      "loss": 3.8693,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2079923877912605e-05,
+      "loss": 3.9623,
+      "step": 3331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2069982061306815e-05,
+      "loss": 4.0317,
+      "step": 3332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.206003902928914e-05,
+      "loss": 3.8455,
+      "step": 3333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.205009478356892e-05,
+      "loss": 3.82,
+      "step": 3334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2040149325855675e-05,
+      "loss": 3.9071,
+      "step": 3335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.203020265785914e-05,
+      "loss": 4.0406,
+      "step": 3336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.202025478128926e-05,
+      "loss": 3.9643,
+      "step": 3337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2010305697856216e-05,
+      "loss": 3.9544,
+      "step": 3338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.200035540927034e-05,
+      "loss": 4.0916,
+      "step": 3339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.199040391724221e-05,
+      "loss": 3.9389,
+      "step": 3340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.198045122348262e-05,
+      "loss": 3.8628,
+      "step": 3341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1970497329702526e-05,
+      "loss": 3.8573,
+      "step": 3342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.196054223761315e-05,
+      "loss": 3.9744,
+      "step": 3343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.195058594892585e-05,
+      "loss": 3.894,
+      "step": 3344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.194062846535226e-05,
+      "loss": 3.9998,
+      "step": 3345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1930669788604165e-05,
+      "loss": 3.992,
+      "step": 3346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1920709920393595e-05,
+      "loss": 3.8247,
+      "step": 3347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.191074886243275e-05,
+      "loss": 3.8871,
+      "step": 3348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.190078661643406e-05,
+      "loss": 4.0399,
+      "step": 3349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1890823184110166e-05,
+      "loss": 3.9778,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.188085856717386e-05,
+      "loss": 4.0857,
+      "step": 3351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1870892767338204e-05,
+      "loss": 4.0873,
+      "step": 3352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.186092578631643e-05,
+      "loss": 3.9719,
+      "step": 3353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.185095762582197e-05,
+      "loss": 4.0336,
+      "step": 3354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1840988287568474e-05,
+      "loss": 3.9267,
+      "step": 3355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.183101777326978e-05,
+      "loss": 4.0036,
+      "step": 3356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.182104608463995e-05,
+      "loss": 3.9964,
+      "step": 3357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.181107322339321e-05,
+      "loss": 4.0037,
+      "step": 3358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.180109919124403e-05,
+      "loss": 3.9813,
+      "step": 3359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.179112398990705e-05,
+      "loss": 3.8695,
+      "step": 3360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.178114762109712e-05,
+      "loss": 4.0287,
+      "step": 3361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.177117008652931e-05,
+      "loss": 4.0254,
+      "step": 3362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.176119138791886e-05,
+      "loss": 3.9424,
+      "step": 3363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1751211526981235e-05,
+      "loss": 3.9943,
+      "step": 3364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.174123050543207e-05,
+      "loss": 3.9515,
+      "step": 3365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.173124832498724e-05,
+      "loss": 3.969,
+      "step": 3366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.172126498736278e-05,
+      "loss": 4.0461,
+      "step": 3367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.171128049427496e-05,
+      "loss": 3.8583,
+      "step": 3368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.17012948474402e-05,
+      "loss": 3.9584,
+      "step": 3369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1691308048575186e-05,
+      "loss": 4.0352,
+      "step": 3370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.168132009939674e-05,
+      "loss": 3.9818,
+      "step": 3371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1671331001621906e-05,
+      "loss": 3.9862,
+      "step": 3372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1661340756967945e-05,
+      "loss": 4.0929,
+      "step": 3373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.165134936715227e-05,
+      "loss": 3.9355,
+      "step": 3374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1641356833892544e-05,
+      "loss": 4.0903,
+      "step": 3375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.163136315890657e-05,
+      "loss": 3.9364,
+      "step": 3376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.16213683439124e-05,
+      "loss": 3.9288,
+      "step": 3377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1611372390628255e-05,
+      "loss": 3.8635,
+      "step": 3378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.160137530077254e-05,
+      "loss": 4.0043,
+      "step": 3379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1591377076063885e-05,
+      "loss": 3.9274,
+      "step": 3380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1581377718221095e-05,
+      "loss": 4.0234,
+      "step": 3381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.157137722896318e-05,
+      "loss": 3.949,
+      "step": 3382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.156137561000933e-05,
+      "loss": 4.0474,
+      "step": 3383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.155137286307896e-05,
+      "loss": 3.9643,
+      "step": 3384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.154136898989164e-05,
+      "loss": 3.9349,
+      "step": 3385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.153136399216715e-05,
+      "loss": 3.8351,
+      "step": 3386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.152135787162548e-05,
+      "loss": 3.8407,
+      "step": 3387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.151135062998678e-05,
+      "loss": 4.0115,
+      "step": 3388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1501342268971434e-05,
+      "loss": 4.01,
+      "step": 3389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.149133279029997e-05,
+      "loss": 3.8795,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.148132219569315e-05,
+      "loss": 4.1283,
+      "step": 3391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1471310486871914e-05,
+      "loss": 3.8794,
+      "step": 3392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.146129766555737e-05,
+      "loss": 4.0097,
+      "step": 3393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.145128373347086e-05,
+      "loss": 4.0319,
+      "step": 3394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1441268692333885e-05,
+      "loss": 3.9434,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.143125254386815e-05,
+      "loss": 4.0516,
+      "step": 3396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.142123528979554e-05,
+      "loss": 4.0987,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.141121693183814e-05,
+      "loss": 3.9173,
+      "step": 3398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.140119747171824e-05,
+      "loss": 3.9804,
+      "step": 3399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1391176911158274e-05,
+      "loss": 3.9457,
+      "step": 3400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.13811552518809e-05,
+      "loss": 3.9103,
+      "step": 3401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.137113249560896e-05,
+      "loss": 4.0242,
+      "step": 3402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.13611086440655e-05,
+      "loss": 3.8795,
+      "step": 3403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.13510836989737e-05,
+      "loss": 3.8922,
+      "step": 3404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1341057662057e-05,
+      "loss": 3.8732,
+      "step": 3405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.133103053503896e-05,
+      "loss": 3.9103,
+      "step": 3406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.132100231964339e-05,
+      "loss": 3.8404,
+      "step": 3407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1310973017594235e-05,
+      "loss": 4.0465,
+      "step": 3408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.130094263061566e-05,
+      "loss": 4.0442,
+      "step": 3409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1290911160432e-05,
+      "loss": 4.0878,
+      "step": 3410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.128087860876778e-05,
+      "loss": 3.8821,
+      "step": 3411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1270844977347705e-05,
+      "loss": 4.0216,
+      "step": 3412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.126081026789669e-05,
+      "loss": 3.9092,
+      "step": 3413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.125077448213981e-05,
+      "loss": 4.0514,
+      "step": 3414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.124073762180232e-05,
+      "loss": 4.0512,
+      "step": 3415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1230699688609695e-05,
+      "loss": 3.8939,
+      "step": 3416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1220660684287555e-05,
+      "loss": 3.9552,
+      "step": 3417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1210620610561735e-05,
+      "loss": 4.0413,
+      "step": 3418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.120057946915822e-05,
+      "loss": 3.9939,
+      "step": 3419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1190537261803214e-05,
+      "loss": 4.0196,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.118049399022309e-05,
+      "loss": 4.0411,
+      "step": 3421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.11704496561444e-05,
+      "loss": 3.9252,
+      "step": 3422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.116040426129388e-05,
+      "loss": 3.9783,
+      "step": 3423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.115035780739845e-05,
+      "loss": 4.0584,
+      "step": 3424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.114031029618521e-05,
+      "loss": 3.9264,
+      "step": 3425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1130261729381435e-05,
+      "loss": 4.0036,
+      "step": 3426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.112021210871462e-05,
+      "loss": 4.0007,
+      "step": 3427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.111016143591238e-05,
+      "loss": 4.0585,
+      "step": 3428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1100109712702556e-05,
+      "loss": 3.993,
+      "step": 3429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.109005694081315e-05,
+      "loss": 3.7466,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.108000312197236e-05,
+      "loss": 3.9402,
+      "step": 3431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.106994825790855e-05,
+      "loss": 3.9776,
+      "step": 3432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1059892350350264e-05,
+      "loss": 3.9824,
+      "step": 3433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.104983540102622e-05,
+      "loss": 4.0618,
+      "step": 3434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.103977741166535e-05,
+      "loss": 3.8837,
+      "step": 3435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.102971838399672e-05,
+      "loss": 4.0214,
+      "step": 3436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.101965831974959e-05,
+      "loss": 3.8353,
+      "step": 3437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.100959722065342e-05,
+      "loss": 3.9392,
+      "step": 3438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.099953508843781e-05,
+      "loss": 4.0231,
+      "step": 3439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0989471924832564e-05,
+      "loss": 3.9447,
+      "step": 3440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.097940773156766e-05,
+      "loss": 4.0341,
+      "step": 3441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0969342510373255e-05,
+      "loss": 4.0455,
+      "step": 3442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.095927626297966e-05,
+      "loss": 4.0543,
+      "step": 3443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.094920899111739e-05,
+      "loss": 3.912,
+      "step": 3444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.093914069651714e-05,
+      "loss": 3.8783,
+      "step": 3445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.092907138090974e-05,
+      "loss": 3.9927,
+      "step": 3446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.091900104602623e-05,
+      "loss": 3.9612,
+      "step": 3447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0908929693597826e-05,
+      "loss": 3.9524,
+      "step": 3448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0898857325355905e-05,
+      "loss": 4.0372,
+      "step": 3449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.088878394303203e-05,
+      "loss": 3.9085,
+      "step": 3450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.087870954835792e-05,
+      "loss": 3.9987,
+      "step": 3451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.086863414306549e-05,
+      "loss": 3.9871,
+      "step": 3452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.085855772888682e-05,
+      "loss": 3.9651,
+      "step": 3453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.084848030755415e-05,
+      "loss": 4.0133,
+      "step": 3454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.083840188079991e-05,
+      "loss": 4.0617,
+      "step": 3455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.082832245035671e-05,
+      "loss": 4.0877,
+      "step": 3456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.081824201795731e-05,
+      "loss": 3.8859,
+      "step": 3457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0808160585334656e-05,
+      "loss": 4.0642,
+      "step": 3458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.079807815422187e-05,
+      "loss": 4.0,
+      "step": 3459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0787994726352234e-05,
+      "loss": 3.9342,
+      "step": 3460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.077791030345919e-05,
+      "loss": 3.9587,
+      "step": 3461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.076782488727638e-05,
+      "loss": 3.966,
+      "step": 3462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.075773847953762e-05,
+      "loss": 4.0996,
+      "step": 3463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.074765108197687e-05,
+      "loss": 3.9083,
+      "step": 3464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0737562696328254e-05,
+      "loss": 4.0152,
+      "step": 3465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.07274733243261e-05,
+      "loss": 3.8949,
+      "step": 3466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0717382967704875e-05,
+      "loss": 4.1084,
+      "step": 3467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0707291628199246e-05,
+      "loss": 3.9457,
+      "step": 3468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0697199307544014e-05,
+      "loss": 3.8877,
+      "step": 3469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.068710600747418e-05,
+      "loss": 4.033,
+      "step": 3470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.067701172972489e-05,
+      "loss": 3.9979,
+      "step": 3471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.066691647603147e-05,
+      "loss": 3.9561,
+      "step": 3472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0656820248129414e-05,
+      "loss": 3.9595,
+      "step": 3473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.064672304775438e-05,
+      "loss": 3.9959,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0636624876642186e-05,
+      "loss": 4.1299,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.062652573652882e-05,
+      "loss": 3.8569,
+      "step": 3476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.061642562915047e-05,
+      "loss": 3.9597,
+      "step": 3477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.060632455624344e-05,
+      "loss": 3.8994,
+      "step": 3478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0596222519544216e-05,
+      "loss": 3.928,
+      "step": 3479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.058611952078947e-05,
+      "loss": 3.9843,
+      "step": 3480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.057601556171601e-05,
+      "loss": 3.8765,
+      "step": 3481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0565910644060847e-05,
+      "loss": 4.1021,
+      "step": 3482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0555804769561106e-05,
+      "loss": 3.9552,
+      "step": 3483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.054569793995413e-05,
+      "loss": 3.8101,
+      "step": 3484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0535590156977376e-05,
+      "loss": 3.8638,
+      "step": 3485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.05254814223685e-05,
+      "loss": 3.9625,
+      "step": 3486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.051537173786531e-05,
+      "loss": 4.019,
+      "step": 3487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0505261105205786e-05,
+      "loss": 4.0605,
+      "step": 3488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0495149526128054e-05,
+      "loss": 3.9407,
+      "step": 3489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0485037002370403e-05,
+      "loss": 3.8257,
+      "step": 3490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.047492353567131e-05,
+      "loss": 3.9928,
+      "step": 3491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.04648091277694e-05,
+      "loss": 3.8341,
+      "step": 3492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0454693780403444e-05,
+      "loss": 4.0177,
+      "step": 3493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0444577495312376e-05,
+      "loss": 3.8806,
+      "step": 3494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0434460274235328e-05,
+      "loss": 3.9117,
+      "step": 3495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0424342118911558e-05,
+      "loss": 3.9697,
+      "step": 3496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0414223031080492e-05,
+      "loss": 4.1646,
+      "step": 3497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0404103012481722e-05,
+      "loss": 3.9865,
+      "step": 3498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0393982064854987e-05,
+      "loss": 3.8465,
+      "step": 3499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0383860189940213e-05,
+      "loss": 4.0148,
+      "step": 3500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0373737389477446e-05,
+      "loss": 4.0682,
+      "step": 3501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0363613665206934e-05,
+      "loss": 3.9647,
+      "step": 3502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.035348901886905e-05,
+      "loss": 3.9511,
+      "step": 3503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0343363452204338e-05,
+      "loss": 3.8953,
+      "step": 3504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0333236966953505e-05,
+      "loss": 3.8627,
+      "step": 3505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.032310956485741e-05,
+      "loss": 4.0208,
+      "step": 3506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.031298124765707e-05,
+      "loss": 4.0178,
+      "step": 3507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0302852017093653e-05,
+      "loss": 4.1802,
+      "step": 3508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0292721874908503e-05,
+      "loss": 3.9941,
+      "step": 3509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.028259082284311e-05,
+      "loss": 3.9442,
+      "step": 3510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.027245886263911e-05,
+      "loss": 3.9802,
+      "step": 3511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0262325996038306e-05,
+      "loss": 3.9049,
+      "step": 3512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0252192224782655e-05,
+      "loss": 3.8858,
+      "step": 3513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0242057550614278e-05,
+      "loss": 3.9777,
+      "step": 3514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0231921975275428e-05,
+      "loss": 3.9553,
+      "step": 3515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0221785500508548e-05,
+      "loss": 3.8894,
+      "step": 3516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0211648128056198e-05,
+      "loss": 4.0433,
+      "step": 3517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0201509859661114e-05,
+      "loss": 4.0727,
+      "step": 3518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.019137069706619e-05,
+      "loss": 4.0731,
+      "step": 3519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0181230642014457e-05,
+      "loss": 3.9196,
+      "step": 3520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0171089696249112e-05,
+      "loss": 4.0,
+      "step": 3521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.01609478615135e-05,
+      "loss": 3.9417,
+      "step": 3522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0150805139551114e-05,
+      "loss": 3.9198,
+      "step": 3523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0140661532105617e-05,
+      "loss": 3.9354,
+      "step": 3524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0130517040920808e-05,
+      "loss": 3.9331,
+      "step": 3525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.012037166774063e-05,
+      "loss": 4.0207,
+      "step": 3526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0110225414309212e-05,
+      "loss": 4.0344,
+      "step": 3527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0100078282370803e-05,
+      "loss": 4.0426,
+      "step": 3528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.008993027366981e-05,
+      "loss": 3.9095,
+      "step": 3529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0079781389950802e-05,
+      "loss": 3.855,
+      "step": 3530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0069631632958474e-05,
+      "loss": 3.9449,
+      "step": 3531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0059481004437694e-05,
+      "loss": 4.0079,
+      "step": 3532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0049329506133482e-05,
+      "loss": 3.8946,
+      "step": 3533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0039177139790993e-05,
+      "loss": 4.1,
+      "step": 3534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.002902390715553e-05,
+      "loss": 3.9857,
+      "step": 3535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0018869809972556e-05,
+      "loss": 3.897,
+      "step": 3536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0008714849987674e-05,
+      "loss": 3.9894,
+      "step": 3537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.999855902894665e-05,
+      "loss": 4.0041,
+      "step": 3538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.998840234859538e-05,
+      "loss": 3.9509,
+      "step": 3539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9978244810679913e-05,
+      "loss": 3.984,
+      "step": 3540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9968086416946446e-05,
+      "loss": 3.9507,
+      "step": 3541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9957927169141336e-05,
+      "loss": 3.861,
+      "step": 3542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.994776706901107e-05,
+      "loss": 3.9186,
+      "step": 3543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.993760611830228e-05,
+      "loss": 3.9353,
+      "step": 3544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9927444318761755e-05,
+      "loss": 3.9229,
+      "step": 3545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9917281672136433e-05,
+      "loss": 3.8506,
+      "step": 3546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9907118180173384e-05,
+      "loss": 3.8961,
+      "step": 3547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9896953844619835e-05,
+      "loss": 4.0841,
+      "step": 3548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9886788667223147e-05,
+      "loss": 3.9398,
+      "step": 3549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.987662264973083e-05,
+      "loss": 4.0939,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9866455793890547e-05,
+      "loss": 4.0162,
+      "step": 3551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9856288101450096e-05,
+      "loss": 3.9688,
+      "step": 3552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.984611957415743e-05,
+      "loss": 3.9198,
+      "step": 3553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9835950213760616e-05,
+      "loss": 3.9349,
+      "step": 3554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9825780022007905e-05,
+      "loss": 3.9627,
+      "step": 3555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.981560900064766e-05,
+      "loss": 4.0811,
+      "step": 3556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9805437151428405e-05,
+      "loss": 3.9751,
+      "step": 3557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.979526447609879e-05,
+      "loss": 3.9057,
+      "step": 3558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9785090976407614e-05,
+      "loss": 3.8953,
+      "step": 3559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9774916654103835e-05,
+      "loss": 4.0497,
+      "step": 3560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9764741510936527e-05,
+      "loss": 4.0189,
+      "step": 3561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.975456554865491e-05,
+      "loss": 3.9315,
+      "step": 3562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9744388769008365e-05,
+      "loss": 4.0096,
+      "step": 3563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9734211173746374e-05,
+      "loss": 3.8933,
+      "step": 3564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.972403276461861e-05,
+      "loss": 4.093,
+      "step": 3565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.971385354337485e-05,
+      "loss": 4.0853,
+      "step": 3566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9703673511765012e-05,
+      "loss": 3.9301,
+      "step": 3567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9693492671539163e-05,
+      "loss": 3.9144,
+      "step": 3568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.968331102444752e-05,
+      "loss": 3.9293,
+      "step": 3569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9673128572240415e-05,
+      "loss": 4.0126,
+      "step": 3570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.966294531666834e-05,
+      "loss": 3.9762,
+      "step": 3571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9652761259481898e-05,
+      "loss": 4.0281,
+      "step": 3572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.964257640243186e-05,
+      "loss": 4.0467,
+      "step": 3573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.963239074726912e-05,
+      "loss": 3.9716,
+      "step": 3574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9622204295744705e-05,
+      "loss": 3.8735,
+      "step": 3575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.961201704960978e-05,
+      "loss": 3.9947,
+      "step": 3576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960182901061567e-05,
+      "loss": 4.0618,
+      "step": 3577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9591640180513793e-05,
+      "loss": 3.9557,
+      "step": 3578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9581450561055747e-05,
+      "loss": 3.9732,
+      "step": 3579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9571260153993226e-05,
+      "loss": 3.8199,
+      "step": 3580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9561068961078097e-05,
+      "loss": 3.9164,
+      "step": 3581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9550876984062335e-05,
+      "loss": 3.9395,
+      "step": 3582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9540684224698057e-05,
+      "loss": 3.953,
+      "step": 3583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9530490684737532e-05,
+      "loss": 3.9749,
+      "step": 3584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9520296365933132e-05,
+      "loss": 3.9599,
+      "step": 3585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9510101270037377e-05,
+      "loss": 3.9909,
+      "step": 3586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.949990539880293e-05,
+      "loss": 3.9952,
+      "step": 3587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9489708753982582e-05,
+      "loss": 3.9979,
+      "step": 3588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9479511337329253e-05,
+      "loss": 3.8922,
+      "step": 3589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9469313150595985e-05,
+      "loss": 4.0358,
+      "step": 3590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9459114195535976e-05,
+      "loss": 4.0286,
+      "step": 3591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9448914473902544e-05,
+      "loss": 3.9897,
+      "step": 3592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.943871398744914e-05,
+      "loss": 3.9785,
+      "step": 3593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9428512737929338e-05,
+      "loss": 3.945,
+      "step": 3594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9418310727096866e-05,
+      "loss": 3.9141,
+      "step": 3595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9408107956705543e-05,
+      "loss": 3.7347,
+      "step": 3596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.939790442850937e-05,
+      "loss": 3.9272,
+      "step": 3597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9387700144262443e-05,
+      "loss": 3.8652,
+      "step": 3598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9377495105718993e-05,
+      "loss": 4.0612,
+      "step": 3599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9367289314633382e-05,
+      "loss": 3.9308,
+      "step": 3600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9357082772760102e-05,
+      "loss": 3.9632,
+      "step": 3601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9346875481853798e-05,
+      "loss": 3.8656,
+      "step": 3602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9336667443669196e-05,
+      "loss": 3.9637,
+      "step": 3603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9326458659961187e-05,
+      "loss": 3.9131,
+      "step": 3604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9316249132484775e-05,
+      "loss": 4.0761,
+      "step": 3605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.93060388629951e-05,
+      "loss": 3.9357,
+      "step": 3606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9295827853247427e-05,
+      "loss": 3.9496,
+      "step": 3607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9285616104997144e-05,
+      "loss": 3.9637,
+      "step": 3608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9275403619999768e-05,
+      "loss": 4.0131,
+      "step": 3609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9265190400010945e-05,
+      "loss": 3.8285,
+      "step": 3610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.925497644678645e-05,
+      "loss": 3.975,
+      "step": 3611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9244761762082168e-05,
+      "loss": 3.8759,
+      "step": 3612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9234546347654135e-05,
+      "loss": 3.8984,
+      "step": 3613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.922433020525849e-05,
+      "loss": 4.0083,
+      "step": 3614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.921411333665151e-05,
+      "loss": 4.0692,
+      "step": 3615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9203895743589597e-05,
+      "loss": 3.9164,
+      "step": 3616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9193677427829276e-05,
+      "loss": 3.9423,
+      "step": 3617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.918345839112717e-05,
+      "loss": 3.8831,
+      "step": 3618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.917323863524008e-05,
+      "loss": 3.9315,
+      "step": 3619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9163018161924887e-05,
+      "loss": 3.8827,
+      "step": 3620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9152796972938617e-05,
+      "loss": 3.7965,
+      "step": 3621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9142575070038392e-05,
+      "loss": 3.9592,
+      "step": 3622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9132352454981492e-05,
+      "loss": 3.9809,
+      "step": 3623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.91221291295253e-05,
+      "loss": 3.8879,
+      "step": 3624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9111905095427327e-05,
+      "loss": 3.8847,
+      "step": 3625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9101680354445187e-05,
+      "loss": 3.9547,
+      "step": 3626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9091454908336657e-05,
+      "loss": 3.8932,
+      "step": 3627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9081228758859584e-05,
+      "loss": 3.9191,
+      "step": 3628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9071001907771982e-05,
+      "loss": 4.0246,
+      "step": 3629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.906077435683195e-05,
+      "loss": 3.9034,
+      "step": 3630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9050546107797738e-05,
+      "loss": 3.9227,
+      "step": 3631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.904031716242769e-05,
+      "loss": 3.9557,
+      "step": 3632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.903008752248027e-05,
+      "loss": 4.0191,
+      "step": 3633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9019857189714095e-05,
+      "loss": 3.9401,
+      "step": 3634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.900962616588787e-05,
+      "loss": 3.9013,
+      "step": 3635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8999394452760414e-05,
+      "loss": 3.9148,
+      "step": 3636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8989162052090696e-05,
+      "loss": 3.9396,
+      "step": 3637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8978928965637768e-05,
+      "loss": 4.04,
+      "step": 3638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.896869519516083e-05,
+      "loss": 4.0594,
+      "step": 3639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.895846074241917e-05,
+      "loss": 4.0125,
+      "step": 3640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8948225609172225e-05,
+      "loss": 4.0049,
+      "step": 3641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8937989797179522e-05,
+      "loss": 3.988,
+      "step": 3642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8927753308200723e-05,
+      "loss": 4.0076,
+      "step": 3643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8917516143995594e-05,
+      "loss": 4.0189,
+      "step": 3644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8907278306324027e-05,
+      "loss": 3.9757,
+      "step": 3645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8897039796946024e-05,
+      "loss": 3.9518,
+      "step": 3646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.888680061762169e-05,
+      "loss": 3.9481,
+      "step": 3647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8876560770111278e-05,
+      "loss": 3.999,
+      "step": 3648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8866320256175126e-05,
+      "loss": 3.9048,
+      "step": 3649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8856079077573706e-05,
+      "loss": 3.9332,
+      "step": 3650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8845837236067586e-05,
+      "loss": 3.9131,
+      "step": 3651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8835594733417453e-05,
+      "loss": 4.0318,
+      "step": 3652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.882535157138413e-05,
+      "loss": 3.9173,
+      "step": 3653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8815107751728514e-05,
+      "loss": 4.0427,
+      "step": 3654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.880486327621166e-05,
+      "loss": 3.9975,
+      "step": 3655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8794618146594687e-05,
+      "loss": 3.9517,
+      "step": 3656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8784372364638873e-05,
+      "loss": 3.995,
+      "step": 3657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8774125932105567e-05,
+      "loss": 4.0305,
+      "step": 3658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.876387885075627e-05,
+      "loss": 3.9475,
+      "step": 3659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8753631122352564e-05,
+      "loss": 4.0108,
+      "step": 3660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8743382748656146e-05,
+      "loss": 4.0,
+      "step": 3661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8733133731428835e-05,
+      "loss": 3.9948,
+      "step": 3662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.872288407243256e-05,
+      "loss": 4.0715,
+      "step": 3663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.871263377342936e-05,
+      "loss": 3.9721,
+      "step": 3664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8702382836181363e-05,
+      "loss": 4.0198,
+      "step": 3665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.869213126245084e-05,
+      "loss": 3.9542,
+      "step": 3666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.868187905400015e-05,
+      "loss": 3.9172,
+      "step": 3667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8671626212591774e-05,
+      "loss": 3.8466,
+      "step": 3668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8661372739988274e-05,
+      "loss": 3.969,
+      "step": 3669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8651118637952358e-05,
+      "loss": 3.9747,
+      "step": 3670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8640863908246823e-05,
+      "loss": 4.0144,
+      "step": 3671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.863060855263457e-05,
+      "loss": 4.062,
+      "step": 3672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.862035257287862e-05,
+      "loss": 3.9412,
+      "step": 3673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8610095970742094e-05,
+      "loss": 3.9129,
+      "step": 3674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8599838747988217e-05,
+      "loss": 3.9276,
+      "step": 3675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8589580906380324e-05,
+      "loss": 3.9977,
+      "step": 3676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8579322447681862e-05,
+      "loss": 4.1137,
+      "step": 3677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.856906337365638e-05,
+      "loss": 3.934,
+      "step": 3678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.855880368606752e-05,
+      "loss": 3.9308,
+      "step": 3679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.854854338667905e-05,
+      "loss": 4.0745,
+      "step": 3680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.853828247725484e-05,
+      "loss": 3.8886,
+      "step": 3681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.852802095955885e-05,
+      "loss": 4.1402,
+      "step": 3682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8517758835355153e-05,
+      "loss": 4.0291,
+      "step": 3683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.850749610640793e-05,
+      "loss": 3.9037,
+      "step": 3684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8497232774481468e-05,
+      "loss": 4.0877,
+      "step": 3685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8486968841340157e-05,
+      "loss": 4.0062,
+      "step": 3686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8476704308748465e-05,
+      "loss": 3.8767,
+      "step": 3687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8466439178471e-05,
+      "loss": 3.8761,
+      "step": 3688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8456173452272456e-05,
+      "loss": 4.0828,
+      "step": 3689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.844590713191762e-05,
+      "loss": 3.9237,
+      "step": 3690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.843564021917141e-05,
+      "loss": 3.9098,
+      "step": 3691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8425372715798816e-05,
+      "loss": 4.0826,
+      "step": 3692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8415104623564937e-05,
+      "loss": 3.9419,
+      "step": 3693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.840483594423498e-05,
+      "loss": 4.062,
+      "step": 3694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8394566679574257e-05,
+      "loss": 4.102,
+      "step": 3695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.838429683134817e-05,
+      "loss": 3.9573,
+      "step": 3696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8374026401322222e-05,
+      "loss": 3.9968,
+      "step": 3697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8363755391262016e-05,
+      "loss": 3.9416,
+      "step": 3698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8353483802933267e-05,
+      "loss": 3.9011,
+      "step": 3699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.834321163810178e-05,
+      "loss": 3.8764,
+      "step": 3700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8332938898533447e-05,
+      "loss": 3.8943,
+      "step": 3701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.832266558599428e-05,
+      "loss": 3.8729,
+      "step": 3702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.831239170225038e-05,
+      "loss": 3.8351,
+      "step": 3703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.830211724906794e-05,
+      "loss": 4.081,
+      "step": 3704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8291842228213272e-05,
+      "loss": 3.9376,
+      "step": 3705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8281566641452763e-05,
+      "loss": 4.024,
+      "step": 3706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8271290490552898e-05,
+      "loss": 3.96,
+      "step": 3707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8261013777280277e-05,
+      "loss": 4.1374,
+      "step": 3708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8250736503401586e-05,
+      "loss": 3.9241,
+      "step": 3709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.82404586706836e-05,
+      "loss": 3.812,
+      "step": 3710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8230180280893203e-05,
+      "loss": 4.0693,
+      "step": 3711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8219901335797366e-05,
+      "loss": 4.0006,
+      "step": 3712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.820962183716317e-05,
+      "loss": 4.1091,
+      "step": 3713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8199341786757766e-05,
+      "loss": 3.8453,
+      "step": 3714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8189061186348415e-05,
+      "loss": 4.0977,
+      "step": 3715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8178780037702474e-05,
+      "loss": 4.0457,
+      "step": 3716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8168498342587396e-05,
+      "loss": 3.9418,
+      "step": 3717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.815821610277073e-05,
+      "loss": 3.9774,
+      "step": 3718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8147933320020088e-05,
+      "loss": 3.9307,
+      "step": 3719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.813764999610322e-05,
+      "loss": 3.9542,
+      "step": 3720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.812736613278794e-05,
+      "loss": 3.8822,
+      "step": 3721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8117081731842172e-05,
+      "loss": 4.0087,
+      "step": 3722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.810679679503392e-05,
+      "loss": 3.9203,
+      "step": 3723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8096511324131273e-05,
+      "loss": 4.0019,
+      "step": 3724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8086225320902437e-05,
+      "loss": 4.067,
+      "step": 3725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8075938787115684e-05,
+      "loss": 3.9147,
+      "step": 3726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8065651724539404e-05,
+      "loss": 3.9979,
+      "step": 3727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8055364134942046e-05,
+      "loss": 3.9127,
+      "step": 3728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8045076020092175e-05,
+      "loss": 3.9609,
+      "step": 3729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8034787381758433e-05,
+      "loss": 4.0123,
+      "step": 3730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8024498221709556e-05,
+      "loss": 3.8686,
+      "step": 3731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8014208541714377e-05,
+      "loss": 3.9798,
+      "step": 3732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8003918343541802e-05,
+      "loss": 4.1657,
+      "step": 3733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.799362762896084e-05,
+      "loss": 3.9605,
+      "step": 3734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7983336399740583e-05,
+      "loss": 3.9849,
+      "step": 3735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7973044657650216e-05,
+      "loss": 3.8867,
+      "step": 3736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7962752404459003e-05,
+      "loss": 4.0491,
+      "step": 3737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7952459641936308e-05,
+      "loss": 4.0719,
+      "step": 3738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.794216637185157e-05,
+      "loss": 3.9518,
+      "step": 3739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.793187259597432e-05,
+      "loss": 3.963,
+      "step": 3740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7921578316074183e-05,
+      "loss": 3.9312,
+      "step": 3741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7911283533920867e-05,
+      "loss": 4.0087,
+      "step": 3742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7900988251284162e-05,
+      "loss": 3.9975,
+      "step": 3743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.789069246993394e-05,
+      "loss": 3.8778,
+      "step": 3744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7880396191640175e-05,
+      "loss": 4.0074,
+      "step": 3745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7870099418172912e-05,
+      "loss": 4.0888,
+      "step": 3746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7859802151302283e-05,
+      "loss": 4.0146,
+      "step": 3747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.784950439279852e-05,
+      "loss": 4.0747,
+      "step": 3748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7839206144431907e-05,
+      "loss": 3.9679,
+      "step": 3749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.782890740797285e-05,
+      "loss": 4.0324,
+      "step": 3750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7818608185191808e-05,
+      "loss": 4.0562,
+      "step": 3751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7808308477859347e-05,
+      "loss": 4.0065,
+      "step": 3752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7798008287746097e-05,
+      "loss": 3.9838,
+      "step": 3753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7787707616622788e-05,
+      "loss": 3.9453,
+      "step": 3754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.777740646626022e-05,
+      "loss": 4.0049,
+      "step": 3755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7767104838429286e-05,
+      "loss": 4.1022,
+      "step": 3756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7756802734900954e-05,
+      "loss": 4.0678,
+      "step": 3757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.774650015744626e-05,
+      "loss": 4.0041,
+      "step": 3758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.773619710783636e-05,
+      "loss": 3.9784,
+      "step": 3759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7725893587842457e-05,
+      "loss": 4.0781,
+      "step": 3760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7715589599235836e-05,
+      "loss": 4.1015,
+      "step": 3761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7705285143787884e-05,
+      "loss": 4.0002,
+      "step": 3762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7694980223270055e-05,
+      "loss": 4.0454,
+      "step": 3763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7684674839453888e-05,
+      "loss": 4.1303,
+      "step": 3764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7674368994110982e-05,
+      "loss": 3.9914,
+      "step": 3765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7664062689013053e-05,
+      "loss": 4.0202,
+      "step": 3766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7653755925931857e-05,
+      "loss": 4.0497,
+      "step": 3767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7643448706639254e-05,
+      "loss": 3.9941,
+      "step": 3768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7633141032907166e-05,
+      "loss": 4.0983,
+      "step": 3769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7622832906507623e-05,
+      "loss": 3.8858,
+      "step": 3770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7612524329212687e-05,
+      "loss": 3.966,
+      "step": 3771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7602215302794532e-05,
+      "loss": 4.0794,
+      "step": 3772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.75919058290254e-05,
+      "loss": 4.1317,
+      "step": 3773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.758159590967761e-05,
+      "loss": 3.9834,
+      "step": 3774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7571285546523557e-05,
+      "loss": 3.8862,
+      "step": 3775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.75609747413357e-05,
+      "loss": 4.0667,
+      "step": 3776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.75506634958866e-05,
+      "loss": 4.0235,
+      "step": 3777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7540351811948877e-05,
+      "loss": 4.0564,
+      "step": 3778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7530039691295222e-05,
+      "loss": 3.9379,
+      "step": 3779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7519727135698426e-05,
+      "loss": 3.9397,
+      "step": 3780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7509414146931312e-05,
+      "loss": 4.0156,
+      "step": 3781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7499100726766817e-05,
+      "loss": 3.9776,
+      "step": 3782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.748878687697794e-05,
+      "loss": 3.9285,
+      "step": 3783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7478472599337747e-05,
+      "loss": 4.0097,
+      "step": 3784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.746815789561938e-05,
+      "loss": 4.1244,
+      "step": 3785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7457842767596048e-05,
+      "loss": 3.9966,
+      "step": 3786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7447527217041058e-05,
+      "loss": 4.0409,
+      "step": 3787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7437211245727767e-05,
+      "loss": 3.9773,
+      "step": 3788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.742689485542961e-05,
+      "loss": 4.0112,
+      "step": 3789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7416578047920077e-05,
+      "loss": 3.8251,
+      "step": 3790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7406260824972773e-05,
+      "loss": 3.8467,
+      "step": 3791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7395943188361344e-05,
+      "loss": 4.0431,
+      "step": 3792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.73856251398595e-05,
+      "loss": 4.0229,
+      "step": 3793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7375306681241035e-05,
+      "loss": 3.9538,
+      "step": 3794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7364987814279814e-05,
+      "loss": 3.8578,
+      "step": 3795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7354668540749774e-05,
+      "loss": 3.9096,
+      "step": 3796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7344348862424908e-05,
+      "loss": 4.0012,
+      "step": 3797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7334028781079302e-05,
+      "loss": 3.976,
+      "step": 3798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7323708298487095e-05,
+      "loss": 3.9661,
+      "step": 3799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.731338741642248e-05,
+      "loss": 3.9677,
+      "step": 3800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.730306613665975e-05,
+      "loss": 3.982,
+      "step": 3801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7292744460973257e-05,
+      "loss": 3.8357,
+      "step": 3802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7282422391137414e-05,
+      "loss": 3.9381,
+      "step": 3803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7272099928926698e-05,
+      "loss": 3.8949,
+      "step": 3804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.726177707611566e-05,
+      "loss": 3.9331,
+      "step": 3805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.725145383447893e-05,
+      "loss": 4.006,
+      "step": 3806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7241130205791182e-05,
+      "loss": 3.8607,
+      "step": 3807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7230806191827162e-05,
+      "loss": 3.9368,
+      "step": 3808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.72204817943617e-05,
+      "loss": 3.9677,
+      "step": 3809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7210157015169684e-05,
+      "loss": 3.8172,
+      "step": 3810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7199831856026043e-05,
+      "loss": 3.8614,
+      "step": 3811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7189506318705814e-05,
+      "loss": 3.9141,
+      "step": 3812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.717918040498406e-05,
+      "loss": 3.9399,
+      "step": 3813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7168854116635933e-05,
+      "loss": 4.1138,
+      "step": 3814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7158527455436634e-05,
+      "loss": 3.9593,
+      "step": 3815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.714820042316145e-05,
+      "loss": 3.9187,
+      "step": 3816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.713787302158571e-05,
+      "loss": 3.8453,
+      "step": 3817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7127545252484805e-05,
+      "loss": 3.8865,
+      "step": 3818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.711721711763421e-05,
+      "loss": 3.9183,
+      "step": 3819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.710688861880945e-05,
+      "loss": 4.039,
+      "step": 3820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7096559757786112e-05,
+      "loss": 3.9949,
+      "step": 3821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.708623053633984e-05,
+      "loss": 4.0151,
+      "step": 3822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7075900956246353e-05,
+      "loss": 3.9276,
+      "step": 3823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7065571019281432e-05,
+      "loss": 3.9329,
+      "step": 3824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.705524072722091e-05,
+      "loss": 3.9984,
+      "step": 3825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.704491008184067e-05,
+      "loss": 3.9772,
+      "step": 3826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7034579084916682e-05,
+      "loss": 3.9916,
+      "step": 3827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7024247738224968e-05,
+      "loss": 4.021,
+      "step": 3828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.70139160435416e-05,
+      "loss": 3.9508,
+      "step": 3829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7003584002642712e-05,
+      "loss": 3.9533,
+      "step": 3830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6993251617304517e-05,
+      "loss": 3.9069,
+      "step": 3831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.698291888930325e-05,
+      "loss": 4.0073,
+      "step": 3832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6972585820415237e-05,
+      "loss": 4.0483,
+      "step": 3833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6962252412416862e-05,
+      "loss": 3.9125,
+      "step": 3834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6951918667084546e-05,
+      "loss": 3.8529,
+      "step": 3835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6941584586194778e-05,
+      "loss": 3.9604,
+      "step": 3836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6931250171524115e-05,
+      "loss": 4.0264,
+      "step": 3837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6920915424849157e-05,
+      "loss": 3.9744,
+      "step": 3838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6910580347946574e-05,
+      "loss": 4.0223,
+      "step": 3839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6900244942593067e-05,
+      "loss": 3.873,
+      "step": 3840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6889909210565434e-05,
+      "loss": 4.1208,
+      "step": 3841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6879573153640496e-05,
+      "loss": 4.0362,
+      "step": 3842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6869236773595145e-05,
+      "loss": 3.8557,
+      "step": 3843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.685890007220632e-05,
+      "loss": 3.8927,
+      "step": 3844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.684856305125103e-05,
+      "loss": 3.9251,
+      "step": 3845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.683822571250631e-05,
+      "loss": 3.9717,
+      "step": 3846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6827888057749278e-05,
+      "loss": 3.9038,
+      "step": 3847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.681755008875711e-05,
+      "loss": 3.8943,
+      "step": 3848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.680721180730701e-05,
+      "loss": 3.9043,
+      "step": 3849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.679687321517624e-05,
+      "loss": 4.0448,
+      "step": 3850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6786534314142136e-05,
+      "loss": 4.0529,
+      "step": 3851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6776195105982076e-05,
+      "loss": 4.0005,
+      "step": 3852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.676585559247349e-05,
+      "loss": 3.9078,
+      "step": 3853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.675551577539384e-05,
+      "loss": 3.8755,
+      "step": 3854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6745175656520686e-05,
+      "loss": 3.8418,
+      "step": 3855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6734835237631606e-05,
+      "loss": 4.0823,
+      "step": 3856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6724494520504238e-05,
+      "loss": 3.9909,
+      "step": 3857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6714153506916267e-05,
+      "loss": 3.91,
+      "step": 3858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6703812198645438e-05,
+      "loss": 3.9391,
+      "step": 3859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.669347059746954e-05,
+      "loss": 3.9121,
+      "step": 3860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6683128705166416e-05,
+      "loss": 3.9188,
+      "step": 3861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.667278652351395e-05,
+      "loss": 3.9261,
+      "step": 3862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6662444054290087e-05,
+      "loss": 3.9384,
+      "step": 3863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6652101299272818e-05,
+      "loss": 4.0423,
+      "step": 3864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6641758260240185e-05,
+      "loss": 4.0695,
+      "step": 3865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6631414938970278e-05,
+      "loss": 4.0319,
+      "step": 3866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6621071337241227e-05,
+      "loss": 3.8861,
+      "step": 3867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6610727456831218e-05,
+      "loss": 3.958,
+      "step": 3868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6600383299518487e-05,
+      "loss": 3.9401,
+      "step": 3869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6590038867081318e-05,
+      "loss": 3.8906,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.657969416129803e-05,
+      "loss": 3.8564,
+      "step": 3871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6569349183947e-05,
+      "loss": 3.9148,
+      "step": 3872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.655900393680666e-05,
+      "loss": 3.9784,
+      "step": 3873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6548658421655465e-05,
+      "loss": 3.9414,
+      "step": 3874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6538312640271933e-05,
+      "loss": 4.0703,
+      "step": 3875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.652796659443462e-05,
+      "loss": 3.9776,
+      "step": 3876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6517620285922144e-05,
+      "loss": 3.9663,
+      "step": 3877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.650727371651314e-05,
+      "loss": 3.9757,
+      "step": 3878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6496926887986324e-05,
+      "loss": 3.9364,
+      "step": 3879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6486579802120405e-05,
+      "loss": 3.9559,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6476232460694196e-05,
+      "loss": 3.9779,
+      "step": 3881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6465884865486512e-05,
+      "loss": 3.9157,
+      "step": 3882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6455537018276212e-05,
+      "loss": 4.1097,
+      "step": 3883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.644518892084224e-05,
+      "loss": 4.0157,
+      "step": 3884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.643484057496353e-05,
+      "loss": 3.8524,
+      "step": 3885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6424491982419098e-05,
+      "loss": 4.0148,
+      "step": 3886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6414143144987974e-05,
+      "loss": 4.071,
+      "step": 3887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640379406444925e-05,
+      "loss": 3.9136,
+      "step": 3888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6393444742582056e-05,
+      "loss": 3.8605,
+      "step": 3889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6383095181165553e-05,
+      "loss": 3.901,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.637274538197896e-05,
+      "loss": 4.0314,
+      "step": 3891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.636239534680151e-05,
+      "loss": 3.9278,
+      "step": 3892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6352045077412517e-05,
+      "loss": 4.0985,
+      "step": 3893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634169457559129e-05,
+      "loss": 3.9746,
+      "step": 3894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6331343843117223e-05,
+      "loss": 4.0513,
+      "step": 3895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6320992881769713e-05,
+      "loss": 4.0773,
+      "step": 3896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6310641693328205e-05,
+      "loss": 4.0809,
+      "step": 3897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6300290279572198e-05,
+      "loss": 3.9411,
+      "step": 3898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.628993864228122e-05,
+      "loss": 4.016,
+      "step": 3899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.627958678323484e-05,
+      "loss": 3.7955,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.626923470421264e-05,
+      "loss": 4.1203,
+      "step": 3901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.62588824069943e-05,
+      "loss": 4.0247,
+      "step": 3902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6248529893359473e-05,
+      "loss": 3.8833,
+      "step": 3903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.623817716508788e-05,
+      "loss": 4.0077,
+      "step": 3904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6227824223959284e-05,
+      "loss": 3.9524,
+      "step": 3905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6217471071753464e-05,
+      "loss": 3.8928,
+      "step": 3906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6207117710250257e-05,
+      "loss": 3.9196,
+      "step": 3907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6196764141229518e-05,
+      "loss": 4.0421,
+      "step": 3908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6186410366471154e-05,
+      "loss": 3.9562,
+      "step": 3909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6176056387755093e-05,
+      "loss": 3.9358,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6165702206861297e-05,
+      "loss": 3.9635,
+      "step": 3911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.615534782556978e-05,
+      "loss": 3.9765,
+      "step": 3912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6144993245660587e-05,
+      "loss": 4.0025,
+      "step": 3913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6134638468913773e-05,
+      "loss": 3.9981,
+      "step": 3914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6124283497109447e-05,
+      "loss": 4.1176,
+      "step": 3915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6113928332027765e-05,
+      "loss": 3.8035,
+      "step": 3916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6103572975448887e-05,
+      "loss": 3.7702,
+      "step": 3917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6093217429153023e-05,
+      "loss": 3.9591,
+      "step": 3918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6082861694920403e-05,
+      "loss": 4.0026,
+      "step": 3919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6072505774531303e-05,
+      "loss": 3.9309,
+      "step": 3920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6062149669766034e-05,
+      "loss": 3.9494,
+      "step": 3921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6051793382404914e-05,
+      "loss": 3.9872,
+      "step": 3922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6041436914228328e-05,
+      "loss": 3.895,
+      "step": 3923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6031080267016668e-05,
+      "loss": 3.9715,
+      "step": 3924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6020723442550343e-05,
+      "loss": 3.902,
+      "step": 3925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6010366442609824e-05,
+      "loss": 3.9133,
+      "step": 3926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.600000926897561e-05,
+      "loss": 3.9394,
+      "step": 3927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.598965192342821e-05,
+      "loss": 3.7799,
+      "step": 3928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.597929440774816e-05,
+      "loss": 4.0932,
+      "step": 3929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5968936723716053e-05,
+      "loss": 3.8728,
+      "step": 3930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5958578873112498e-05,
+      "loss": 4.1027,
+      "step": 3931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.594822085771812e-05,
+      "loss": 3.887,
+      "step": 3932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5937862679313573e-05,
+      "loss": 3.8959,
+      "step": 3933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5927504339679564e-05,
+      "loss": 3.843,
+      "step": 3934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5917145840596814e-05,
+      "loss": 3.9101,
+      "step": 3935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5906787183846053e-05,
+      "loss": 3.9849,
+      "step": 3936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.589642837120807e-05,
+      "loss": 3.9213,
+      "step": 3937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5886069404463648e-05,
+      "loss": 4.0402,
+      "step": 3938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5875710285393634e-05,
+      "loss": 3.9686,
+      "step": 3939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.586535101577886e-05,
+      "loss": 3.9944,
+      "step": 3940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.585499159740022e-05,
+      "loss": 3.9216,
+      "step": 3941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5844632032038612e-05,
+      "loss": 3.9634,
+      "step": 3942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5834272321474968e-05,
+      "loss": 3.9756,
+      "step": 3943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.582391246749023e-05,
+      "loss": 3.9993,
+      "step": 3944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5813552471865397e-05,
+      "loss": 3.8586,
+      "step": 3945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5803192336381465e-05,
+      "loss": 3.9094,
+      "step": 3946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.579283206281945e-05,
+      "loss": 3.9988,
+      "step": 3947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5782471652960416e-05,
+      "loss": 3.9996,
+      "step": 3948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5772111108585433e-05,
+      "loss": 4.0278,
+      "step": 3949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5761750431475606e-05,
+      "loss": 3.838,
+      "step": 3950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5751389623412043e-05,
+      "loss": 3.8869,
+      "step": 3951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5741028686175888e-05,
+      "loss": 3.981,
+      "step": 3952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5730667621548322e-05,
+      "loss": 4.0348,
+      "step": 3953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.572030643131051e-05,
+      "loss": 3.9171,
+      "step": 3954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5709945117243677e-05,
+      "loss": 3.9817,
+      "step": 3955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5699583681129053e-05,
+      "loss": 3.9943,
+      "step": 3956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.568922212474788e-05,
+      "loss": 3.9739,
+      "step": 3957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5678860449881424e-05,
+      "loss": 3.904,
+      "step": 3958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5668498658311003e-05,
+      "loss": 4.0578,
+      "step": 3959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.565813675181791e-05,
+      "loss": 3.9823,
+      "step": 3960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5647774732183473e-05,
+      "loss": 3.8976,
+      "step": 3961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5637412601189043e-05,
+      "loss": 3.9779,
+      "step": 3962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.562705036061601e-05,
+      "loss": 4.1019,
+      "step": 3963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5616688012245747e-05,
+      "loss": 4.1779,
+      "step": 3964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5606325557859668e-05,
+      "loss": 3.9788,
+      "step": 3965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5595962999239187e-05,
+      "loss": 3.9581,
+      "step": 3966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5585600338165766e-05,
+      "loss": 4.2263,
+      "step": 3967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5575237576420862e-05,
+      "loss": 4.0522,
+      "step": 3968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5564874715785943e-05,
+      "loss": 3.8385,
+      "step": 3969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5554511758042515e-05,
+      "loss": 3.8174,
+      "step": 3970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.554414870497208e-05,
+      "loss": 3.8797,
+      "step": 3971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5533785558356175e-05,
+      "loss": 3.9492,
+      "step": 3972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5523422319976352e-05,
+      "loss": 4.0064,
+      "step": 3973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5513058991614163e-05,
+      "loss": 4.0572,
+      "step": 3974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.550269557505118e-05,
+      "loss": 3.7921,
+      "step": 3975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5492332072069003e-05,
+      "loss": 3.9701,
+      "step": 3976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5481968484449236e-05,
+      "loss": 3.9187,
+      "step": 3977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5471604813973505e-05,
+      "loss": 3.9959,
+      "step": 3978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.546124106242343e-05,
+      "loss": 4.0563,
+      "step": 3979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5450877231580665e-05,
+      "loss": 4.0305,
+      "step": 3980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.544051332322689e-05,
+      "loss": 3.9794,
+      "step": 3981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5430149339143762e-05,
+      "loss": 4.0521,
+      "step": 3982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5419785281112972e-05,
+      "loss": 3.9468,
+      "step": 3983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5409421150916235e-05,
+      "loss": 3.8384,
+      "step": 3984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5399056950335244e-05,
+      "loss": 3.8766,
+      "step": 3985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5388692681151747e-05,
+      "loss": 3.9313,
+      "step": 3986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5378328345147468e-05,
+      "loss": 3.9013,
+      "step": 3987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.536796394410416e-05,
+      "loss": 3.9742,
+      "step": 3988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5357599479803585e-05,
+      "loss": 3.9006,
+      "step": 3989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5347234954027503e-05,
+      "loss": 3.8717,
+      "step": 3990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5336870368557726e-05,
+      "loss": 3.998,
+      "step": 3991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.532650572517602e-05,
+      "loss": 3.9032,
+      "step": 3992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5316141025664192e-05,
+      "loss": 3.8416,
+      "step": 3993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5305776271804055e-05,
+      "loss": 4.0578,
+      "step": 3994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5295411465377433e-05,
+      "loss": 3.9552,
+      "step": 3995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.528504660816616e-05,
+      "loss": 4.0347,
+      "step": 3996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.527468170195207e-05,
+      "loss": 3.9588,
+      "step": 3997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.526431674851701e-05,
+      "loss": 4.0328,
+      "step": 3998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.525395174964283e-05,
+      "loss": 4.1715,
+      "step": 3999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.524358670711141e-05,
+      "loss": 4.1672,
+      "step": 4000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5233221622704604e-05,
+      "loss": 3.9548,
+      "step": 4001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.52228564982043e-05,
+      "loss": 3.9426,
+      "step": 4002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.521249133539238e-05,
+      "loss": 3.9748,
+      "step": 4003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5202126136050742e-05,
+      "loss": 3.9099,
+      "step": 4004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.519176090196127e-05,
+      "loss": 3.9379,
+      "step": 4005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.518139563490588e-05,
+      "loss": 3.9203,
+      "step": 4006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5171030336666478e-05,
+      "loss": 3.9569,
+      "step": 4007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.516066500902497e-05,
+      "loss": 3.787,
+      "step": 4008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5150299653763294e-05,
+      "loss": 3.7855,
+      "step": 4009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.513993427266336e-05,
+      "loss": 3.9394,
+      "step": 4010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5129568867507102e-05,
+      "loss": 4.0551,
+      "step": 4011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5119203440076455e-05,
+      "loss": 4.0024,
+      "step": 4012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5108837992153354e-05,
+      "loss": 4.0408,
+      "step": 4013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5098472525519746e-05,
+      "loss": 4.0465,
+      "step": 4014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5088107041957564e-05,
+      "loss": 4.0299,
+      "step": 4015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5077741543248763e-05,
+      "loss": 3.9877,
+      "step": 4016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5067376031175282e-05,
+      "loss": 4.0596,
+      "step": 4017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.505701050751909e-05,
+      "loss": 3.8838,
+      "step": 4018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.504664497406212e-05,
+      "loss": 3.8713,
+      "step": 4019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5036279432586352e-05,
+      "loss": 3.9648,
+      "step": 4020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5025913884873727e-05,
+      "loss": 3.9992,
+      "step": 4021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.501554833270619e-05,
+      "loss": 4.0525,
+      "step": 4022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5005182777865727e-05,
+      "loss": 3.9177,
+      "step": 4023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4994817222134275e-05,
+      "loss": 3.9691,
+      "step": 4024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4984451667293813e-05,
+      "loss": 3.9448,
+      "step": 4025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4974086115126282e-05,
+      "loss": 3.9786,
+      "step": 4026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4963720567413654e-05,
+      "loss": 4.1321,
+      "step": 4027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.495335502593788e-05,
+      "loss": 3.9395,
+      "step": 4028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4942989492480915e-05,
+      "loss": 3.9504,
+      "step": 4029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4932623968824723e-05,
+      "loss": 3.8922,
+      "step": 4030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.492225845675125e-05,
+      "loss": 3.9434,
+      "step": 4031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4911892958042442e-05,
+      "loss": 4.0326,
+      "step": 4032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.490152747448026e-05,
+      "loss": 3.9895,
+      "step": 4033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4891162007846645e-05,
+      "loss": 3.9941,
+      "step": 4034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4880796559923547e-05,
+      "loss": 4.0939,
+      "step": 4035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4870431132492904e-05,
+      "loss": 4.0124,
+      "step": 4036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4860065727336645e-05,
+      "loss": 4.0517,
+      "step": 4037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4849700346236715e-05,
+      "loss": 3.8688,
+      "step": 4038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4839334990975037e-05,
+      "loss": 3.9422,
+      "step": 4039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4828969663333535e-05,
+      "loss": 3.9863,
+      "step": 4040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4818604365094132e-05,
+      "loss": 4.1297,
+      "step": 4041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4808239098038745e-05,
+      "loss": 4.0743,
+      "step": 4042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4797873863949264e-05,
+      "loss": 4.0079,
+      "step": 4043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.478750866460762e-05,
+      "loss": 3.9435,
+      "step": 4044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4777143501795702e-05,
+      "loss": 3.9781,
+      "step": 4045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.47667783772954e-05,
+      "loss": 3.963,
+      "step": 4046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4756413292888595e-05,
+      "loss": 3.9624,
+      "step": 4047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4746048250357174e-05,
+      "loss": 4.0018,
+      "step": 4048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4735683251483002e-05,
+      "loss": 4.0785,
+      "step": 4049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.472531829804794e-05,
+      "loss": 3.9147,
+      "step": 4050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4714953391833848e-05,
+      "loss": 3.8651,
+      "step": 4051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4704588534622562e-05,
+      "loss": 3.9915,
+      "step": 4052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4694223728195947e-05,
+      "loss": 4.0633,
+      "step": 4053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4683858974335817e-05,
+      "loss": 4.0723,
+      "step": 4054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4673494274823987e-05,
+      "loss": 3.9571,
+      "step": 4055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.466312963144228e-05,
+      "loss": 4.1461,
+      "step": 4056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.46527650459725e-05,
+      "loss": 4.0763,
+      "step": 4057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4642400520196424e-05,
+      "loss": 4.0197,
+      "step": 4058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4632036055895852e-05,
+      "loss": 3.8791,
+      "step": 4059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4621671654852545e-05,
+      "loss": 3.8951,
+      "step": 4060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.461130731884826e-05,
+      "loss": 4.0913,
+      "step": 4061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4600943049664755e-05,
+      "loss": 3.8254,
+      "step": 4062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4590578849083774e-05,
+      "loss": 4.0294,
+      "step": 4063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.458021471888703e-05,
+      "loss": 3.9984,
+      "step": 4064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.456985066085624e-05,
+      "loss": 3.9851,
+      "step": 4065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4559486676773117e-05,
+      "loss": 4.0485,
+      "step": 4066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.454912276841934e-05,
+      "loss": 3.9712,
+      "step": 4067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.453875893757658e-05,
+      "loss": 4.0018,
+      "step": 4068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4528395186026508e-05,
+      "loss": 3.9453,
+      "step": 4069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4518031515550763e-05,
+      "loss": 4.0294,
+      "step": 4070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4507667927930996e-05,
+      "loss": 3.8595,
+      "step": 4071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.449730442494882e-05,
+      "loss": 3.9325,
+      "step": 4072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4486941008385843e-05,
+      "loss": 4.0711,
+      "step": 4073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4476577680023653e-05,
+      "loss": 4.0323,
+      "step": 4074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4466214441643827e-05,
+      "loss": 4.0637,
+      "step": 4075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4455851295027925e-05,
+      "loss": 4.0595,
+      "step": 4076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4445488241957498e-05,
+      "loss": 3.907,
+      "step": 4077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4435125284214073e-05,
+      "loss": 4.0189,
+      "step": 4078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.442476242357915e-05,
+      "loss": 3.9733,
+      "step": 4079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.441439966183423e-05,
+      "loss": 3.9684,
+      "step": 4080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.440403700076081e-05,
+      "loss": 4.0162,
+      "step": 4081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.439367444214034e-05,
+      "loss": 4.0518,
+      "step": 4082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4383311987754255e-05,
+      "loss": 4.0409,
+      "step": 4083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4372949639383995e-05,
+      "loss": 4.0946,
+      "step": 4084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.436258739881096e-05,
+      "loss": 3.9416,
+      "step": 4085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4352225267816536e-05,
+      "loss": 3.9147,
+      "step": 4086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4341863248182104e-05,
+      "loss": 3.9495,
+      "step": 4087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.433150134168901e-05,
+      "loss": 3.9927,
+      "step": 4088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.432113955011857e-05,
+      "loss": 3.912,
+      "step": 4089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.431077787525213e-05,
+      "loss": 3.7738,
+      "step": 4090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.430041631887095e-05,
+      "loss": 3.8967,
+      "step": 4091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4290054882756325e-05,
+      "loss": 4.1256,
+      "step": 4092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4279693568689496e-05,
+      "loss": 4.0427,
+      "step": 4093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4269332378451687e-05,
+      "loss": 4.0102,
+      "step": 4094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4258971313824118e-05,
+      "loss": 3.8911,
+      "step": 4095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4248610376587973e-05,
+      "loss": 3.9266,
+      "step": 4096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4238249568524403e-05,
+      "loss": 3.8839,
+      "step": 4097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4227888891414563e-05,
+      "loss": 3.8703,
+      "step": 4098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4217528347039583e-05,
+      "loss": 4.0417,
+      "step": 4099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4207167937180554e-05,
+      "loss": 4.0616,
+      "step": 4100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.419680766361854e-05,
+      "loss": 3.9697,
+      "step": 4101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4186447528134605e-05,
+      "loss": 4.1208,
+      "step": 4102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.417608753250977e-05,
+      "loss": 4.1813,
+      "step": 4103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.416572767852504e-05,
+      "loss": 4.1088,
+      "step": 4104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4155367967961394e-05,
+      "loss": 3.9396,
+      "step": 4105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.414500840259979e-05,
+      "loss": 3.9978,
+      "step": 4106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.413464898422114e-05,
+      "loss": 3.8552,
+      "step": 4107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4124289714606368e-05,
+      "loss": 3.9429,
+      "step": 4108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.411393059553635e-05,
+      "loss": 4.0211,
+      "step": 4109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.410357162879194e-05,
+      "loss": 4.0515,
+      "step": 4110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4093212816153952e-05,
+      "loss": 4.0749,
+      "step": 4111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.408285415940319e-05,
+      "loss": 4.122,
+      "step": 4112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.407249566032044e-05,
+      "loss": 4.0228,
+      "step": 4113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4062137320686436e-05,
+      "loss": 3.9972,
+      "step": 4114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4051779142281892e-05,
+      "loss": 3.9389,
+      "step": 4115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4041421126887515e-05,
+      "loss": 3.8793,
+      "step": 4116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4031063276283946e-05,
+      "loss": 3.8973,
+      "step": 4117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4020705592251843e-05,
+      "loss": 3.9928,
+      "step": 4118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.40103480765718e-05,
+      "loss": 3.9146,
+      "step": 4119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3999990731024396e-05,
+      "loss": 3.9987,
+      "step": 4120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3989633557390182e-05,
+      "loss": 3.9178,
+      "step": 4121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3979276557449666e-05,
+      "loss": 3.8698,
+      "step": 4122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3968919732983348e-05,
+      "loss": 3.9711,
+      "step": 4123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3958563085771678e-05,
+      "loss": 3.7583,
+      "step": 4124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.394820661759509e-05,
+      "loss": 3.7321,
+      "step": 4125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3937850330233965e-05,
+      "loss": 4.0107,
+      "step": 4126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3927494225468693e-05,
+      "loss": 3.9755,
+      "step": 4127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.39171383050796e-05,
+      "loss": 4.0401,
+      "step": 4128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.390678257084698e-05,
+      "loss": 4.008,
+      "step": 4129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3896427024551115e-05,
+      "loss": 4.0011,
+      "step": 4130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.388607166797224e-05,
+      "loss": 3.9473,
+      "step": 4131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.387571650289056e-05,
+      "loss": 4.0976,
+      "step": 4132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3865361531086236e-05,
+      "loss": 4.0881,
+      "step": 4133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3855006754339426e-05,
+      "loss": 3.9588,
+      "step": 4134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3844652174430217e-05,
+      "loss": 4.0514,
+      "step": 4135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.383429779313871e-05,
+      "loss": 4.0602,
+      "step": 4136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3823943612244913e-05,
+      "loss": 3.9742,
+      "step": 4137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3813589633528855e-05,
+      "loss": 3.9721,
+      "step": 4138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3803235858770488e-05,
+      "loss": 3.9731,
+      "step": 4139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.379288228974975e-05,
+      "loss": 3.9046,
+      "step": 4140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3782528928246542e-05,
+      "loss": 3.9114,
+      "step": 4141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3772175776040728e-05,
+      "loss": 4.075,
+      "step": 4142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3761822834912126e-05,
+      "loss": 3.7756,
+      "step": 4143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.375147010664053e-05,
+      "loss": 3.8332,
+      "step": 4144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.37411175930057e-05,
+      "loss": 3.9725,
+      "step": 4145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.373076529578736e-05,
+      "loss": 3.9502,
+      "step": 4146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3720413216765167e-05,
+      "loss": 3.997,
+      "step": 4147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3710061357718783e-05,
+      "loss": 4.0663,
+      "step": 4148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3699709720427808e-05,
+      "loss": 3.9491,
+      "step": 4149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3689358306671798e-05,
+      "loss": 3.8861,
+      "step": 4150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36790071182303e-05,
+      "loss": 4.0338,
+      "step": 4151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3668656156882786e-05,
+      "loss": 4.085,
+      "step": 4152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3658305424408718e-05,
+      "loss": 4.1035,
+      "step": 4153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.364795492258749e-05,
+      "loss": 3.9543,
+      "step": 4154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.363760465319849e-05,
+      "loss": 4.0437,
+      "step": 4155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.362725461802105e-05,
+      "loss": 3.9813,
+      "step": 4156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3616904818834453e-05,
+      "loss": 3.9193,
+      "step": 4157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.360655525741795e-05,
+      "loss": 3.9623,
+      "step": 4158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3596205935550756e-05,
+      "loss": 4.0311,
+      "step": 4159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3585856855012035e-05,
+      "loss": 3.9945,
+      "step": 4160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.357550801758091e-05,
+      "loss": 4.0864,
+      "step": 4161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3565159425036478e-05,
+      "loss": 3.936,
+      "step": 4162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3554811079157764e-05,
+      "loss": 4.0532,
+      "step": 4163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.354446298172379e-05,
+      "loss": 4.0393,
+      "step": 4164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3534115134513497e-05,
+      "loss": 4.0704,
+      "step": 4165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.352376753930581e-05,
+      "loss": 3.9338,
+      "step": 4166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3513420197879597e-05,
+      "loss": 3.9445,
+      "step": 4167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3503073112013685e-05,
+      "loss": 3.9962,
+      "step": 4168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.349272628348686e-05,
+      "loss": 4.0698,
+      "step": 4169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3482379714077865e-05,
+      "loss": 4.0888,
+      "step": 4170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3472033405565388e-05,
+      "loss": 3.9264,
+      "step": 4171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3461687359728073e-05,
+      "loss": 4.0304,
+      "step": 4172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3451341578344538e-05,
+      "loss": 3.9586,
+      "step": 4173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3440996063193344e-05,
+      "loss": 3.929,
+      "step": 4174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3430650816053e-05,
+      "loss": 3.9806,
+      "step": 4175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3420305838701972e-05,
+      "loss": 4.0663,
+      "step": 4176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3409961132918688e-05,
+      "loss": 3.9662,
+      "step": 4177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.339961670048152e-05,
+      "loss": 4.0385,
+      "step": 4178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3389272543168785e-05,
+      "loss": 4.0932,
+      "step": 4179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3378928662758782e-05,
+      "loss": 3.9781,
+      "step": 4180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3368585061029724e-05,
+      "loss": 3.9583,
+      "step": 4181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3358241739759814e-05,
+      "loss": 3.8804,
+      "step": 4182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.334789870072718e-05,
+      "loss": 4.0655,
+      "step": 4183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3337555945709916e-05,
+      "loss": 3.8812,
+      "step": 4184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.332721347648606e-05,
+      "loss": 4.0468,
+      "step": 4185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3316871294833593e-05,
+      "loss": 3.9727,
+      "step": 4186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3306529402530467e-05,
+      "loss": 4.0061,
+      "step": 4187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3296187801354568e-05,
+      "loss": 3.9692,
+      "step": 4188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.328584649308374e-05,
+      "loss": 4.0,
+      "step": 4189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3275505479495768e-05,
+      "loss": 4.0781,
+      "step": 4190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.326516476236839e-05,
+      "loss": 4.0081,
+      "step": 4191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3254824343479316e-05,
+      "loss": 4.0029,
+      "step": 4192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.324448422460616e-05,
+      "loss": 4.1008,
+      "step": 4193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.323414440752652e-05,
+      "loss": 3.9061,
+      "step": 4194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.322380489401793e-05,
+      "loss": 3.96,
+      "step": 4195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3213465685857873e-05,
+      "loss": 3.9339,
+      "step": 4196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3203126784823764e-05,
+      "loss": 3.8088,
+      "step": 4197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3192788192693002e-05,
+      "loss": 4.042,
+      "step": 4198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31824499112429e-05,
+      "loss": 3.8844,
+      "step": 4199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3172111942250717e-05,
+      "loss": 3.8441,
+      "step": 4200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.316177428749369e-05,
+      "loss": 4.0451,
+      "step": 4201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.315143694874898e-05,
+      "loss": 3.9852,
+      "step": 4202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.314109992779368e-05,
+      "loss": 4.0354,
+      "step": 4203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3130763226404857e-05,
+      "loss": 3.9694,
+      "step": 4204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.312042684635951e-05,
+      "loss": 3.8814,
+      "step": 4205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3110090789434575e-05,
+      "loss": 3.8877,
+      "step": 4206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3099755057406935e-05,
+      "loss": 3.9881,
+      "step": 4207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.308941965205344e-05,
+      "loss": 4.0628,
+      "step": 4208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3079084575150845e-05,
+      "loss": 3.9154,
+      "step": 4209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3068749828475887e-05,
+      "loss": 3.9765,
+      "step": 4210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3058415413805228e-05,
+      "loss": 4.028,
+      "step": 4211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.304808133291546e-05,
+      "loss": 3.9856,
+      "step": 4212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3037747587583144e-05,
+      "loss": 4.0501,
+      "step": 4213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.302741417958477e-05,
+      "loss": 3.9796,
+      "step": 4214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.301708111069676e-05,
+      "loss": 4.0307,
+      "step": 4215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.30067483826955e-05,
+      "loss": 4.0961,
+      "step": 4216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2996415997357297e-05,
+      "loss": 3.9195,
+      "step": 4217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2986083956458403e-05,
+      "loss": 3.9325,
+      "step": 4218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.297575226177503e-05,
+      "loss": 4.0064,
+      "step": 4219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.296542091508332e-05,
+      "loss": 4.031,
+      "step": 4220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2955089918159334e-05,
+      "loss": 3.9975,
+      "step": 4221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29447592727791e-05,
+      "loss": 3.9692,
+      "step": 4222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2934428980718573e-05,
+      "loss": 3.8742,
+      "step": 4223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.292409904375365e-05,
+      "loss": 4.1299,
+      "step": 4224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2913769463660166e-05,
+      "loss": 4.0155,
+      "step": 4225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29034402422139e-05,
+      "loss": 3.9965,
+      "step": 4226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2893111381190562e-05,
+      "loss": 3.9636,
+      "step": 4227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2882782882365793e-05,
+      "loss": 3.8974,
+      "step": 4228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2872454747515198e-05,
+      "loss": 3.8887,
+      "step": 4229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2862126978414293e-05,
+      "loss": 4.0583,
+      "step": 4230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2851799576838552e-05,
+      "loss": 3.9349,
+      "step": 4231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.284147254456337e-05,
+      "loss": 3.9435,
+      "step": 4232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2831145883364073e-05,
+      "loss": 3.8807,
+      "step": 4233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2820819595015948e-05,
+      "loss": 3.9483,
+      "step": 4234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.28104936812942e-05,
+      "loss": 3.9969,
+      "step": 4235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.280016814397396e-05,
+      "loss": 3.8642,
+      "step": 4236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.278984298483032e-05,
+      "loss": 4.0231,
+      "step": 4237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.27795182056383e-05,
+      "loss": 3.8148,
+      "step": 4238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2769193808172844e-05,
+      "loss": 4.0979,
+      "step": 4239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2758869794208827e-05,
+      "loss": 3.9576,
+      "step": 4240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.274854616552108e-05,
+      "loss": 4.1099,
+      "step": 4241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2738222923884347e-05,
+      "loss": 3.823,
+      "step": 4242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.272790007107331e-05,
+      "loss": 3.8523,
+      "step": 4243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2717577608862598e-05,
+      "loss": 3.959,
+      "step": 4244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2707255539026752e-05,
+      "loss": 4.0221,
+      "step": 4245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.269693386334025e-05,
+      "loss": 4.0397,
+      "step": 4246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2686612583577525e-05,
+      "loss": 3.9355,
+      "step": 4247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2676291701512914e-05,
+      "loss": 3.9333,
+      "step": 4248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.26659712189207e-05,
+      "loss": 3.985,
+      "step": 4249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2655651137575095e-05,
+      "loss": 3.9916,
+      "step": 4250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2645331459250232e-05,
+      "loss": 4.0578,
+      "step": 4251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2635012185720195e-05,
+      "loss": 4.0454,
+      "step": 4252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2624693318758977e-05,
+      "loss": 4.1275,
+      "step": 4253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.261437486014051e-05,
+      "loss": 4.0358,
+      "step": 4254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2604056811638658e-05,
+      "loss": 3.9059,
+      "step": 4255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2593739175027222e-05,
+      "loss": 3.9916,
+      "step": 4256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2583421952079926e-05,
+      "loss": 3.9208,
+      "step": 4257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25731051445704e-05,
+      "loss": 3.9596,
+      "step": 4258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.256278875427224e-05,
+      "loss": 4.0797,
+      "step": 4259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2552472782958948e-05,
+      "loss": 4.095,
+      "step": 4260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2542157232403958e-05,
+      "loss": 4.0161,
+      "step": 4261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2531842104380634e-05,
+      "loss": 3.9038,
+      "step": 4262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.252152740066227e-05,
+      "loss": 3.9916,
+      "step": 4263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2511213123022067e-05,
+      "loss": 3.9569,
+      "step": 4264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2500899273233185e-05,
+      "loss": 4.0091,
+      "step": 4265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2490585853068687e-05,
+      "loss": 3.997,
+      "step": 4266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2480272864301583e-05,
+      "loss": 3.9879,
+      "step": 4267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.246996030870478e-05,
+      "loss": 4.1086,
+      "step": 4268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.245964818805113e-05,
+      "loss": 4.0156,
+      "step": 4269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2449336504113404e-05,
+      "loss": 3.9219,
+      "step": 4270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.243902525866431e-05,
+      "loss": 3.916,
+      "step": 4271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.242871445347646e-05,
+      "loss": 3.9113,
+      "step": 4272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2418404090322403e-05,
+      "loss": 3.8908,
+      "step": 4273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2408094170974603e-05,
+      "loss": 4.069,
+      "step": 4274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2397784697205474e-05,
+      "loss": 3.9847,
+      "step": 4275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2387475670787316e-05,
+      "loss": 3.8375,
+      "step": 4276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2377167093492386e-05,
+      "loss": 3.9963,
+      "step": 4277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2366858967092837e-05,
+      "loss": 3.9813,
+      "step": 4278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2356551293360752e-05,
+      "loss": 4.0325,
+      "step": 4279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2346244074068152e-05,
+      "loss": 4.0597,
+      "step": 4280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.233593731098696e-05,
+      "loss": 3.7643,
+      "step": 4281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2325631005889024e-05,
+      "loss": 4.0081,
+      "step": 4282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2315325160546118e-05,
+      "loss": 3.9111,
+      "step": 4283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2305019776729944e-05,
+      "loss": 3.7956,
+      "step": 4284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2294714856212118e-05,
+      "loss": 3.983,
+      "step": 4285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2284410400764173e-05,
+      "loss": 4.0479,
+      "step": 4286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2274106412157552e-05,
+      "loss": 4.0612,
+      "step": 4287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2263802892163646e-05,
+      "loss": 4.0003,
+      "step": 4288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2253499842553746e-05,
+      "loss": 3.9506,
+      "step": 4289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.224319726509906e-05,
+      "loss": 3.9475,
+      "step": 4290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2232895161570723e-05,
+      "loss": 4.1155,
+      "step": 4291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.222259353373978e-05,
+      "loss": 3.9447,
+      "step": 4292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2212292383377214e-05,
+      "loss": 3.9035,
+      "step": 4293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2201991712253906e-05,
+      "loss": 3.8523,
+      "step": 4294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.219169152214066e-05,
+      "loss": 3.9764,
+      "step": 4295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.21813918148082e-05,
+      "loss": 4.0008,
+      "step": 4296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2171092592027156e-05,
+      "loss": 4.0798,
+      "step": 4297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.21607938555681e-05,
+      "loss": 4.1523,
+      "step": 4298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2150495607201495e-05,
+      "loss": 3.9587,
+      "step": 4299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.214019784869772e-05,
+      "loss": 3.9205,
+      "step": 4300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2129900581827097e-05,
+      "loss": 4.0542,
+      "step": 4301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2119603808359824e-05,
+      "loss": 3.9677,
+      "step": 4302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.210930753006606e-05,
+      "loss": 4.033,
+      "step": 4303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2099011748715844e-05,
+      "loss": 3.9696,
+      "step": 4304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2088716466079135e-05,
+      "loss": 4.002,
+      "step": 4305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.207842168392582e-05,
+      "loss": 3.9339,
+      "step": 4306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.206812740402569e-05,
+      "loss": 3.9368,
+      "step": 4307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.205783362814844e-05,
+      "loss": 3.8765,
+      "step": 4308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2047540358063705e-05,
+      "loss": 3.9208,
+      "step": 4309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.203724759554101e-05,
+      "loss": 3.8298,
+      "step": 4310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.202695534234979e-05,
+      "loss": 4.0108,
+      "step": 4311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.201666360025942e-05,
+      "loss": 3.9606,
+      "step": 4312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2006372371039164e-05,
+      "loss": 3.9293,
+      "step": 4313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1996081656458204e-05,
+      "loss": 4.0154,
+      "step": 4314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1985791458285625e-05,
+      "loss": 4.1073,
+      "step": 4315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1975501778290446e-05,
+      "loss": 4.0245,
+      "step": 4316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1965212618241576e-05,
+      "loss": 3.928,
+      "step": 4317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.195492397990783e-05,
+      "loss": 3.9455,
+      "step": 4318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.194463586505796e-05,
+      "loss": 3.8943,
+      "step": 4319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.19343482754606e-05,
+      "loss": 3.9751,
+      "step": 4320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1924061212884315e-05,
+      "loss": 4.0282,
+      "step": 4321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.191377467909757e-05,
+      "loss": 4.0471,
+      "step": 4322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.190348867586873e-05,
+      "loss": 3.925,
+      "step": 4323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.189320320496609e-05,
+      "loss": 4.1015,
+      "step": 4324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1882918268157834e-05,
+      "loss": 3.929,
+      "step": 4325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1872633867212062e-05,
+      "loss": 4.0573,
+      "step": 4326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1862350003896788e-05,
+      "loss": 3.8903,
+      "step": 4327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1852066679979924e-05,
+      "loss": 4.1051,
+      "step": 4328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.184178389722928e-05,
+      "loss": 3.8591,
+      "step": 4329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.18315016574126e-05,
+      "loss": 4.0031,
+      "step": 4330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.182121996229753e-05,
+      "loss": 3.8067,
+      "step": 4331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1810938813651594e-05,
+      "loss": 3.9373,
+      "step": 4332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1800658213242243e-05,
+      "loss": 3.9726,
+      "step": 4333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.179037816283684e-05,
+      "loss": 3.8918,
+      "step": 4334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.178009866420264e-05,
+      "loss": 3.9711,
+      "step": 4335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1769819719106803e-05,
+      "loss": 3.9508,
+      "step": 4336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1759541329316408e-05,
+      "loss": 4.0641,
+      "step": 4337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1749263496598426e-05,
+      "loss": 4.0721,
+      "step": 4338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1738986222719722e-05,
+      "loss": 3.9714,
+      "step": 4339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.17287095094471e-05,
+      "loss": 3.9902,
+      "step": 4340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.171843335854724e-05,
+      "loss": 3.8769,
+      "step": 4341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1708157771786734e-05,
+      "loss": 4.0382,
+      "step": 4342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1697882750932064e-05,
+      "loss": 4.0435,
+      "step": 4343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1687608297749625e-05,
+      "loss": 3.9049,
+      "step": 4344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.167733441400573e-05,
+      "loss": 3.8109,
+      "step": 4345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1667061101466566e-05,
+      "loss": 4.081,
+      "step": 4346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.165678836189823e-05,
+      "loss": 3.9597,
+      "step": 4347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.164651619706673e-05,
+      "loss": 3.9992,
+      "step": 4348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1636244608737983e-05,
+      "loss": 3.8332,
+      "step": 4349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1625973598677783e-05,
+      "loss": 3.8699,
+      "step": 4350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.161570316865183e-05,
+      "loss": 4.0917,
+      "step": 4351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1605433320425745e-05,
+      "loss": 3.958,
+      "step": 4352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1595164055765025e-05,
+      "loss": 3.94,
+      "step": 4353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.158489537643507e-05,
+      "loss": 3.9419,
+      "step": 4354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1574627284201193e-05,
+      "loss": 3.8364,
+      "step": 4355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.15643597808286e-05,
+      "loss": 3.9407,
+      "step": 4356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.155409286808238e-05,
+      "loss": 3.9651,
+      "step": 4357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1543826547727546e-05,
+      "loss": 3.9843,
+      "step": 4358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1533560821529003e-05,
+      "loss": 3.9111,
+      "step": 4359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1523295691251544e-05,
+      "loss": 3.9638,
+      "step": 4360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1513031158659852e-05,
+      "loss": 4.0508,
+      "step": 4361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1502767225518534e-05,
+      "loss": 4.0017,
+      "step": 4362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1492503893592074e-05,
+      "loss": 3.9525,
+      "step": 4363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1482241164644856e-05,
+      "loss": 3.9126,
+      "step": 4364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.147197904044116e-05,
+      "loss": 3.8378,
+      "step": 4365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.146171752274516e-05,
+      "loss": 3.9897,
+      "step": 4366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.145145661332095e-05,
+      "loss": 4.0834,
+      "step": 4367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1441196313932484e-05,
+      "loss": 4.0486,
+      "step": 4368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1430936626343626e-05,
+      "loss": 3.8182,
+      "step": 4369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.142067755231814e-05,
+      "loss": 4.0616,
+      "step": 4370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.141041909361968e-05,
+      "loss": 3.7657,
+      "step": 4371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.140016125201179e-05,
+      "loss": 3.9785,
+      "step": 4372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.138990402925791e-05,
+      "loss": 3.9233,
+      "step": 4373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1379647427121388e-05,
+      "loss": 4.0095,
+      "step": 4374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1369391447365436e-05,
+      "loss": 4.0204,
+      "step": 4375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1359136091753176e-05,
+      "loss": 3.9861,
+      "step": 4376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1348881362047645e-05,
+      "loss": 3.8848,
+      "step": 4377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1338627260011732e-05,
+      "loss": 3.9466,
+      "step": 4378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1328373787408235e-05,
+      "loss": 4.0997,
+      "step": 4379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1318120945999853e-05,
+      "loss": 4.0423,
+      "step": 4380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1307868737549167e-05,
+      "loss": 3.8207,
+      "step": 4381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.129761716381864e-05,
+      "loss": 3.9135,
+      "step": 4382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.128736622657065e-05,
+      "loss": 3.9201,
+      "step": 4383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.127711592756745e-05,
+      "loss": 3.8807,
+      "step": 4384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1266866268571168e-05,
+      "loss": 4.022,
+      "step": 4385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1256617251343863e-05,
+      "loss": 3.9575,
+      "step": 4386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1246368877647445e-05,
+      "loss": 3.8826,
+      "step": 4387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1236121149243735e-05,
+      "loss": 3.9474,
+      "step": 4388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1225874067894436e-05,
+      "loss": 3.8632,
+      "step": 4389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1215627635361136e-05,
+      "loss": 4.0085,
+      "step": 4390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.120538185340532e-05,
+      "loss": 4.0757,
+      "step": 4391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1195136723788354e-05,
+      "loss": 3.9093,
+      "step": 4392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.118489224827149e-05,
+      "loss": 3.9532,
+      "step": 4393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.117464842861587e-05,
+      "loss": 3.9606,
+      "step": 4394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1164405266582546e-05,
+      "loss": 4.0282,
+      "step": 4395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1154162763932423e-05,
+      "loss": 3.8857,
+      "step": 4396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1143920922426297e-05,
+      "loss": 3.9486,
+      "step": 4397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1133679743824876e-05,
+      "loss": 4.1681,
+      "step": 4398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.112343922988873e-05,
+      "loss": 4.0139,
+      "step": 4399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1113199382378313e-05,
+      "loss": 4.0264,
+      "step": 4400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.110296020305399e-05,
+      "loss": 3.9163,
+      "step": 4401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1092721693675985e-05,
+      "loss": 3.9725,
+      "step": 4402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1082483856004405e-05,
+      "loss": 4.049,
+      "step": 4403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.107224669179928e-05,
+      "loss": 3.9195,
+      "step": 4404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1062010202820477e-05,
+      "loss": 3.8247,
+      "step": 4405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1051774390827778e-05,
+      "loss": 3.8949,
+      "step": 4406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1041539257580834e-05,
+      "loss": 4.0068,
+      "step": 4407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1031304804839177e-05,
+      "loss": 4.0104,
+      "step": 4408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1021071034362238e-05,
+      "loss": 3.963,
+      "step": 4409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1010837947909316e-05,
+      "loss": 3.8567,
+      "step": 4410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1000605547239592e-05,
+      "loss": 4.0717,
+      "step": 4411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.099037383411214e-05,
+      "loss": 3.9025,
+      "step": 4412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0980142810285904e-05,
+      "loss": 3.9989,
+      "step": 4413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0969912477519734e-05,
+      "loss": 3.9775,
+      "step": 4414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0959682837572317e-05,
+      "loss": 3.9978,
+      "step": 4415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.094945389220227e-05,
+      "loss": 3.9055,
+      "step": 4416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0939225643168057e-05,
+      "loss": 3.9393,
+      "step": 4417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0928998092228024e-05,
+      "loss": 3.8425,
+      "step": 4418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.091877124114042e-05,
+      "loss": 3.8805,
+      "step": 4419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0908545091663356e-05,
+      "loss": 3.9452,
+      "step": 4420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.089831964555482e-05,
+      "loss": 3.7829,
+      "step": 4421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.088809490457268e-05,
+      "loss": 3.9057,
+      "step": 4422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.08778708704747e-05,
+      "loss": 3.8705,
+      "step": 4423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.086764754501851e-05,
+      "loss": 3.9714,
+      "step": 4424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0857424929961613e-05,
+      "loss": 4.0127,
+      "step": 4425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.084720302706139e-05,
+      "loss": 3.9721,
+      "step": 4426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0836981838075115e-05,
+      "loss": 4.0776,
+      "step": 4427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0826761364759924e-05,
+      "loss": 3.934,
+      "step": 4428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.081654160887283e-05,
+      "loss": 3.995,
+      "step": 4429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.080632257217074e-05,
+      "loss": 3.8921,
+      "step": 4430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0796104256410402e-05,
+      "loss": 3.9978,
+      "step": 4431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.078588666334849e-05,
+      "loss": 4.0142,
+      "step": 4432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.077566979474151e-05,
+      "loss": 3.9728,
+      "step": 4433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.076545365234587e-05,
+      "loss": 3.873,
+      "step": 4434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.075523823791784e-05,
+      "loss": 4.0462,
+      "step": 4435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0745023553213557e-05,
+      "loss": 3.8033,
+      "step": 4436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0734809599989064e-05,
+      "loss": 3.8861,
+      "step": 4437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.072459638000024e-05,
+      "loss": 3.9866,
+      "step": 4438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0714383895002865e-05,
+      "loss": 4.0898,
+      "step": 4439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0704172146752576e-05,
+      "loss": 4.037,
+      "step": 4440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.06939611370049e-05,
+      "loss": 3.909,
+      "step": 4441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0683750867515227e-05,
+      "loss": 4.0128,
+      "step": 4442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0673541340038822e-05,
+      "loss": 4.1494,
+      "step": 4443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.066333255633081e-05,
+      "loss": 3.9662,
+      "step": 4444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.065312451814621e-05,
+      "loss": 3.9617,
+      "step": 4445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.06429172272399e-05,
+      "loss": 3.8637,
+      "step": 4446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0632710685366623e-05,
+      "loss": 3.9471,
+      "step": 4447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.062250489428102e-05,
+      "loss": 4.0161,
+      "step": 4448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.061229985573757e-05,
+      "loss": 3.9989,
+      "step": 4449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0602095571490633e-05,
+      "loss": 3.8921,
+      "step": 4450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0591892043294453e-05,
+      "loss": 3.9934,
+      "step": 4451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0581689272903143e-05,
+      "loss": 4.013,
+      "step": 4452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0571487262070665e-05,
+      "loss": 3.8668,
+      "step": 4453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0561286012550864e-05,
+      "loss": 3.9896,
+      "step": 4454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0551085526097462e-05,
+      "loss": 3.8513,
+      "step": 4455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0540885804464033e-05,
+      "loss": 4.027,
+      "step": 4456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.053068684940402e-05,
+      "loss": 3.8406,
+      "step": 4457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0520488662670763e-05,
+      "loss": 3.9177,
+      "step": 4458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0510291246017417e-05,
+      "loss": 4.118,
+      "step": 4459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.050009460119707e-05,
+      "loss": 4.0093,
+      "step": 4460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.048989872996263e-05,
+      "loss": 4.0181,
+      "step": 4461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0479703634066873e-05,
+      "loss": 3.8442,
+      "step": 4462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0469509315262474e-05,
+      "loss": 3.9161,
+      "step": 4463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0459315775301945e-05,
+      "loss": 4.0142,
+      "step": 4464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.044912301593767e-05,
+      "loss": 3.968,
+      "step": 4465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0438931038921913e-05,
+      "loss": 3.9033,
+      "step": 4466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0428739846006784e-05,
+      "loss": 4.0225,
+      "step": 4467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0418549438944262e-05,
+      "loss": 3.9958,
+      "step": 4468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0408359819486206e-05,
+      "loss": 3.9488,
+      "step": 4469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0398170989384334e-05,
+      "loss": 3.7981,
+      "step": 4470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.038798295039022e-05,
+      "loss": 3.808,
+      "step": 4471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.03777957042553e-05,
+      "loss": 3.7752,
+      "step": 4472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0367609252730887e-05,
+      "loss": 4.1694,
+      "step": 4473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0357423597568147e-05,
+      "loss": 3.8985,
+      "step": 4474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0347238740518108e-05,
+      "loss": 3.9175,
+      "step": 4475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0337054683331673e-05,
+      "loss": 3.8951,
+      "step": 4476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0326871427759584e-05,
+      "loss": 3.8908,
+      "step": 4477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0316688975552482e-05,
+      "loss": 3.9177,
+      "step": 4478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.030650732846084e-05,
+      "loss": 3.8559,
+      "step": 4479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0296326488234997e-05,
+      "loss": 3.8634,
+      "step": 4480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.028614645662516e-05,
+      "loss": 3.9557,
+      "step": 4481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.02759672353814e-05,
+      "loss": 3.9993,
+      "step": 4482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.026578882625363e-05,
+      "loss": 4.0066,
+      "step": 4483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.025561123099165e-05,
+      "loss": 3.9122,
+      "step": 4484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.02454344513451e-05,
+      "loss": 3.9585,
+      "step": 4485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0235258489063486e-05,
+      "loss": 3.9934,
+      "step": 4486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0225083345896164e-05,
+      "loss": 3.9515,
+      "step": 4487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0214909023592388e-05,
+      "loss": 3.9298,
+      "step": 4488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.020473552390122e-05,
+      "loss": 3.976,
+      "step": 4489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0194562848571604e-05,
+      "loss": 3.9798,
+      "step": 4490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0184390999352347e-05,
+      "loss": 3.9296,
+      "step": 4491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0174219977992104e-05,
+      "loss": 3.9957,
+      "step": 4492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0164049786239386e-05,
+      "loss": 3.898,
+      "step": 4493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.015388042584258e-05,
+      "loss": 3.8254,
+      "step": 4494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.014371189854991e-05,
+      "loss": 3.9717,
+      "step": 4495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0133544206109452e-05,
+      "loss": 3.8348,
+      "step": 4496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0123377350269176e-05,
+      "loss": 3.9874,
+      "step": 4497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.011321133277686e-05,
+      "loss": 4.0893,
+      "step": 4498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0103046155380174e-05,
+      "loss": 4.0134,
+      "step": 4499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0092881819826622e-05,
+      "loss": 3.7738,
+      "step": 4500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.008271832786357e-05,
+      "loss": 3.9622,
+      "step": 4501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.007255568123825e-05,
+      "loss": 3.9741,
+      "step": 4502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0062393881697734e-05,
+      "loss": 4.001,
+      "step": 4503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.005223293098894e-05,
+      "loss": 4.0855,
+      "step": 4504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0042072830858663e-05,
+      "loss": 4.0635,
+      "step": 4505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0031913583053553e-05,
+      "loss": 4.0466,
+      "step": 4506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0021755189320096e-05,
+      "loss": 4.0468,
+      "step": 4507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0011597651404625e-05,
+      "loss": 3.9249,
+      "step": 4508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0001440971053357e-05,
+      "loss": 3.887,
+      "step": 4509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9991285150012332e-05,
+      "loss": 3.956,
+      "step": 4510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9981130190027454e-05,
+      "loss": 3.9949,
+      "step": 4511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.997097609284448e-05,
+      "loss": 3.8742,
+      "step": 4512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9960822860209023e-05,
+      "loss": 4.0408,
+      "step": 4513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.995067049386652e-05,
+      "loss": 4.0993,
+      "step": 4514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9940518995562308e-05,
+      "loss": 4.0266,
+      "step": 4515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9930368367041528e-05,
+      "loss": 4.0296,
+      "step": 4516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9920218610049207e-05,
+      "loss": 3.9152,
+      "step": 4517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9910069726330196e-05,
+      "loss": 4.0394,
+      "step": 4518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.98999217176292e-05,
+      "loss": 3.9036,
+      "step": 4519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9889774585690794e-05,
+      "loss": 3.9307,
+      "step": 4520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9879628332259375e-05,
+      "loss": 3.8578,
+      "step": 4521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9869482959079205e-05,
+      "loss": 3.9443,
+      "step": 4522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9859338467894395e-05,
+      "loss": 3.9264,
+      "step": 4523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.984919486044889e-05,
+      "loss": 3.9566,
+      "step": 4524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.983905213848651e-05,
+      "loss": 4.0179,
+      "step": 4525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.982891030375089e-05,
+      "loss": 3.8757,
+      "step": 4526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.981876935798555e-05,
+      "loss": 3.8098,
+      "step": 4527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808629302933817e-05,
+      "loss": 4.0537,
+      "step": 4528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9798490140338888e-05,
+      "loss": 3.9578,
+      "step": 4529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.978835187194381e-05,
+      "loss": 4.0504,
+      "step": 4530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.977821449949146e-05,
+      "loss": 3.8667,
+      "step": 4531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9768078024724578e-05,
+      "loss": 3.87,
+      "step": 4532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9757942449385725e-05,
+      "loss": 4.1375,
+      "step": 4533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9747807775217344e-05,
+      "loss": 3.9788,
+      "step": 4534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.97376740039617e-05,
+      "loss": 4.0989,
+      "step": 4535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.97275411373609e-05,
+      "loss": 3.8735,
+      "step": 4536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9717409177156897e-05,
+      "loss": 3.8719,
+      "step": 4537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.97072781250915e-05,
+      "loss": 3.8872,
+      "step": 4538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9697147982906356e-05,
+      "loss": 3.9789,
+      "step": 4539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.968701875234294e-05,
+      "loss": 3.7668,
+      "step": 4540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9676890435142605e-05,
+      "loss": 3.909,
+      "step": 4541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9666763033046497e-05,
+      "loss": 3.9915,
+      "step": 4542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9656636547795664e-05,
+      "loss": 3.8796,
+      "step": 4543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9646510981130954e-05,
+      "loss": 3.8841,
+      "step": 4544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.963638633479307e-05,
+      "loss": 3.9517,
+      "step": 4545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962626261052256e-05,
+      "loss": 3.8061,
+      "step": 4546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616139810059793e-05,
+      "loss": 3.9703,
+      "step": 4547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960601793514502e-05,
+      "loss": 3.8546,
+      "step": 4548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.959589698751829e-05,
+      "loss": 4.0902,
+      "step": 4549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9585776968919517e-05,
+      "loss": 3.8901,
+      "step": 4550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.957565788108844e-05,
+      "loss": 3.862,
+      "step": 4551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9565539725764674e-05,
+      "loss": 3.926,
+      "step": 4552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9555422504687627e-05,
+      "loss": 3.8508,
+      "step": 4553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9545306219596565e-05,
+      "loss": 3.8877,
+      "step": 4554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9535190872230607e-05,
+      "loss": 3.8786,
+      "step": 4555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.952507646432869e-05,
+      "loss": 4.0246,
+      "step": 4556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9514962997629603e-05,
+      "loss": 3.8856,
+      "step": 4557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9504850473871956e-05,
+      "loss": 3.9017,
+      "step": 4558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9494738894794224e-05,
+      "loss": 3.9734,
+      "step": 4559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9484628262134698e-05,
+      "loss": 3.9626,
+      "step": 4560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9474518577631503e-05,
+      "loss": 3.7846,
+      "step": 4561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9464409843022627e-05,
+      "loss": 3.8703,
+      "step": 4562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9454302060045878e-05,
+      "loss": 3.867,
+      "step": 4563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9444195230438897e-05,
+      "loss": 3.9679,
+      "step": 4564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.943408935593916e-05,
+      "loss": 3.9361,
+      "step": 4565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.942398443828399e-05,
+      "loss": 3.8744,
+      "step": 4566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.941388047921054e-05,
+      "loss": 3.9333,
+      "step": 4567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.940377748045579e-05,
+      "loss": 3.9299,
+      "step": 4568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9393675443756572e-05,
+      "loss": 4.0104,
+      "step": 4569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9383574370849534e-05,
+      "loss": 3.9781,
+      "step": 4570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.937347426347118e-05,
+      "loss": 3.9062,
+      "step": 4571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.936337512335782e-05,
+      "loss": 3.9008,
+      "step": 4572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.935327695224563e-05,
+      "loss": 3.9913,
+      "step": 4573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9343179751870595e-05,
+      "loss": 3.8203,
+      "step": 4574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9333083523968533e-05,
+      "loss": 3.9042,
+      "step": 4575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9322988270275117e-05,
+      "loss": 3.8319,
+      "step": 4576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.931289399252583e-05,
+      "loss": 4.0111,
+      "step": 4577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9302800692455995e-05,
+      "loss": 3.9161,
+      "step": 4578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.929270837180076e-05,
+      "loss": 3.9546,
+      "step": 4579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9282617032295124e-05,
+      "loss": 3.8839,
+      "step": 4580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9272526675673906e-05,
+      "loss": 3.815,
+      "step": 4581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9262437303671755e-05,
+      "loss": 3.9892,
+      "step": 4582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9252348918023138e-05,
+      "loss": 3.8864,
+      "step": 4583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9242261520462383e-05,
+      "loss": 3.8692,
+      "step": 4584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.923217511272362e-05,
+      "loss": 3.922,
+      "step": 4585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9222089696540817e-05,
+      "loss": 3.9166,
+      "step": 4586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9212005273647782e-05,
+      "loss": 3.9846,
+      "step": 4587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9201921845778133e-05,
+      "loss": 3.7772,
+      "step": 4588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.919183941466534e-05,
+      "loss": 3.9086,
+      "step": 4589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9181757982042686e-05,
+      "loss": 3.8719,
+      "step": 4590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.917167754964329e-05,
+      "loss": 3.8975,
+      "step": 4591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.916159811920009e-05,
+      "loss": 3.9227,
+      "step": 4592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9151519692445856e-05,
+      "loss": 3.9547,
+      "step": 4593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.914144227111319e-05,
+      "loss": 3.8209,
+      "step": 4594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9131365856934517e-05,
+      "loss": 3.896,
+      "step": 4595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.912129045164209e-05,
+      "loss": 4.0087,
+      "step": 4596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9111216056967978e-05,
+      "loss": 3.8478,
+      "step": 4597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.910114267464409e-05,
+      "loss": 3.9422,
+      "step": 4598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9091070306402177e-05,
+      "loss": 3.9157,
+      "step": 4599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.908099895397377e-05,
+      "loss": 3.948,
+      "step": 4600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9070928619090266e-05,
+      "loss": 3.9754,
+      "step": 4601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.906085930348287e-05,
+      "loss": 4.0505,
+      "step": 4602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9050791008882614e-05,
+      "loss": 3.7259,
+      "step": 4603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9040723737020344e-05,
+      "loss": 3.8343,
+      "step": 4604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9030657489626754e-05,
+      "loss": 3.8886,
+      "step": 4605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9020592268432348e-05,
+      "loss": 3.9061,
+      "step": 4606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.901052807516744e-05,
+      "loss": 4.1424,
+      "step": 4607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9000464911562194e-05,
+      "loss": 3.9369,
+      "step": 4608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8990402779346588e-05,
+      "loss": 3.797,
+      "step": 4609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8980341680250412e-05,
+      "loss": 3.9148,
+      "step": 4610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8970281616003286e-05,
+      "loss": 3.9687,
+      "step": 4611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8960222588334657e-05,
+      "loss": 3.8824,
+      "step": 4612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8950164598973784e-05,
+      "loss": 3.9047,
+      "step": 4613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8940107649649745e-05,
+      "loss": 3.9631,
+      "step": 4614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.893005174209146e-05,
+      "loss": 3.9779,
+      "step": 4615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8919996878027635e-05,
+      "loss": 3.8841,
+      "step": 4616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8909943059186845e-05,
+      "loss": 4.123,
+      "step": 4617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.889989028729745e-05,
+      "loss": 4.1131,
+      "step": 4618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8889838564087623e-05,
+      "loss": 3.9837,
+      "step": 4619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.887978789128539e-05,
+      "loss": 3.9166,
+      "step": 4620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8869738270618567e-05,
+      "loss": 3.871,
+      "step": 4621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.88596897038148e-05,
+      "loss": 3.9639,
+      "step": 4622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.884964219260156e-05,
+      "loss": 4.0115,
+      "step": 4623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8839595738706132e-05,
+      "loss": 4.0869,
+      "step": 4624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.88295503438556e-05,
+      "loss": 3.8815,
+      "step": 4625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8819506009776906e-05,
+      "loss": 3.8152,
+      "step": 4626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8809462738196785e-05,
+      "loss": 3.9796,
+      "step": 4627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8799420530841782e-05,
+      "loss": 4.0295,
+      "step": 4628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.878937938943827e-05,
+      "loss": 3.8983,
+      "step": 4629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.877933931571245e-05,
+      "loss": 3.9543,
+      "step": 4630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8769300311390314e-05,
+      "loss": 3.959,
+      "step": 4631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.875926237819768e-05,
+      "loss": 3.9658,
+      "step": 4632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.87492255178602e-05,
+      "loss": 4.0421,
+      "step": 4633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8739189732103317e-05,
+      "loss": 3.8923,
+      "step": 4634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.872915502265229e-05,
+      "loss": 4.0645,
+      "step": 4635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8719121391232225e-05,
+      "loss": 3.9024,
+      "step": 4636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8709088839568003e-05,
+      "loss": 3.8248,
+      "step": 4637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8699057369384344e-05,
+      "loss": 3.8516,
+      "step": 4638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8689026982405768e-05,
+      "loss": 4.0293,
+      "step": 4639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.867899768035661e-05,
+      "loss": 3.8921,
+      "step": 4640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.866896946496104e-05,
+      "loss": 3.746,
+      "step": 4641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8658942337943012e-05,
+      "loss": 3.8812,
+      "step": 4642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8648916301026305e-05,
+      "loss": 3.8712,
+      "step": 4643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8638891355934504e-05,
+      "loss": 3.8615,
+      "step": 4644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8628867504391037e-05,
+      "loss": 3.8757,
+      "step": 4645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.861884474811911e-05,
+      "loss": 4.0001,
+      "step": 4646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8608823088841735e-05,
+      "loss": 3.9694,
+      "step": 4647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598802528281772e-05,
+      "loss": 4.0424,
+      "step": 4648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8588783068161864e-05,
+      "loss": 3.8843,
+      "step": 4649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8578764710204465e-05,
+      "loss": 4.0264,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.856874745613186e-05,
+      "loss": 3.9476,
+      "step": 4651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8558731307666128e-05,
+      "loss": 3.9448,
+      "step": 4652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548716266529143e-05,
+      "loss": 3.8113,
+      "step": 4653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538702334442633e-05,
+      "loss": 3.8664,
+      "step": 4654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852868951312809e-05,
+      "loss": 3.936,
+      "step": 4655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851867780430685e-05,
+      "loss": 3.9653,
+      "step": 4656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508667209700036e-05,
+      "loss": 3.9573,
+      "step": 4657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498657731028575e-05,
+      "loss": 3.8851,
+      "step": 4658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8488649370013223e-05,
+      "loss": 3.9834,
+      "step": 4659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.847864212837453e-05,
+      "loss": 3.7683,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8468636007832856e-05,
+      "loss": 3.7837,
+      "step": 4661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8458631010108364e-05,
+      "loss": 3.8846,
+      "step": 4662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.844862713692104e-05,
+      "loss": 3.933,
+      "step": 4663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8438624389990668e-05,
+      "loss": 3.963,
+      "step": 4664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8428622771036825e-05,
+      "loss": 3.9151,
+      "step": 4665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8418622281778907e-05,
+      "loss": 3.8744,
+      "step": 4666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8408622923936124e-05,
+      "loss": 3.8621,
+      "step": 4667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.839862469922747e-05,
+      "loss": 3.9401,
+      "step": 4668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8388627609371758e-05,
+      "loss": 4.0522,
+      "step": 4669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.837863165608761e-05,
+      "loss": 3.8899,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8368636841093434e-05,
+      "loss": 3.9644,
+      "step": 4671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8358643166107466e-05,
+      "loss": 3.9898,
+      "step": 4672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8348650632847726e-05,
+      "loss": 3.8235,
+      "step": 4673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8338659243032064e-05,
+      "loss": 3.9206,
+      "step": 4674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8328668998378096e-05,
+      "loss": 3.9446,
+      "step": 4675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8318679900603265e-05,
+      "loss": 4.1351,
+      "step": 4676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8308691951424823e-05,
+      "loss": 3.8922,
+      "step": 4677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8298705152559804e-05,
+      "loss": 3.9134,
+      "step": 4678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8288719505725055e-05,
+      "loss": 3.9663,
+      "step": 4679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8278735012637228e-05,
+      "loss": 3.8708,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8268751675012763e-05,
+      "loss": 4.0239,
+      "step": 4681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8258769494567932e-05,
+      "loss": 3.976,
+      "step": 4682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8248788473018767e-05,
+      "loss": 4.0011,
+      "step": 4683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8238808612081143e-05,
+      "loss": 4.1366,
+      "step": 4684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8228829913470695e-05,
+      "loss": 3.9416,
+      "step": 4685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8218852378902877e-05,
+      "loss": 3.9974,
+      "step": 4686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.820887601009296e-05,
+      "loss": 3.8761,
+      "step": 4687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8198900808755982e-05,
+      "loss": 3.937,
+      "step": 4688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8188926776606795e-05,
+      "loss": 3.9522,
+      "step": 4689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8178953915360057e-05,
+      "loss": 3.9556,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8168982226730217e-05,
+      "loss": 3.8226,
+      "step": 4691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8159011712431528e-05,
+      "loss": 3.9493,
+      "step": 4692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8149042374178038e-05,
+      "loss": 3.8221,
+      "step": 4693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8139074213683576e-05,
+      "loss": 3.8827,
+      "step": 4694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8129107232661802e-05,
+      "loss": 3.8965,
+      "step": 4695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.811914143282615e-05,
+      "loss": 4.0758,
+      "step": 4696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.810917681588985e-05,
+      "loss": 4.0738,
+      "step": 4697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8099213383565948e-05,
+      "loss": 3.9458,
+      "step": 4698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8089251137567254e-05,
+      "loss": 3.898,
+      "step": 4699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8079290079606414e-05,
+      "loss": 3.9026,
+      "step": 4700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8069330211395834e-05,
+      "loss": 3.8492,
+      "step": 4701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.805937153464775e-05,
+      "loss": 4.019,
+      "step": 4702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.804941405107416e-05,
+      "loss": 3.9935,
+      "step": 4703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8039457762386863e-05,
+      "loss": 3.952,
+      "step": 4704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.802950267029748e-05,
+      "loss": 4.0004,
+      "step": 4705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8019548776517394e-05,
+      "loss": 3.8447,
+      "step": 4706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8009596082757796e-05,
+      "loss": 3.8294,
+      "step": 4707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.799964459072967e-05,
+      "loss": 3.977,
+      "step": 4708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.798969430214379e-05,
+      "loss": 3.871,
+      "step": 4709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7979745218710737e-05,
+      "loss": 3.8567,
+      "step": 4710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7969797342140867e-05,
+      "loss": 3.8375,
+      "step": 4711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.795985067414433e-05,
+      "loss": 3.9627,
+      "step": 4712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7949905216431083e-05,
+      "loss": 3.8974,
+      "step": 4713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7939960970710865e-05,
+      "loss": 3.9597,
+      "step": 4714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7930017938693194e-05,
+      "loss": 4.0264,
+      "step": 4715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7920076122087408e-05,
+      "loss": 3.9527,
+      "step": 4716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7910135522602614e-05,
+      "loss": 4.0951,
+      "step": 4717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7900196141947707e-05,
+      "loss": 3.9099,
+      "step": 4718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7890257981831393e-05,
+      "loss": 3.7998,
+      "step": 4719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7880321043962165e-05,
+      "loss": 4.0295,
+      "step": 4720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7870385330048285e-05,
+      "loss": 3.9278,
+      "step": 4721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7860450841797815e-05,
+      "loss": 3.848,
+      "step": 4722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.785051758091862e-05,
+      "loss": 3.9205,
+      "step": 4723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.784058554911834e-05,
+      "loss": 3.9675,
+      "step": 4724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.783065474810439e-05,
+      "loss": 3.8421,
+      "step": 4725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7820725179584014e-05,
+      "loss": 4.0603,
+      "step": 4726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7810796845264198e-05,
+      "loss": 4.0378,
+      "step": 4727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7800869746851758e-05,
+      "loss": 3.9366,
+      "step": 4728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.779094388605327e-05,
+      "loss": 4.014,
+      "step": 4729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7781019264575095e-05,
+      "loss": 3.9384,
+      "step": 4730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7771095884123405e-05,
+      "loss": 3.9612,
+      "step": 4731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7761173746404137e-05,
+      "loss": 3.7616,
+      "step": 4732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7751252853123012e-05,
+      "loss": 3.9326,
+      "step": 4733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7741333205985568e-05,
+      "loss": 3.8256,
+      "step": 4734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7731414806697092e-05,
+      "loss": 3.8347,
+      "step": 4735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7721497656962665e-05,
+      "loss": 3.9804,
+      "step": 4736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.771158175848718e-05,
+      "loss": 3.9694,
+      "step": 4737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7701667112975283e-05,
+      "loss": 3.8669,
+      "step": 4738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7691753722131424e-05,
+      "loss": 3.9251,
+      "step": 4739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7681841587659817e-05,
+      "loss": 3.9077,
+      "step": 4740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7671930711264487e-05,
+      "loss": 3.8583,
+      "step": 4741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7662021094649222e-05,
+      "loss": 3.9479,
+      "step": 4742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7652112739517594e-05,
+      "loss": 3.8793,
+      "step": 4743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7642205647572977e-05,
+      "loss": 3.9224,
+      "step": 4744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7632299820518506e-05,
+      "loss": 3.9372,
+      "step": 4745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.76223952600571e-05,
+      "loss": 3.8646,
+      "step": 4746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7612491967891487e-05,
+      "loss": 4.1287,
+      "step": 4747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7602589945724147e-05,
+      "loss": 3.9553,
+      "step": 4748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7592689195257352e-05,
+      "loss": 3.9124,
+      "step": 4749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.758278971819316e-05,
+      "loss": 4.0211,
+      "step": 4750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.75728915162334e-05,
+      "loss": 3.9036,
+      "step": 4751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7562994591079697e-05,
+      "loss": 4.0369,
+      "step": 4752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7553098944433436e-05,
+      "loss": 3.8986,
+      "step": 4753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7543204577995794e-05,
+      "loss": 4.0598,
+      "step": 4754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.753331149346773e-05,
+      "loss": 3.9661,
+      "step": 4755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.752341969254999e-05,
+      "loss": 3.8726,
+      "step": 4756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.751352917694308e-05,
+      "loss": 3.9376,
+      "step": 4757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.750363994834729e-05,
+      "loss": 3.9893,
+      "step": 4758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7493752008462703e-05,
+      "loss": 3.9483,
+      "step": 4759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.748386535898917e-05,
+      "loss": 4.027,
+      "step": 4760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7473980001626304e-05,
+      "loss": 3.9801,
+      "step": 4761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7464095938073537e-05,
+      "loss": 3.916,
+      "step": 4762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7454213170030038e-05,
+      "loss": 3.8238,
+      "step": 4763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7444331699194764e-05,
+      "loss": 3.9602,
+      "step": 4764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7434451527266466e-05,
+      "loss": 4.0012,
+      "step": 4765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7424572655943665e-05,
+      "loss": 3.8945,
+      "step": 4766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.741469508692465e-05,
+      "loss": 3.9755,
+      "step": 4767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7404818821907482e-05,
+      "loss": 3.7617,
+      "step": 4768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7394943862590004e-05,
+      "loss": 4.055,
+      "step": 4769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.738507021066985e-05,
+      "loss": 3.9066,
+      "step": 4770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7375197867844402e-05,
+      "loss": 3.9789,
+      "step": 4771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7365326835810834e-05,
+      "loss": 3.9052,
+      "step": 4772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7355457116266083e-05,
+      "loss": 3.9818,
+      "step": 4773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.734558871090689e-05,
+      "loss": 3.8464,
+      "step": 4774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.733572162142973e-05,
+      "loss": 4.016,
+      "step": 4775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7325855849530874e-05,
+      "loss": 3.9327,
+      "step": 4776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.731599139690637e-05,
+      "loss": 4.0434,
+      "step": 4777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7306128265252027e-05,
+      "loss": 3.8978,
+      "step": 4778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.729626645626342e-05,
+      "loss": 3.8005,
+      "step": 4779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7286405971635927e-05,
+      "loss": 3.9281,
+      "step": 4780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7276546813064672e-05,
+      "loss": 3.8569,
+      "step": 4781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.726668898224455e-05,
+      "loss": 3.9725,
+      "step": 4782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7256832480870238e-05,
+      "loss": 3.887,
+      "step": 4783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7246977310636204e-05,
+      "loss": 3.8391,
+      "step": 4784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7237123473236645e-05,
+      "loss": 3.8961,
+      "step": 4785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7227270970365557e-05,
+      "loss": 4.0016,
+      "step": 4786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7217419803716696e-05,
+      "loss": 3.9683,
+      "step": 4787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.72075699749836e-05,
+      "loss": 4.0468,
+      "step": 4788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7197721485859563e-05,
+      "loss": 4.027,
+      "step": 4789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7187874338037646e-05,
+      "loss": 3.9416,
+      "step": 4790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7178028533210703e-05,
+      "loss": 3.9607,
+      "step": 4791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7168184073071326e-05,
+      "loss": 3.9385,
+      "step": 4792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7158340959311912e-05,
+      "loss": 3.7139,
+      "step": 4793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7148499193624587e-05,
+      "loss": 3.8442,
+      "step": 4794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.713865877770128e-05,
+      "loss": 3.8958,
+      "step": 4795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7128819713233664e-05,
+      "loss": 3.8861,
+      "step": 4796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7118982001913187e-05,
+      "loss": 3.9173,
+      "step": 4797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7109145645431074e-05,
+      "loss": 3.9365,
+      "step": 4798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.70993106454783e-05,
+      "loss": 4.0729,
+      "step": 4799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7089477003745617e-05,
+      "loss": 4.0074,
+      "step": 4800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7079644721923537e-05,
+      "loss": 3.9951,
+      "step": 4801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7069813801702362e-05,
+      "loss": 4.0904,
+      "step": 4802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7059984244772127e-05,
+      "loss": 3.9056,
+      "step": 4803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7050156052822643e-05,
+      "loss": 3.9198,
+      "step": 4804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.70403292275435e-05,
+      "loss": 3.9926,
+      "step": 4805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7030503770624047e-05,
+      "loss": 3.9377,
+      "step": 4806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.702067968375337e-05,
+      "loss": 3.8538,
+      "step": 4807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7010856968620373e-05,
+      "loss": 4.0444,
+      "step": 4808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7001035626913676e-05,
+      "loss": 3.9394,
+      "step": 4809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.699121566032168e-05,
+      "loss": 4.0472,
+      "step": 4810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6981397070532567e-05,
+      "loss": 3.9871,
+      "step": 4811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.697157985923425e-05,
+      "loss": 3.8301,
+      "step": 4812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.696176402811444e-05,
+      "loss": 3.9696,
+      "step": 4813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6951949578860575e-05,
+      "loss": 3.8665,
+      "step": 4814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6942136513159875e-05,
+      "loss": 3.7384,
+      "step": 4815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6932324832699326e-05,
+      "loss": 3.9886,
+      "step": 4816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.692251453916567e-05,
+      "loss": 3.9722,
+      "step": 4817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.69127056342454e-05,
+      "loss": 3.9402,
+      "step": 4818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6902898119624794e-05,
+      "loss": 3.9473,
+      "step": 4819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.689309199698986e-05,
+      "loss": 3.8116,
+      "step": 4820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6883287268026404e-05,
+      "loss": 3.92,
+      "step": 4821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6873483934419958e-05,
+      "loss": 3.9525,
+      "step": 4822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6863681997855844e-05,
+      "loss": 3.8485,
+      "step": 4823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6853881460019117e-05,
+      "loss": 3.9885,
+      "step": 4824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.68440823225946e-05,
+      "loss": 4.0273,
+      "step": 4825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6834284587266884e-05,
+      "loss": 3.9999,
+      "step": 4826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6824488255720316e-05,
+      "loss": 3.9443,
+      "step": 4827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6814693329638993e-05,
+      "loss": 3.7755,
+      "step": 4828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6804899810706777e-05,
+      "loss": 4.0002,
+      "step": 4829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6795107700607284e-05,
+      "loss": 3.9221,
+      "step": 4830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6785317001023907e-05,
+      "loss": 3.8996,
+      "step": 4831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.677552771363977e-05,
+      "loss": 3.9192,
+      "step": 4832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6765739840137758e-05,
+      "loss": 3.8592,
+      "step": 4833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6755953382200536e-05,
+      "loss": 3.896,
+      "step": 4834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.67461683415105e-05,
+      "loss": 3.9488,
+      "step": 4835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6736384719749805e-05,
+      "loss": 3.911,
+      "step": 4836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6726602518600383e-05,
+      "loss": 3.9233,
+      "step": 4837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6716821739743888e-05,
+      "loss": 3.9098,
+      "step": 4838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6707042384861775e-05,
+      "loss": 3.838,
+      "step": 4839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6697264455635213e-05,
+      "loss": 3.8447,
+      "step": 4840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.668748795374515e-05,
+      "loss": 3.7828,
+      "step": 4841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6677712880872277e-05,
+      "loss": 3.9541,
+      "step": 4842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.666793923869703e-05,
+      "loss": 3.7917,
+      "step": 4843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6658167028899628e-05,
+      "loss": 3.9477,
+      "step": 4844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.664839625316002e-05,
+      "loss": 3.8737,
+      "step": 4845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.663862691315791e-05,
+      "loss": 3.7816,
+      "step": 4846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6628859010572774e-05,
+      "loss": 3.951,
+      "step": 4847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6619092547083815e-05,
+      "loss": 3.9695,
+      "step": 4848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.660932752437001e-05,
+      "loss": 3.8578,
+      "step": 4849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.659956394411008e-05,
+      "loss": 3.9151,
+      "step": 4850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.658980180798249e-05,
+      "loss": 3.9293,
+      "step": 4851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.658004111766547e-05,
+      "loss": 3.9652,
+      "step": 4852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6570281874836994e-05,
+      "loss": 3.8779,
+      "step": 4853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.656052408117479e-05,
+      "loss": 3.9117,
+      "step": 4854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6550767738356337e-05,
+      "loss": 3.874,
+      "step": 4855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6541012848058863e-05,
+      "loss": 3.7886,
+      "step": 4856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6531259411959332e-05,
+      "loss": 3.8798,
+      "step": 4857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6521507431734492e-05,
+      "loss": 3.9518,
+      "step": 4858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.651175690906082e-05,
+      "loss": 3.9933,
+      "step": 4859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.650200784561454e-05,
+      "loss": 3.8914,
+      "step": 4860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6492260243071627e-05,
+      "loss": 3.9272,
+      "step": 4861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.648251410310781e-05,
+      "loss": 3.8845,
+      "step": 4862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6472769427398566e-05,
+      "loss": 4.06,
+      "step": 4863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6463026217619104e-05,
+      "loss": 3.9051,
+      "step": 4864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6453284475444415e-05,
+      "loss": 3.8747,
+      "step": 4865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.644354420254919e-05,
+      "loss": 4.0228,
+      "step": 4866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6433805400607924e-05,
+      "loss": 3.9654,
+      "step": 4867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6424068071294817e-05,
+      "loss": 3.9624,
+      "step": 4868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6414332216283826e-05,
+      "loss": 3.9063,
+      "step": 4869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6404597837248664e-05,
+      "loss": 4.0098,
+      "step": 4870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.639486493586278e-05,
+      "loss": 3.9376,
+      "step": 4871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6385133513799368e-05,
+      "loss": 3.8841,
+      "step": 4872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6375403572731384e-05,
+      "loss": 3.904,
+      "step": 4873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.636567511433151e-05,
+      "loss": 4.0303,
+      "step": 4874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.635594814027217e-05,
+      "loss": 3.907,
+      "step": 4875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6346222652225564e-05,
+      "loss": 3.9385,
+      "step": 4876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6336498651863606e-05,
+      "loss": 3.957,
+      "step": 4877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.632677614085797e-05,
+      "loss": 3.9506,
+      "step": 4878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.631705512088006e-05,
+      "loss": 3.9758,
+      "step": 4879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6307335593601036e-05,
+      "loss": 4.0373,
+      "step": 4880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6297617560691804e-05,
+      "loss": 4.0467,
+      "step": 4881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6287901023822988e-05,
+      "loss": 3.9565,
+      "step": 4882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6278185984664995e-05,
+      "loss": 4.0966,
+      "step": 4883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6268472444887933e-05,
+      "loss": 3.9555,
+      "step": 4884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.625876040616169e-05,
+      "loss": 4.049,
+      "step": 4885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6249049870155873e-05,
+      "loss": 3.9099,
+      "step": 4886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6239340838539827e-05,
+      "loss": 3.9367,
+      "step": 4887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6229633312982657e-05,
+      "loss": 3.9633,
+      "step": 4888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.62199272951532e-05,
+      "loss": 3.8834,
+      "step": 4889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6210222786720013e-05,
+      "loss": 3.9283,
+      "step": 4890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6200519789351444e-05,
+      "loss": 3.9896,
+      "step": 4891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6190818304715532e-05,
+      "loss": 3.8236,
+      "step": 4892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6181118334480073e-05,
+      "loss": 3.9328,
+      "step": 4893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6171419880312605e-05,
+      "loss": 3.9985,
+      "step": 4894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6161722943880416e-05,
+      "loss": 3.9201,
+      "step": 4895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.615202752685052e-05,
+      "loss": 3.8844,
+      "step": 4896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6142333630889658e-05,
+      "loss": 3.8473,
+      "step": 4897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6132641257664337e-05,
+      "loss": 3.8767,
+      "step": 4898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6122950408840785e-05,
+      "loss": 3.9696,
+      "step": 4899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.611326108608496e-05,
+      "loss": 4.0035,
+      "step": 4900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.610357329106259e-05,
+      "loss": 3.9762,
+      "step": 4901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6093887025439106e-05,
+      "loss": 3.9516,
+      "step": 4902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608420229087968e-05,
+      "loss": 3.9666,
+      "step": 4903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607451908904925e-05,
+      "loss": 3.8772,
+      "step": 4904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6064837421612456e-05,
+      "loss": 3.9049,
+      "step": 4905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6055157290233702e-05,
+      "loss": 4.1045,
+      "step": 4906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6045478696577104e-05,
+      "loss": 3.8782,
+      "step": 4907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6035801642306526e-05,
+      "loss": 3.9945,
+      "step": 4908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6026126129085568e-05,
+      "loss": 3.8565,
+      "step": 4909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6016452158577565e-05,
+      "loss": 4.0469,
+      "step": 4910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.600677973244558e-05,
+      "loss": 3.8837,
+      "step": 4911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599710885235241e-05,
+      "loss": 4.0341,
+      "step": 4912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5987439519960606e-05,
+      "loss": 3.8374,
+      "step": 4913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5977771736932436e-05,
+      "loss": 3.8265,
+      "step": 4914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5968105504929893e-05,
+      "loss": 3.8549,
+      "step": 4915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5958440825614722e-05,
+      "loss": 3.8353,
+      "step": 4916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5948777700648394e-05,
+      "loss": 3.8336,
+      "step": 4917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5939116131692107e-05,
+      "loss": 3.7735,
+      "step": 4918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.59294561204068e-05,
+      "loss": 3.971,
+      "step": 4919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5919797668453145e-05,
+      "loss": 3.8522,
+      "step": 4920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5910140777491527e-05,
+      "loss": 3.9055,
+      "step": 4921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5900485449182094e-05,
+      "loss": 3.9411,
+      "step": 4922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5890831685184702e-05,
+      "loss": 3.9814,
+      "step": 4923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5881179487158954e-05,
+      "loss": 3.7868,
+      "step": 4924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5871528856764162e-05,
+      "loss": 4.0386,
+      "step": 4925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.586187979565938e-05,
+      "loss": 3.8916,
+      "step": 4926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5852232305503406e-05,
+      "loss": 4.0542,
+      "step": 4927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.584258638795475e-05,
+      "loss": 3.9312,
+      "step": 4928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5832942044671647e-05,
+      "loss": 3.9112,
+      "step": 4929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5823299277312087e-05,
+      "loss": 3.7825,
+      "step": 4930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5813658087533756e-05,
+      "loss": 4.1175,
+      "step": 4931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5804018476994108e-05,
+      "loss": 3.8284,
+      "step": 4932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5794380447350288e-05,
+      "loss": 3.9104,
+      "step": 4933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5784744000259194e-05,
+      "loss": 4.0238,
+      "step": 4934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5775109137377443e-05,
+      "loss": 3.9402,
+      "step": 4935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5765475860361365e-05,
+      "loss": 3.9424,
+      "step": 4936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.575584417086705e-05,
+      "loss": 3.8808,
+      "step": 4937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5746214070550286e-05,
+      "loss": 3.9934,
+      "step": 4938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.57365855610666e-05,
+      "loss": 3.8485,
+      "step": 4939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5726958644071248e-05,
+      "loss": 3.8554,
+      "step": 4940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5717333321219207e-05,
+      "loss": 3.8902,
+      "step": 4941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5707709594165186e-05,
+      "loss": 3.993,
+      "step": 4942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.569808746456361e-05,
+      "loss": 3.8838,
+      "step": 4943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5688466934068633e-05,
+      "loss": 3.8948,
+      "step": 4944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5678848004334143e-05,
+      "loss": 3.8768,
+      "step": 4945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.566923067701374e-05,
+      "loss": 3.7648,
+      "step": 4946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5659614953760752e-05,
+      "loss": 3.9182,
+      "step": 4947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.565000083622824e-05,
+      "loss": 3.8791,
+      "step": 4948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.564038832606897e-05,
+      "loss": 3.9159,
+      "step": 4949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.563077742493546e-05,
+      "loss": 3.9868,
+      "step": 4950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5621168134479926e-05,
+      "loss": 3.9085,
+      "step": 4951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5611560456354323e-05,
+      "loss": 3.8348,
+      "step": 4952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5601954392210317e-05,
+      "loss": 3.7862,
+      "step": 4953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5592349943699295e-05,
+      "loss": 3.8756,
+      "step": 4954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.558274711247239e-05,
+      "loss": 3.9185,
+      "step": 4955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.557314590018043e-05,
+      "loss": 3.8732,
+      "step": 4956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5563546308473967e-05,
+      "loss": 3.9129,
+      "step": 4957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.555394833900329e-05,
+      "loss": 3.8671,
+      "step": 4958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5544351993418403e-05,
+      "loss": 4.0111,
+      "step": 4959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.553475727336904e-05,
+      "loss": 3.8756,
+      "step": 4960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.552516418050462e-05,
+      "loss": 3.9673,
+      "step": 4961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5515572716474322e-05,
+      "loss": 3.9014,
+      "step": 4962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5505982882927027e-05,
+      "loss": 3.8002,
+      "step": 4963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5496394681511345e-05,
+      "loss": 3.7804,
+      "step": 4964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5486808113875585e-05,
+      "loss": 4.0124,
+      "step": 4965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5477223181667796e-05,
+      "loss": 3.9446,
+      "step": 4966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5467639886535746e-05,
+      "loss": 3.7597,
+      "step": 4967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5458058230126894e-05,
+      "loss": 3.9682,
+      "step": 4968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5448478214088458e-05,
+      "loss": 4.06,
+      "step": 4969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5438899840067354e-05,
+      "loss": 3.862,
+      "step": 4970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.542932310971021e-05,
+      "loss": 3.8663,
+      "step": 4971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5419748024663372e-05,
+      "loss": 3.9856,
+      "step": 4972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.541017458657291e-05,
+      "loss": 3.8933,
+      "step": 4973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5400602797084622e-05,
+      "loss": 3.9239,
+      "step": 4974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5391032657843988e-05,
+      "loss": 3.8216,
+      "step": 4975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.538146417049624e-05,
+      "loss": 3.9729,
+      "step": 4976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5371897336686302e-05,
+      "loss": 3.9879,
+      "step": 4977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.536233215805884e-05,
+      "loss": 3.8899,
+      "step": 4978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5352768636258203e-05,
+      "loss": 3.8125,
+      "step": 4979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5343206772928487e-05,
+      "loss": 4.0787,
+      "step": 4980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5333646569713474e-05,
+      "loss": 3.8559,
+      "step": 4981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.532408802825668e-05,
+      "loss": 4.0386,
+      "step": 4982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5314531150201315e-05,
+      "loss": 3.8886,
+      "step": 4983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.530497593719034e-05,
+      "loss": 3.9004,
+      "step": 4984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.52954223908664e-05,
+      "loss": 3.8027,
+      "step": 4985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5285870512871836e-05,
+      "loss": 3.9193,
+      "step": 4986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5276320304848756e-05,
+      "loss": 3.8525,
+      "step": 4987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5266771768438947e-05,
+      "loss": 3.9467,
+      "step": 4988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5257224905283912e-05,
+      "loss": 3.8319,
+      "step": 4989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5247679717024854e-05,
+      "loss": 3.9254,
+      "step": 4990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.523813620530272e-05,
+      "loss": 4.0143,
+      "step": 4991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5228594371758136e-05,
+      "loss": 3.8924,
+      "step": 4992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5219054218031458e-05,
+      "loss": 3.8678,
+      "step": 4993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5209515745762759e-05,
+      "loss": 3.7766,
+      "step": 4994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.519997895659179e-05,
+      "loss": 3.8273,
+      "step": 4995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.519044385215806e-05,
+      "loss": 3.832,
+      "step": 4996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5180910434100748e-05,
+      "loss": 3.9422,
+      "step": 4997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5171378704058773e-05,
+      "loss": 3.8663,
+      "step": 4998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5161848663670742e-05,
+      "loss": 3.8325,
+      "step": 4999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5152320314574974e-05,
+      "loss": 3.9274,
+      "step": 5000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.514279365840951e-05,
+      "loss": 3.8054,
+      "step": 5001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5133268696812092e-05,
+      "loss": 3.9338,
+      "step": 5002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5123745431420169e-05,
+      "loss": 3.9336,
+      "step": 5003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.51142238638709e-05,
+      "loss": 3.8771,
+      "step": 5004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5104703995801145e-05,
+      "loss": 3.8915,
+      "step": 5005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5095185828847502e-05,
+      "loss": 4.0413,
+      "step": 5006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5085669364646243e-05,
+      "loss": 3.8788,
+      "step": 5007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5076154604833351e-05,
+      "loss": 3.9398,
+      "step": 5008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5066641551044532e-05,
+      "loss": 3.9524,
+      "step": 5009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5057130204915194e-05,
+      "loss": 4.0159,
+      "step": 5010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5047620568080428e-05,
+      "loss": 3.788,
+      "step": 5011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5038112642175073e-05,
+      "loss": 3.8732,
+      "step": 5012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5028606428833643e-05,
+      "loss": 3.9731,
+      "step": 5013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5019101929690357e-05,
+      "loss": 3.9111,
+      "step": 5014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5009599146379164e-05,
+      "loss": 4.007,
+      "step": 5015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5000098080533697e-05,
+      "loss": 4.0194,
+      "step": 5016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4990598733787303e-05,
+      "loss": 3.8548,
+      "step": 5017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.498110110777302e-05,
+      "loss": 3.8182,
+      "step": 5018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.497160520412361e-05,
+      "loss": 3.8779,
+      "step": 5019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4962111024471523e-05,
+      "loss": 3.9579,
+      "step": 5020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4952618570448923e-05,
+      "loss": 3.8679,
+      "step": 5021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.494312784368766e-05,
+      "loss": 3.9055,
+      "step": 5022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.493363884581931e-05,
+      "loss": 3.9985,
+      "step": 5023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4924151578475149e-05,
+      "loss": 3.9247,
+      "step": 5024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.491466604328614e-05,
+      "loss": 3.8873,
+      "step": 5025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4905182241882957e-05,
+      "loss": 3.8594,
+      "step": 5026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4895700175895977e-05,
+      "loss": 3.9384,
+      "step": 5027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4886219846955276e-05,
+      "loss": 3.8442,
+      "step": 5028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4876741256690629e-05,
+      "loss": 3.9562,
+      "step": 5029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4867264406731524e-05,
+      "loss": 4.007,
+      "step": 5030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4857789298707133e-05,
+      "loss": 3.8873,
+      "step": 5031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4848315934246332e-05,
+      "loss": 3.7646,
+      "step": 5032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4838844314977717e-05,
+      "loss": 4.1158,
+      "step": 5033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4829374442529564e-05,
+      "loss": 3.978,
+      "step": 5034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4819906318529859e-05,
+      "loss": 3.9919,
+      "step": 5035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4810439944606264e-05,
+      "loss": 3.8835,
+      "step": 5036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4800975322386177e-05,
+      "loss": 3.8694,
+      "step": 5037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.479151245349667e-05,
+      "loss": 3.9341,
+      "step": 5038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4782051339564512e-05,
+      "loss": 3.9383,
+      "step": 5039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4772591982216194e-05,
+      "loss": 3.9493,
+      "step": 5040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4763134383077876e-05,
+      "loss": 3.9743,
+      "step": 5041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4753678543775429e-05,
+      "loss": 4.007,
+      "step": 5042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4744224465934431e-05,
+      "loss": 3.9833,
+      "step": 5043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.473477215118014e-05,
+      "loss": 3.8543,
+      "step": 5044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4725321601137526e-05,
+      "loss": 3.9598,
+      "step": 5045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4715872817431242e-05,
+      "loss": 3.8784,
+      "step": 5046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.470642580168564e-05,
+      "loss": 3.9481,
+      "step": 5047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.469698055552478e-05,
+      "loss": 3.9837,
+      "step": 5048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4687537080572406e-05,
+      "loss": 3.7882,
+      "step": 5049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4678095378451957e-05,
+      "loss": 4.0309,
+      "step": 5050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4668655450786566e-05,
+      "loss": 3.8467,
+      "step": 5051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4659217299199084e-05,
+      "loss": 3.9268,
+      "step": 5052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4649780925312034e-05,
+      "loss": 3.9589,
+      "step": 5053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4640346330747623e-05,
+      "loss": 3.9858,
+      "step": 5054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4630913517127787e-05,
+      "loss": 3.9343,
+      "step": 5055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4621482486074122e-05,
+      "loss": 3.781,
+      "step": 5056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4612053239207929e-05,
+      "loss": 3.8964,
+      "step": 5057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.460262577815022e-05,
+      "loss": 3.7364,
+      "step": 5058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.459320010452167e-05,
+      "loss": 3.8228,
+      "step": 5059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4583776219942666e-05,
+      "loss": 4.0611,
+      "step": 5060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4574354126033293e-05,
+      "loss": 3.8823,
+      "step": 5061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.456493382441331e-05,
+      "loss": 3.8539,
+      "step": 5062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4555515316702176e-05,
+      "loss": 3.8947,
+      "step": 5063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4546098604519043e-05,
+      "loss": 3.9594,
+      "step": 5064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.453668368948275e-05,
+      "loss": 3.8434,
+      "step": 5065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4527270573211824e-05,
+      "loss": 3.8988,
+      "step": 5066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4517859257324506e-05,
+      "loss": 3.8108,
+      "step": 5067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.45084497434387e-05,
+      "loss": 3.9046,
+      "step": 5068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4499042033172004e-05,
+      "loss": 3.8214,
+      "step": 5069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4489636128141725e-05,
+      "loss": 3.9117,
+      "step": 5070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.448023202996485e-05,
+      "loss": 3.9717,
+      "step": 5071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.447082974025804e-05,
+      "loss": 3.8904,
+      "step": 5072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.446142926063766e-05,
+      "loss": 3.849,
+      "step": 5073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4452030592719757e-05,
+      "loss": 3.8687,
+      "step": 5074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4442633738120087e-05,
+      "loss": 3.9195,
+      "step": 5075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4433238698454071e-05,
+      "loss": 3.9678,
+      "step": 5076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.442384547533682e-05,
+      "loss": 3.9945,
+      "step": 5077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4414454070383144e-05,
+      "loss": 3.7032,
+      "step": 5078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4405064485207516e-05,
+      "loss": 3.7829,
+      "step": 5079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4395676721424146e-05,
+      "loss": 3.9859,
+      "step": 5080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4386290780646871e-05,
+      "loss": 3.8706,
+      "step": 5081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4376906664489264e-05,
+      "loss": 3.9509,
+      "step": 5082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4367524374564556e-05,
+      "loss": 3.8051,
+      "step": 5083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4358143912485672e-05,
+      "loss": 3.9458,
+      "step": 5084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.434876527986522e-05,
+      "loss": 4.0363,
+      "step": 5085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4339388478315496e-05,
+      "loss": 3.9352,
+      "step": 5086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.433001350944847e-05,
+      "loss": 3.998,
+      "step": 5087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4320640374875826e-05,
+      "loss": 3.8362,
+      "step": 5088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4311269076208903e-05,
+      "loss": 3.9981,
+      "step": 5089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4301899615058748e-05,
+      "loss": 3.9187,
+      "step": 5090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4292531993036071e-05,
+      "loss": 3.9471,
+      "step": 5091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4283166211751275e-05,
+      "loss": 3.8902,
+      "step": 5092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4273802272814451e-05,
+      "loss": 3.7108,
+      "step": 5093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4264440177835365e-05,
+      "loss": 3.8598,
+      "step": 5094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4255079928423456e-05,
+      "loss": 3.9997,
+      "step": 5095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4245721526187882e-05,
+      "loss": 3.99,
+      "step": 5096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4236364972737447e-05,
+      "loss": 3.9526,
+      "step": 5097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4227010269680662e-05,
+      "loss": 3.8898,
+      "step": 5098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4217657418625707e-05,
+      "loss": 3.8647,
+      "step": 5099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.420830642118044e-05,
+      "loss": 3.912,
+      "step": 5100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4198957278952407e-05,
+      "loss": 3.942,
+      "step": 5101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4189609993548825e-05,
+      "loss": 3.9244,
+      "step": 5102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4180264566576617e-05,
+      "loss": 3.9868,
+      "step": 5103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4170920999642362e-05,
+      "loss": 3.9866,
+      "step": 5104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4161579294352334e-05,
+      "loss": 3.9264,
+      "step": 5105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.415223945231246e-05,
+      "loss": 3.8404,
+      "step": 5106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4142901475128394e-05,
+      "loss": 3.8625,
+      "step": 5107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4133565364405433e-05,
+      "loss": 3.8823,
+      "step": 5108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.412423112174856e-05,
+      "loss": 3.8972,
+      "step": 5109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4114898748762428e-05,
+      "loss": 3.9131,
+      "step": 5110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4105568247051405e-05,
+      "loss": 3.9089,
+      "step": 5111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4096239618219493e-05,
+      "loss": 3.9615,
+      "step": 5112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4086912863870402e-05,
+      "loss": 3.8286,
+      "step": 5113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4077587985607505e-05,
+      "loss": 3.9706,
+      "step": 5114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4068264985033858e-05,
+      "loss": 3.9619,
+      "step": 5115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4058943863752178e-05,
+      "loss": 3.7996,
+      "step": 5116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4049624623364891e-05,
+      "loss": 3.9594,
+      "step": 5117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4040307265474087e-05,
+      "loss": 3.8693,
+      "step": 5118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.403099179168152e-05,
+      "loss": 3.9722,
+      "step": 5119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.402167820358863e-05,
+      "loss": 3.9079,
+      "step": 5120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4012366502796525e-05,
+      "loss": 3.9165,
+      "step": 5121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4003056690906e-05,
+      "loss": 3.8699,
+      "step": 5122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3993748769517506e-05,
+      "loss": 4.0126,
+      "step": 5123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3984442740231201e-05,
+      "loss": 3.9788,
+      "step": 5124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3975138604646887e-05,
+      "loss": 4.0052,
+      "step": 5125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3965836364364068e-05,
+      "loss": 3.8436,
+      "step": 5126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3956536020981897e-05,
+      "loss": 3.7856,
+      "step": 5127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3947237576099209e-05,
+      "loss": 4.1009,
+      "step": 5128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3937941031314516e-05,
+      "loss": 4.0004,
+      "step": 5129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3928646388226003e-05,
+      "loss": 3.9251,
+      "step": 5130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3919353648431516e-05,
+      "loss": 3.9505,
+      "step": 5131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3910062813528605e-05,
+      "loss": 3.9698,
+      "step": 5132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.390077388511446e-05,
+      "loss": 3.9441,
+      "step": 5133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3891486864785949e-05,
+      "loss": 3.8941,
+      "step": 5134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3882201754139637e-05,
+      "loss": 3.9587,
+      "step": 5135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.387291855477173e-05,
+      "loss": 3.8553,
+      "step": 5136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3863637268278123e-05,
+      "loss": 3.8852,
+      "step": 5137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.385435789625436e-05,
+      "loss": 4.0033,
+      "step": 5138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.38450804402957e-05,
+      "loss": 3.9487,
+      "step": 5139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3835804901997029e-05,
+      "loss": 4.1152,
+      "step": 5140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.382653128295292e-05,
+      "loss": 3.876,
+      "step": 5141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.381725958475762e-05,
+      "loss": 3.9309,
+      "step": 5142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.380798980900503e-05,
+      "loss": 3.9531,
+      "step": 5143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3798721957288746e-05,
+      "loss": 4.0825,
+      "step": 5144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.378945603120202e-05,
+      "loss": 3.9862,
+      "step": 5145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3780192032337752e-05,
+      "loss": 3.9671,
+      "step": 5146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3770929962288554e-05,
+      "loss": 3.966,
+      "step": 5147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3761669822646678e-05,
+      "loss": 3.7997,
+      "step": 5148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.375241161500404e-05,
+      "loss": 3.789,
+      "step": 5149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3743155340952243e-05,
+      "loss": 3.8675,
+      "step": 5150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3733901002082544e-05,
+      "loss": 3.8566,
+      "step": 5151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3724648599985856e-05,
+      "loss": 3.9107,
+      "step": 5152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3715398136252794e-05,
+      "loss": 3.9578,
+      "step": 5153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.370614961247362e-05,
+      "loss": 4.0357,
+      "step": 5154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3696903030238262e-05,
+      "loss": 3.8645,
+      "step": 5155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3687658391136304e-05,
+      "loss": 3.9764,
+      "step": 5156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3678415696757016e-05,
+      "loss": 3.8884,
+      "step": 5157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.366917494868932e-05,
+      "loss": 3.7948,
+      "step": 5158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.36599361485218e-05,
+      "loss": 3.9293,
+      "step": 5159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3650699297842729e-05,
+      "loss": 3.9676,
+      "step": 5160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3641464398240022e-05,
+      "loss": 3.9291,
+      "step": 5161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3632231451301256e-05,
+      "loss": 3.9152,
+      "step": 5162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3623000458613699e-05,
+      "loss": 4.0322,
+      "step": 5163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3613771421764255e-05,
+      "loss": 3.9465,
+      "step": 5164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3604544342339506e-05,
+      "loss": 3.9122,
+      "step": 5165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.359531922192569e-05,
+      "loss": 3.7937,
+      "step": 5166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3586096062108709e-05,
+      "loss": 3.9444,
+      "step": 5167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3576874864474143e-05,
+      "loss": 3.9256,
+      "step": 5168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3567655630607212e-05,
+      "loss": 3.8995,
+      "step": 5169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3558438362092818e-05,
+      "loss": 3.9689,
+      "step": 5170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3549223060515503e-05,
+      "loss": 3.8348,
+      "step": 5171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3540009727459499e-05,
+      "loss": 3.8923,
+      "step": 5172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.353079836450868e-05,
+      "loss": 4.0018,
+      "step": 5173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3521588973246574e-05,
+      "loss": 3.8129,
+      "step": 5174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3512381555256403e-05,
+      "loss": 3.8405,
+      "step": 5175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3503176112121018e-05,
+      "loss": 3.8077,
+      "step": 5176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3493972645422943e-05,
+      "loss": 3.9275,
+      "step": 5177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3484771156744357e-05,
+      "loss": 4.0283,
+      "step": 5178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.347557164766711e-05,
+      "loss": 3.8793,
+      "step": 5179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3466374119772684e-05,
+      "loss": 3.9502,
+      "step": 5180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.345717857464226e-05,
+      "loss": 3.9393,
+      "step": 5181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3447985013856664e-05,
+      "loss": 3.9161,
+      "step": 5182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3438793438996367e-05,
+      "loss": 3.8766,
+      "step": 5183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3429603851641507e-05,
+      "loss": 3.9646,
+      "step": 5184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3420416253371878e-05,
+      "loss": 3.7572,
+      "step": 5185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3411230645766936e-05,
+      "loss": 3.8981,
+      "step": 5186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3402047030405801e-05,
+      "loss": 3.9388,
+      "step": 5187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3392865408867223e-05,
+      "loss": 3.9351,
+      "step": 5188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.338368578272965e-05,
+      "loss": 3.9412,
+      "step": 5189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3374508153571153e-05,
+      "loss": 3.7812,
+      "step": 5190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3365332522969487e-05,
+      "loss": 4.1151,
+      "step": 5191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3356158892502038e-05,
+      "loss": 3.9649,
+      "step": 5192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3346987263745864e-05,
+      "loss": 3.9222,
+      "step": 5193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3337817638277675e-05,
+      "loss": 3.9044,
+      "step": 5194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3328650017673822e-05,
+      "loss": 3.7934,
+      "step": 5195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3319484403510346e-05,
+      "loss": 3.8485,
+      "step": 5196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3310320797362915e-05,
+      "loss": 3.9413,
+      "step": 5197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3301159200806857e-05,
+      "loss": 3.9835,
+      "step": 5198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3291999615417148e-05,
+      "loss": 4.0802,
+      "step": 5199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3282842042768446e-05,
+      "loss": 3.9181,
+      "step": 5200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.327368648443503e-05,
+      "loss": 3.9539,
+      "step": 5201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3264532941990854e-05,
+      "loss": 4.0221,
+      "step": 5202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3255381417009502e-05,
+      "loss": 3.827,
+      "step": 5203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3246231911064249e-05,
+      "loss": 3.7622,
+      "step": 5204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3237084425727997e-05,
+      "loss": 4.0798,
+      "step": 5205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3227938962573294e-05,
+      "loss": 3.9509,
+      "step": 5206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3218795523172362e-05,
+      "loss": 3.8761,
+      "step": 5207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3209654109097045e-05,
+      "loss": 3.9156,
+      "step": 5208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3200514721918883e-05,
+      "loss": 3.8019,
+      "step": 5209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3191377363209023e-05,
+      "loss": 3.8945,
+      "step": 5210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3182242034538301e-05,
+      "loss": 3.9322,
+      "step": 5211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3173108737477175e-05,
+      "loss": 3.9625,
+      "step": 5212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3163977473595767e-05,
+      "loss": 4.1541,
+      "step": 5213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3154848244463847e-05,
+      "loss": 4.0389,
+      "step": 5214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3145721051650833e-05,
+      "loss": 4.004,
+      "step": 5215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3136595896725787e-05,
+      "loss": 3.9114,
+      "step": 5216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3127472781257438e-05,
+      "loss": 3.9086,
+      "step": 5217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3118351706814168e-05,
+      "loss": 3.9173,
+      "step": 5218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3109232674963983e-05,
+      "loss": 4.0099,
+      "step": 5219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.310011568727455e-05,
+      "loss": 3.9625,
+      "step": 5220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3091000745313186e-05,
+      "loss": 3.8799,
+      "step": 5221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3081887850646856e-05,
+      "loss": 3.965,
+      "step": 5222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.307277700484217e-05,
+      "loss": 3.7473,
+      "step": 5223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3063668209465379e-05,
+      "loss": 4.0826,
+      "step": 5224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3054561466082412e-05,
+      "loss": 3.9225,
+      "step": 5225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3045456776258813e-05,
+      "loss": 4.0809,
+      "step": 5226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.303635414155977e-05,
+      "loss": 3.9593,
+      "step": 5227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3027253563550157e-05,
+      "loss": 3.9706,
+      "step": 5228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3018155043794455e-05,
+      "loss": 3.8605,
+      "step": 5229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.300905858385681e-05,
+      "loss": 3.9552,
+      "step": 5230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2999964185300989e-05,
+      "loss": 3.9758,
+      "step": 5231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2990871849690456e-05,
+      "loss": 3.9127,
+      "step": 5232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.298178157858827e-05,
+      "loss": 3.9386,
+      "step": 5233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.297269337355716e-05,
+      "loss": 3.8985,
+      "step": 5234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2963607236159492e-05,
+      "loss": 3.9988,
+      "step": 5235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2954523167957266e-05,
+      "loss": 3.9825,
+      "step": 5236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2945441170512163e-05,
+      "loss": 4.0057,
+      "step": 5237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2936361245385459e-05,
+      "loss": 3.8598,
+      "step": 5238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2927283394138123e-05,
+      "loss": 3.9909,
+      "step": 5239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.291820761833073e-05,
+      "loss": 3.8104,
+      "step": 5240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2909133919523508e-05,
+      "loss": 3.8708,
+      "step": 5241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2900062299276339e-05,
+      "loss": 4.0411,
+      "step": 5242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2890992759148734e-05,
+      "loss": 4.0267,
+      "step": 5243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2881925300699854e-05,
+      "loss": 3.9581,
+      "step": 5244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.287285992548849e-05,
+      "loss": 3.9687,
+      "step": 5245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2863796635073092e-05,
+      "loss": 3.7945,
+      "step": 5246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2854735431011758e-05,
+      "loss": 3.8192,
+      "step": 5247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2845676314862203e-05,
+      "loss": 3.9025,
+      "step": 5248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2836619288181795e-05,
+      "loss": 3.9079,
+      "step": 5249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2827564352527544e-05,
+      "loss": 3.9626,
+      "step": 5250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2818511509456094e-05,
+      "loss": 3.9206,
+      "step": 5251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2809460760523723e-05,
+      "loss": 3.9209,
+      "step": 5252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2800412107286386e-05,
+      "loss": 3.9073,
+      "step": 5253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2791365551299625e-05,
+      "loss": 3.9786,
+      "step": 5254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2782321094118671e-05,
+      "loss": 3.9628,
+      "step": 5255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.277327873729836e-05,
+      "loss": 3.8534,
+      "step": 5256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2764238482393181e-05,
+      "loss": 3.8882,
+      "step": 5257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2755200330957251e-05,
+      "loss": 3.8409,
+      "step": 5258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2746164284544332e-05,
+      "loss": 4.0267,
+      "step": 5259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2737130344707843e-05,
+      "loss": 3.9787,
+      "step": 5260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2728098513000806e-05,
+      "loss": 3.8814,
+      "step": 5261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2719068790975904e-05,
+      "loss": 3.9276,
+      "step": 5262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.271004118018545e-05,
+      "loss": 3.935,
+      "step": 5263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2701015682181386e-05,
+      "loss": 3.9141,
+      "step": 5264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2691992298515318e-05,
+      "loss": 3.9907,
+      "step": 5265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.268297103073846e-05,
+      "loss": 3.9039,
+      "step": 5266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2673951880401663e-05,
+      "loss": 3.9769,
+      "step": 5267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2664934849055443e-05,
+      "loss": 3.9713,
+      "step": 5268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.265591993824992e-05,
+      "loss": 3.7874,
+      "step": 5269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.264690714953487e-05,
+      "loss": 3.948,
+      "step": 5270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2637896484459688e-05,
+      "loss": 4.004,
+      "step": 5271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2628887944573414e-05,
+      "loss": 3.9097,
+      "step": 5272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2619881531424713e-05,
+      "loss": 3.8645,
+      "step": 5273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2610877246561898e-05,
+      "loss": 3.9563,
+      "step": 5274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.260187509153292e-05,
+      "loss": 3.8523,
+      "step": 5275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2592875067885346e-05,
+      "loss": 3.8893,
+      "step": 5276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.258387717716638e-05,
+      "loss": 3.8645,
+      "step": 5277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2574881420922874e-05,
+      "loss": 3.8723,
+      "step": 5278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2565887800701293e-05,
+      "loss": 3.9319,
+      "step": 5279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2556896318047733e-05,
+      "loss": 3.9318,
+      "step": 5280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2547906974507967e-05,
+      "loss": 3.9723,
+      "step": 5281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2538919771627333e-05,
+      "loss": 3.8998,
+      "step": 5282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2529934710950863e-05,
+      "loss": 3.9751,
+      "step": 5283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2520951794023183e-05,
+      "loss": 4.0254,
+      "step": 5284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2511971022388559e-05,
+      "loss": 3.9653,
+      "step": 5285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2502992397590891e-05,
+      "loss": 4.1315,
+      "step": 5286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2494015921173704e-05,
+      "loss": 3.8878,
+      "step": 5287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2485041594680155e-05,
+      "loss": 3.8348,
+      "step": 5288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.247606941965305e-05,
+      "loss": 4.0146,
+      "step": 5289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2467099397634801e-05,
+      "loss": 3.9502,
+      "step": 5290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2458131530167453e-05,
+      "loss": 3.9531,
+      "step": 5291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.24491658187927e-05,
+      "loss": 3.9191,
+      "step": 5292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2440202265051844e-05,
+      "loss": 3.9864,
+      "step": 5293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2431240870485824e-05,
+      "loss": 3.924,
+      "step": 5294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2422281636635202e-05,
+      "loss": 3.8581,
+      "step": 5295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2413324565040186e-05,
+      "loss": 3.9356,
+      "step": 5296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2404369657240597e-05,
+      "loss": 4.0238,
+      "step": 5297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2395416914775881e-05,
+      "loss": 3.8837,
+      "step": 5298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2386466339185127e-05,
+      "loss": 4.0878,
+      "step": 5299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2377517932007035e-05,
+      "loss": 3.9726,
+      "step": 5300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2368571694779935e-05,
+      "loss": 3.9362,
+      "step": 5301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2359627629041807e-05,
+      "loss": 3.8234,
+      "step": 5302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2350685736330218e-05,
+      "loss": 3.9023,
+      "step": 5303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.23417460181824e-05,
+      "loss": 3.8961,
+      "step": 5304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2332808476135194e-05,
+      "loss": 3.9252,
+      "step": 5305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2323873111725062e-05,
+      "loss": 3.9639,
+      "step": 5306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2314939926488095e-05,
+      "loss": 4.0485,
+      "step": 5307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.230600892196001e-05,
+      "loss": 3.9303,
+      "step": 5308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2297080099676145e-05,
+      "loss": 3.9735,
+      "step": 5309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2288153461171478e-05,
+      "loss": 3.8463,
+      "step": 5310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2279229007980606e-05,
+      "loss": 4.0595,
+      "step": 5311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.227030674163774e-05,
+      "loss": 3.9732,
+      "step": 5312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2261386663676722e-05,
+      "loss": 3.8588,
+      "step": 5313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2252468775631013e-05,
+      "loss": 3.8847,
+      "step": 5314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2243553079033704e-05,
+      "loss": 3.9624,
+      "step": 5315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2234639575417497e-05,
+      "loss": 4.0099,
+      "step": 5316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2225728266314745e-05,
+      "loss": 3.8448,
+      "step": 5317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2216819153257398e-05,
+      "loss": 3.7976,
+      "step": 5318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2207912237777022e-05,
+      "loss": 4.0456,
+      "step": 5319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.219900752140484e-05,
+      "loss": 3.7973,
+      "step": 5320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.219010500567167e-05,
+      "loss": 3.9048,
+      "step": 5321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2181204692107954e-05,
+      "loss": 3.9544,
+      "step": 5322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2172306582243756e-05,
+      "loss": 3.9973,
+      "step": 5323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2163410677608762e-05,
+      "loss": 3.8943,
+      "step": 5324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2154516979732294e-05,
+      "loss": 3.7658,
+      "step": 5325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2145625490143275e-05,
+      "loss": 3.924,
+      "step": 5326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2136736210370255e-05,
+      "loss": 3.8438,
+      "step": 5327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2127849141941397e-05,
+      "loss": 3.9411,
+      "step": 5328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2118964286384508e-05,
+      "loss": 3.9162,
+      "step": 5329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2110081645226986e-05,
+      "loss": 3.898,
+      "step": 5330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2101201219995854e-05,
+      "loss": 3.9177,
+      "step": 5331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2092323012217773e-05,
+      "loss": 3.9638,
+      "step": 5332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.208344702341901e-05,
+      "loss": 4.0043,
+      "step": 5333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2074573255125444e-05,
+      "loss": 3.8042,
+      "step": 5334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2065701708862578e-05,
+      "loss": 3.8885,
+      "step": 5335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2056832386155536e-05,
+      "loss": 3.913,
+      "step": 5336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.204796528852905e-05,
+      "loss": 3.8308,
+      "step": 5337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.203910041750749e-05,
+      "loss": 3.9612,
+      "step": 5338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2030237774614817e-05,
+      "loss": 4.017,
+      "step": 5339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2021377361374637e-05,
+      "loss": 3.8579,
+      "step": 5340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.201251917931015e-05,
+      "loss": 3.9469,
+      "step": 5341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2003663229944179e-05,
+      "loss": 3.943,
+      "step": 5342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1994809514799165e-05,
+      "loss": 3.9764,
+      "step": 5343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1985958035397168e-05,
+      "loss": 3.8728,
+      "step": 5344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1977108793259844e-05,
+      "loss": 3.8608,
+      "step": 5345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1968261789908502e-05,
+      "loss": 3.9411,
+      "step": 5346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1959417026864025e-05,
+      "loss": 4.01,
+      "step": 5347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1950574505646952e-05,
+      "loss": 4.0369,
+      "step": 5348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1941734227777404e-05,
+      "loss": 3.9887,
+      "step": 5349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1932896194775126e-05,
+      "loss": 3.9017,
+      "step": 5350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1924060408159477e-05,
+      "loss": 4.0394,
+      "step": 5351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1915226869449425e-05,
+      "loss": 3.9251,
+      "step": 5352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1906395580163577e-05,
+      "loss": 3.8635,
+      "step": 5353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1897566541820122e-05,
+      "loss": 4.0595,
+      "step": 5354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1888739755936873e-05,
+      "loss": 4.0188,
+      "step": 5355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1879915224031248e-05,
+      "loss": 4.0216,
+      "step": 5356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.187109294762031e-05,
+      "loss": 3.7641,
+      "step": 5357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1862272928220696e-05,
+      "loss": 3.9248,
+      "step": 5358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1853455167348673e-05,
+      "loss": 3.8778,
+      "step": 5359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1844639666520104e-05,
+      "loss": 3.9509,
+      "step": 5360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1835826427250495e-05,
+      "loss": 3.9817,
+      "step": 5361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1827015451054938e-05,
+      "loss": 3.8979,
+      "step": 5362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1818206739448138e-05,
+      "loss": 3.9372,
+      "step": 5363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1809400293944415e-05,
+      "loss": 3.7858,
+      "step": 5364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1800596116057689e-05,
+      "loss": 3.7829,
+      "step": 5365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1791794207301524e-05,
+      "loss": 3.9265,
+      "step": 5366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1782994569189047e-05,
+      "loss": 3.8593,
+      "step": 5367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1774197203233037e-05,
+      "loss": 3.9713,
+      "step": 5368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1765402110945852e-05,
+      "loss": 3.8404,
+      "step": 5369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1756609293839477e-05,
+      "loss": 3.85,
+      "step": 5370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1747818753425494e-05,
+      "loss": 3.8792,
+      "step": 5371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1739030491215097e-05,
+      "loss": 3.8176,
+      "step": 5372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1730244508719088e-05,
+      "loss": 3.894,
+      "step": 5373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.172146080744789e-05,
+      "loss": 3.8335,
+      "step": 5374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.171267938891151e-05,
+      "loss": 3.8525,
+      "step": 5375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.170390025461959e-05,
+      "loss": 3.9563,
+      "step": 5376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.169512340608136e-05,
+      "loss": 4.0507,
+      "step": 5377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1686348844805659e-05,
+      "loss": 4.0269,
+      "step": 5378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1677576572300936e-05,
+      "loss": 4.0527,
+      "step": 5379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1668806590075249e-05,
+      "loss": 3.9526,
+      "step": 5380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1660038899636247e-05,
+      "loss": 3.9056,
+      "step": 5381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1651273502491216e-05,
+      "loss": 3.942,
+      "step": 5382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1642510400147024e-05,
+      "loss": 3.9183,
+      "step": 5383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1633749594110138e-05,
+      "loss": 3.9204,
+      "step": 5384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.162499108588666e-05,
+      "loss": 3.937,
+      "step": 5385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.161623487698227e-05,
+      "loss": 3.8995,
+      "step": 5386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1607480968902264e-05,
+      "loss": 3.9696,
+      "step": 5387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1598729363151531e-05,
+      "loss": 3.8873,
+      "step": 5388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1589980061234588e-05,
+      "loss": 3.8053,
+      "step": 5389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1581233064655536e-05,
+      "loss": 3.8786,
+      "step": 5390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1572488374918084e-05,
+      "loss": 3.9511,
+      "step": 5391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1563745993525543e-05,
+      "loss": 3.9389,
+      "step": 5392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1555005921980824e-05,
+      "loss": 3.933,
+      "step": 5393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1546268161786467e-05,
+      "loss": 3.8518,
+      "step": 5394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.153753271444458e-05,
+      "loss": 3.9124,
+      "step": 5395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1528799581456879e-05,
+      "loss": 3.888,
+      "step": 5396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1520068764324712e-05,
+      "loss": 3.9735,
+      "step": 5397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1511340264548998e-05,
+      "loss": 3.9471,
+      "step": 5398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1502614083630264e-05,
+      "loss": 4.0598,
+      "step": 5399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1493890223068645e-05,
+      "loss": 3.9556,
+      "step": 5400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1485168684363875e-05,
+      "loss": 3.9314,
+      "step": 5401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1476449469015277e-05,
+      "loss": 3.9936,
+      "step": 5402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.146773257852179e-05,
+      "loss": 3.885,
+      "step": 5403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1459018014381965e-05,
+      "loss": 3.999,
+      "step": 5404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.145030577809392e-05,
+      "loss": 3.9436,
+      "step": 5405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1441595871155397e-05,
+      "loss": 3.8959,
+      "step": 5406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1432888295063723e-05,
+      "loss": 3.904,
+      "step": 5407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1424183051315838e-05,
+      "loss": 3.8652,
+      "step": 5408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1415480141408257e-05,
+      "loss": 3.8748,
+      "step": 5409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1406779566837138e-05,
+      "loss": 3.8723,
+      "step": 5410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1398081329098199e-05,
+      "loss": 3.968,
+      "step": 5411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1389385429686753e-05,
+      "loss": 3.7299,
+      "step": 5412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.138069187009775e-05,
+      "loss": 3.8869,
+      "step": 5413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1372000651825704e-05,
+      "loss": 3.9698,
+      "step": 5414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1363311776364735e-05,
+      "loss": 3.9618,
+      "step": 5415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.135462524520856e-05,
+      "loss": 3.8928,
+      "step": 5416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.134594105985049e-05,
+      "loss": 3.8594,
+      "step": 5417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1337259221783453e-05,
+      "loss": 3.9525,
+      "step": 5418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1328579732499945e-05,
+      "loss": 3.8812,
+      "step": 5419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1319902593492073e-05,
+      "loss": 3.9359,
+      "step": 5420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1311227806251531e-05,
+      "loss": 3.8048,
+      "step": 5421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1302555372269633e-05,
+      "loss": 3.8729,
+      "step": 5422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.129388529303726e-05,
+      "loss": 3.9844,
+      "step": 5423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.128521757004489e-05,
+      "loss": 3.8137,
+      "step": 5424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1276552204782626e-05,
+      "loss": 3.9824,
+      "step": 5425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.126788919874013e-05,
+      "loss": 3.9174,
+      "step": 5426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1259228553406681e-05,
+      "loss": 3.7911,
+      "step": 5427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.125057027027114e-05,
+      "loss": 3.66,
+      "step": 5428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1241914350821966e-05,
+      "loss": 3.9692,
+      "step": 5429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1233260796547202e-05,
+      "loss": 3.958,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1224609608934505e-05,
+      "loss": 3.9455,
+      "step": 5431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1215960789471125e-05,
+      "loss": 3.8885,
+      "step": 5432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1207314339643885e-05,
+      "loss": 4.1253,
+      "step": 5433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1198670260939206e-05,
+      "loss": 3.8045,
+      "step": 5434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1190028554843107e-05,
+      "loss": 3.796,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1181389222841201e-05,
+      "loss": 3.8601,
+      "step": 5436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1172752266418685e-05,
+      "loss": 3.9154,
+      "step": 5437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1164117687060347e-05,
+      "loss": 3.9509,
+      "step": 5438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1155485486250575e-05,
+      "loss": 3.8604,
+      "step": 5439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1146855665473355e-05,
+      "loss": 3.9486,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1138228226212249e-05,
+      "loss": 3.9394,
+      "step": 5441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.112960316995041e-05,
+      "loss": 3.9684,
+      "step": 5442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1120980498170582e-05,
+      "loss": 4.0005,
+      "step": 5443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.111236021235511e-05,
+      "loss": 4.0041,
+      "step": 5444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1103742313985906e-05,
+      "loss": 4.0042,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1095126804544507e-05,
+      "loss": 3.8206,
+      "step": 5446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.108651368551201e-05,
+      "loss": 3.9974,
+      "step": 5447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1077902958369108e-05,
+      "loss": 3.8408,
+      "step": 5448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1069294624596077e-05,
+      "loss": 3.8072,
+      "step": 5449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.106068868567281e-05,
+      "loss": 3.9089,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1052085143078753e-05,
+      "loss": 3.8356,
+      "step": 5451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.104348399829295e-05,
+      "loss": 3.9011,
+      "step": 5452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1034885252794055e-05,
+      "loss": 3.8264,
+      "step": 5453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1026288908060284e-05,
+      "loss": 3.902,
+      "step": 5454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1017694965569447e-05,
+      "loss": 3.8766,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1009103426798939e-05,
+      "loss": 3.8854,
+      "step": 5456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.100051429322575e-05,
+      "loss": 3.8799,
+      "step": 5457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0991927566326443e-05,
+      "loss": 3.9122,
+      "step": 5458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0983343247577188e-05,
+      "loss": 3.9487,
+      "step": 5459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0974761338453717e-05,
+      "loss": 3.7939,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0966181840431376e-05,
+      "loss": 3.973,
+      "step": 5461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0957604754985071e-05,
+      "loss": 3.9703,
+      "step": 5462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.09490300835893e-05,
+      "loss": 3.9599,
+      "step": 5463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.094045782771815e-05,
+      "loss": 3.8537,
+      "step": 5464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0931887988845294e-05,
+      "loss": 3.9039,
+      "step": 5465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0923320568443972e-05,
+      "loss": 3.9346,
+      "step": 5466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.091475556798704e-05,
+      "loss": 3.9543,
+      "step": 5467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0906192988946923e-05,
+      "loss": 3.9419,
+      "step": 5468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0897632832795617e-05,
+      "loss": 3.8695,
+      "step": 5469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.088907510100472e-05,
+      "loss": 3.8894,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.08805197950454e-05,
+      "loss": 3.8841,
+      "step": 5471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0871966916388416e-05,
+      "loss": 3.8362,
+      "step": 5472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0863416466504092e-05,
+      "loss": 3.8774,
+      "step": 5473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0854868446862373e-05,
+      "loss": 4.032,
+      "step": 5474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0846322858932756e-05,
+      "loss": 3.8839,
+      "step": 5475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0837779704184311e-05,
+      "loss": 3.9787,
+      "step": 5476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0829238984085726e-05,
+      "loss": 3.9722,
+      "step": 5477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0820700700105243e-05,
+      "loss": 3.967,
+      "step": 5478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0812164853710688e-05,
+      "loss": 3.8947,
+      "step": 5479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0803631446369478e-05,
+      "loss": 3.9788,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0795100479548587e-05,
+      "loss": 3.8962,
+      "step": 5481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0786571954714614e-05,
+      "loss": 3.9238,
+      "step": 5482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0778045873333695e-05,
+      "loss": 3.9202,
+      "step": 5483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0769522236871569e-05,
+      "loss": 3.8114,
+      "step": 5484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.076100104679354e-05,
+      "loss": 3.9708,
+      "step": 5485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0752482304564496e-05,
+      "loss": 3.909,
+      "step": 5486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0743966011648927e-05,
+      "loss": 3.879,
+      "step": 5487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0735452169510859e-05,
+      "loss": 3.9242,
+      "step": 5488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0726940779613943e-05,
+      "loss": 3.7274,
+      "step": 5489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.071843184342137e-05,
+      "loss": 3.9463,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0709925362395934e-05,
+      "loss": 3.8841,
+      "step": 5491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.070142133799999e-05,
+      "loss": 3.9276,
+      "step": 5492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0692919771695483e-05,
+      "loss": 3.9642,
+      "step": 5493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.068442066494392e-05,
+      "loss": 3.9843,
+      "step": 5494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0675924019206416e-05,
+      "loss": 3.911,
+      "step": 5495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0667429835943619e-05,
+      "loss": 4.06,
+      "step": 5496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0658938116615803e-05,
+      "loss": 4.0598,
+      "step": 5497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0650448862682777e-05,
+      "loss": 3.9111,
+      "step": 5498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0641962075603949e-05,
+      "loss": 3.8548,
+      "step": 5499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0633477756838292e-05,
+      "loss": 3.8095,
+      "step": 5500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.062499590784436e-05,
+      "loss": 4.0034,
+      "step": 5501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.061651653008027e-05,
+      "loss": 3.8922,
+      "step": 5502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0608039625003746e-05,
+      "loss": 3.7982,
+      "step": 5503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0599565194072047e-05,
+      "loss": 4.0913,
+      "step": 5504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0591093238742047e-05,
+      "loss": 3.8718,
+      "step": 5505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0582623760470158e-05,
+      "loss": 3.9476,
+      "step": 5506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.057415676071239e-05,
+      "loss": 3.8665,
+      "step": 5507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0565692240924308e-05,
+      "loss": 3.9509,
+      "step": 5508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0557230202561061e-05,
+      "loss": 3.9365,
+      "step": 5509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0548770647077385e-05,
+      "loss": 3.8311,
+      "step": 5510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.054031357592757e-05,
+      "loss": 3.8686,
+      "step": 5511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0531858990565477e-05,
+      "loss": 3.883,
+      "step": 5512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0523406892444548e-05,
+      "loss": 3.9888,
+      "step": 5513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.051495728301781e-05,
+      "loss": 3.8057,
+      "step": 5514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.050651016373784e-05,
+      "loss": 3.8428,
+      "step": 5515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0498065536056794e-05,
+      "loss": 3.97,
+      "step": 5516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0489623401426396e-05,
+      "loss": 3.9594,
+      "step": 5517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.048118376129796e-05,
+      "loss": 3.9471,
+      "step": 5518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0472746617122355e-05,
+      "loss": 3.888,
+      "step": 5519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0464311970350022e-05,
+      "loss": 3.7195,
+      "step": 5520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.045587982243097e-05,
+      "loss": 3.8878,
+      "step": 5521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0447450174814787e-05,
+      "loss": 3.8933,
+      "step": 5522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.043902302895062e-05,
+      "loss": 3.8611,
+      "step": 5523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.04305983862872e-05,
+      "loss": 3.8009,
+      "step": 5524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0422176248272825e-05,
+      "loss": 4.0004,
+      "step": 5525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0413756616355359e-05,
+      "loss": 3.8785,
+      "step": 5526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0405339491982225e-05,
+      "loss": 3.7687,
+      "step": 5527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0396924876600427e-05,
+      "loss": 3.9629,
+      "step": 5528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0388512771656539e-05,
+      "loss": 3.9193,
+      "step": 5529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0380103178596688e-05,
+      "loss": 4.0128,
+      "step": 5530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0371696098866595e-05,
+      "loss": 3.9819,
+      "step": 5531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0363291533911523e-05,
+      "loss": 3.8585,
+      "step": 5532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0354889485176328e-05,
+      "loss": 3.9321,
+      "step": 5533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.034648995410541e-05,
+      "loss": 3.878,
+      "step": 5534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0338092942142746e-05,
+      "loss": 4.0581,
+      "step": 5535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0329698450731881e-05,
+      "loss": 3.8852,
+      "step": 5536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0321306481315925e-05,
+      "loss": 3.8895,
+      "step": 5537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0312917035337546e-05,
+      "loss": 3.8969,
+      "step": 5538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0304530114239008e-05,
+      "loss": 3.9215,
+      "step": 5539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0296145719462106e-05,
+      "loss": 3.812,
+      "step": 5540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0287763852448207e-05,
+      "loss": 3.9735,
+      "step": 5541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0279384514638269e-05,
+      "loss": 3.8432,
+      "step": 5542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0271007707472788e-05,
+      "loss": 3.892,
+      "step": 5543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0262633432391838e-05,
+      "loss": 3.8677,
+      "step": 5544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.025426169083504e-05,
+      "loss": 3.9016,
+      "step": 5545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0245892484241615e-05,
+      "loss": 3.9162,
+      "step": 5546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0237525814050316e-05,
+      "loss": 3.8871,
+      "step": 5547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.022916168169947e-05,
+      "loss": 3.8678,
+      "step": 5548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0220800088626969e-05,
+      "loss": 3.8344,
+      "step": 5549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.021244103627027e-05,
+      "loss": 3.9296,
+      "step": 5550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0204084526066379e-05,
+      "loss": 3.8939,
+      "step": 5551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0195730559451894e-05,
+      "loss": 3.8095,
+      "step": 5552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0187379137862946e-05,
+      "loss": 3.9184,
+      "step": 5553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0179030262735254e-05,
+      "loss": 3.9592,
+      "step": 5554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0170683935504078e-05,
+      "loss": 3.9462,
+      "step": 5555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0162340157604253e-05,
+      "loss": 3.8554,
+      "step": 5556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0153998930470166e-05,
+      "loss": 3.9091,
+      "step": 5557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0145660255535772e-05,
+      "loss": 3.9581,
+      "step": 5558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0137324134234577e-05,
+      "loss": 3.9681,
+      "step": 5559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0128990567999666e-05,
+      "loss": 3.8765,
+      "step": 5560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0120659558263687e-05,
+      "loss": 3.9933,
+      "step": 5561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0112331106458826e-05,
+      "loss": 3.9427,
+      "step": 5562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0104005214016837e-05,
+      "loss": 3.9344,
+      "step": 5563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0095681882369043e-05,
+      "loss": 4.0153,
+      "step": 5564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.008736111294632e-05,
+      "loss": 3.9626,
+      "step": 5565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0079042907179093e-05,
+      "loss": 3.8989,
+      "step": 5566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0070727266497379e-05,
+      "loss": 3.8502,
+      "step": 5567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0062414192330723e-05,
+      "loss": 3.894,
+      "step": 5568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.005410368610823e-05,
+      "loss": 4.0106,
+      "step": 5569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.004579574925859e-05,
+      "loss": 3.9183,
+      "step": 5570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0037490383210024e-05,
+      "loss": 4.026,
+      "step": 5571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.002918758939032e-05,
+      "loss": 4.073,
+      "step": 5572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.002088736922683e-05,
+      "loss": 3.895,
+      "step": 5573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0012589724146444e-05,
+      "loss": 3.9958,
+      "step": 5574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.000429465557564e-05,
+      "loss": 3.9486,
+      "step": 5575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.99600216494043e-06,
+      "loss": 4.0591,
+      "step": 5576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.98771225366639e-06,
+      "loss": 3.9509,
+      "step": 5577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.979424923178643e-06,
+      "loss": 3.9319,
+      "step": 5578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.971140174901891e-06,
+      "loss": 3.7769,
+      "step": 5579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.962858010260375e-06,
+      "loss": 3.7882,
+      "step": 5580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.954578430677882e-06,
+      "loss": 3.889,
+      "step": 5581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.94630143757779e-06,
+      "loss": 4.0428,
+      "step": 5582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.938027032382995e-06,
+      "loss": 4.0127,
+      "step": 5583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.929755216515971e-06,
+      "loss": 3.9383,
+      "step": 5584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.92148599139873e-06,
+      "loss": 3.9343,
+      "step": 5585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.913219358452855e-06,
+      "loss": 3.979,
+      "step": 5586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.904955319099462e-06,
+      "loss": 3.7227,
+      "step": 5587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.896693874759258e-06,
+      "loss": 3.8746,
+      "step": 5588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.888435026852458e-06,
+      "loss": 3.9393,
+      "step": 5589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.880178776798877e-06,
+      "loss": 3.9177,
+      "step": 5590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.871925126017845e-06,
+      "loss": 3.8413,
+      "step": 5591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.863674075928266e-06,
+      "loss": 3.7718,
+      "step": 5592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855425627948586e-06,
+      "loss": 4.0254,
+      "step": 5593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.847179783496815e-06,
+      "loss": 3.8371,
+      "step": 5594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.838936543990495e-06,
+      "loss": 4.044,
+      "step": 5595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.830695910846755e-06,
+      "loss": 3.9127,
+      "step": 5596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.822457885482236e-06,
+      "loss": 3.9448,
+      "step": 5597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.814222469313167e-06,
+      "loss": 3.876,
+      "step": 5598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.805989663755308e-06,
+      "loss": 3.8848,
+      "step": 5599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.797759470223967e-06,
+      "loss": 3.9005,
+      "step": 5600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.789531890134012e-06,
+      "loss": 3.9432,
+      "step": 5601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.781306924899853e-06,
+      "loss": 3.9653,
+      "step": 5602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.773084575935471e-06,
+      "loss": 3.8115,
+      "step": 5603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.764864844654379e-06,
+      "loss": 3.9829,
+      "step": 5604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.756647732469636e-06,
+      "loss": 3.7589,
+      "step": 5605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.748433240793859e-06,
+      "loss": 4.0633,
+      "step": 5606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.740221371039226e-06,
+      "loss": 3.9988,
+      "step": 5607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.732012124617448e-06,
+      "loss": 3.9507,
+      "step": 5608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.723805502939787e-06,
+      "loss": 4.0298,
+      "step": 5609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.715601507417047e-06,
+      "loss": 3.9367,
+      "step": 5610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.70740013945961e-06,
+      "loss": 3.7922,
+      "step": 5611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.699201400477374e-06,
+      "loss": 3.8119,
+      "step": 5612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.691005291879801e-06,
+      "loss": 3.9017,
+      "step": 5613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.682811815075895e-06,
+      "loss": 3.8842,
+      "step": 5614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.674620971474202e-06,
+      "loss": 3.8028,
+      "step": 5615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.666432762482839e-06,
+      "loss": 3.9166,
+      "step": 5616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.658247189509436e-06,
+      "loss": 3.7918,
+      "step": 5617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.650064253961209e-06,
+      "loss": 3.9893,
+      "step": 5618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.641883957244888e-06,
+      "loss": 3.9158,
+      "step": 5619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.63370630076676e-06,
+      "loss": 4.01,
+      "step": 5620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.625531285932659e-06,
+      "loss": 4.0884,
+      "step": 5621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.61735891414797e-06,
+      "loss": 3.8634,
+      "step": 5622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.609189186817605e-06,
+      "loss": 3.9406,
+      "step": 5623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.60102210534605e-06,
+      "loss": 3.9464,
+      "step": 5624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.59285767113731e-06,
+      "loss": 4.0131,
+      "step": 5625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.584695885594957e-06,
+      "loss": 3.9589,
+      "step": 5626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.576536750122094e-06,
+      "loss": 3.7878,
+      "step": 5627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.568380266121365e-06,
+      "loss": 3.8734,
+      "step": 5628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.560226434994971e-06,
+      "loss": 3.8872,
+      "step": 5629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.55207525814464e-06,
+      "loss": 3.9551,
+      "step": 5630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.543926736971656e-06,
+      "loss": 3.9148,
+      "step": 5631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.535780872876857e-06,
+      "loss": 4.0325,
+      "step": 5632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.5276376672606e-06,
+      "loss": 3.9248,
+      "step": 5633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.519497121522791e-06,
+      "loss": 3.9403,
+      "step": 5634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.5113592370629e-06,
+      "loss": 4.0105,
+      "step": 5635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.503224015279915e-06,
+      "loss": 3.9308,
+      "step": 5636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.49509145757238e-06,
+      "loss": 3.9332,
+      "step": 5637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.486961565338359e-06,
+      "loss": 3.9387,
+      "step": 5638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.478834339975497e-06,
+      "loss": 3.9278,
+      "step": 5639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.470709782880952e-06,
+      "loss": 3.9415,
+      "step": 5640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.462587895451425e-06,
+      "loss": 3.935,
+      "step": 5641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.454468679083161e-06,
+      "loss": 3.9145,
+      "step": 5642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.446352135171944e-06,
+      "loss": 3.7709,
+      "step": 5643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.438238265113116e-06,
+      "loss": 3.8806,
+      "step": 5644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.430127070301531e-06,
+      "loss": 4.0286,
+      "step": 5645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.422018552131611e-06,
+      "loss": 3.9975,
+      "step": 5646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.413912711997297e-06,
+      "loss": 3.9772,
+      "step": 5647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.405809551292077e-06,
+      "loss": 3.8347,
+      "step": 5648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.39770907140898e-06,
+      "loss": 3.9999,
+      "step": 5649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.38961127374057e-06,
+      "loss": 3.9301,
+      "step": 5650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.381516159678955e-06,
+      "loss": 3.8343,
+      "step": 5651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.373423730615765e-06,
+      "loss": 4.0192,
+      "step": 5652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.3653339879422e-06,
+      "loss": 3.8477,
+      "step": 5653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.357246933048977e-06,
+      "loss": 4.0603,
+      "step": 5654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.349162567326355e-06,
+      "loss": 3.8812,
+      "step": 5655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.341080892164131e-06,
+      "loss": 4.0107,
+      "step": 5656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.333001908951633e-06,
+      "loss": 3.925,
+      "step": 5657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.324925619077732e-06,
+      "loss": 4.158,
+      "step": 5658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.316852023930833e-06,
+      "loss": 3.7207,
+      "step": 5659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.308781124898894e-06,
+      "loss": 3.8652,
+      "step": 5660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.300712923369386e-06,
+      "loss": 3.8695,
+      "step": 5661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.29264742072932e-06,
+      "loss": 3.6915,
+      "step": 5662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.284584618365267e-06,
+      "loss": 4.0129,
+      "step": 5663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.276524517663307e-06,
+      "loss": 3.9686,
+      "step": 5664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.268467120009064e-06,
+      "loss": 3.8776,
+      "step": 5665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.260412426787691e-06,
+      "loss": 3.9382,
+      "step": 5666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.252360439383898e-06,
+      "loss": 3.8809,
+      "step": 5667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.244311159181906e-06,
+      "loss": 3.8983,
+      "step": 5668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.236264587565485e-06,
+      "loss": 3.9995,
+      "step": 5669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.22822072591793e-06,
+      "loss": 3.8431,
+      "step": 5670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.220179575622066e-06,
+      "loss": 4.0773,
+      "step": 5671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.212141138060273e-06,
+      "loss": 4.0205,
+      "step": 5672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.20410541461445e-06,
+      "loss": 4.0284,
+      "step": 5673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.19607240666602e-06,
+      "loss": 3.8692,
+      "step": 5674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.188042115595966e-06,
+      "loss": 3.8733,
+      "step": 5675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.18001454278478e-06,
+      "loss": 3.9796,
+      "step": 5676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.171989689612495e-06,
+      "loss": 3.903,
+      "step": 5677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.163967557458675e-06,
+      "loss": 3.7769,
+      "step": 5678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.15594814770242e-06,
+      "loss": 3.7987,
+      "step": 5679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.147931461722351e-06,
+      "loss": 3.9415,
+      "step": 5680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.139917500896635e-06,
+      "loss": 3.9063,
+      "step": 5681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.131906266602978e-06,
+      "loss": 3.899,
+      "step": 5682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.123897760218588e-06,
+      "loss": 3.9598,
+      "step": 5683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.115891983120228e-06,
+      "loss": 3.9028,
+      "step": 5684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.10788893668418e-06,
+      "loss": 3.9269,
+      "step": 5685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.099888622286263e-06,
+      "loss": 3.919,
+      "step": 5686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.091891041301808e-06,
+      "loss": 3.8928,
+      "step": 5687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.083896195105718e-06,
+      "loss": 3.8455,
+      "step": 5688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.075904085072376e-06,
+      "loss": 3.8756,
+      "step": 5689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.06791471257574e-06,
+      "loss": 3.8993,
+      "step": 5690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.059928078989266e-06,
+      "loss": 4.054,
+      "step": 5691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.051944185685949e-06,
+      "loss": 3.8574,
+      "step": 5692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.043963034038311e-06,
+      "loss": 3.8805,
+      "step": 5693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.035984625418405e-06,
+      "loss": 3.9922,
+      "step": 5694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.028008961197804e-06,
+      "loss": 3.9291,
+      "step": 5695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.020036042747637e-06,
+      "loss": 3.9319,
+      "step": 5696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.012065871438527e-06,
+      "loss": 3.9805,
+      "step": 5697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.004098448640643e-06,
+      "loss": 3.8279,
+      "step": 5698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.996133775723666e-06,
+      "loss": 3.9603,
+      "step": 5699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.988171854056838e-06,
+      "loss": 3.8912,
+      "step": 5700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.98021268500889e-06,
+      "loss": 3.8737,
+      "step": 5701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.97225626994809e-06,
+      "loss": 3.8757,
+      "step": 5702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.964302610242256e-06,
+      "loss": 3.8547,
+      "step": 5703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.956351707258706e-06,
+      "loss": 3.8095,
+      "step": 5704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.948403562364292e-06,
+      "loss": 3.9729,
+      "step": 5705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.94045817692539e-06,
+      "loss": 4.0081,
+      "step": 5706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.932515552307905e-06,
+      "loss": 3.9085,
+      "step": 5707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.92457568987726e-06,
+      "loss": 3.7969,
+      "step": 5708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.916638590998424e-06,
+      "loss": 3.9816,
+      "step": 5709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.90870425703586e-06,
+      "loss": 3.9288,
+      "step": 5710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.90077268935359e-06,
+      "loss": 3.9211,
+      "step": 5711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.892843889315133e-06,
+      "loss": 3.9309,
+      "step": 5712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.884917858283539e-06,
+      "loss": 4.0518,
+      "step": 5713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.87699459762139e-06,
+      "loss": 3.9159,
+      "step": 5714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.869074108690783e-06,
+      "loss": 3.8305,
+      "step": 5715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.861156392853334e-06,
+      "loss": 3.9506,
+      "step": 5716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.853241451470198e-06,
+      "loss": 3.8841,
+      "step": 5717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.845329285902054e-06,
+      "loss": 3.8512,
+      "step": 5718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.837419897509085e-06,
+      "loss": 3.9729,
+      "step": 5719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.829513287651011e-06,
+      "loss": 3.8981,
+      "step": 5720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.821609457687069e-06,
+      "loss": 3.8889,
+      "step": 5721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.813708408976015e-06,
+      "loss": 4.1067,
+      "step": 5722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.805810142876122e-06,
+      "loss": 3.8396,
+      "step": 5723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.797914660745219e-06,
+      "loss": 3.8838,
+      "step": 5724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.790021963940612e-06,
+      "loss": 3.9524,
+      "step": 5725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.782132053819145e-06,
+      "loss": 3.9727,
+      "step": 5726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.774244931737196e-06,
+      "loss": 3.8777,
+      "step": 5727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.766360599050655e-06,
+      "loss": 4.097,
+      "step": 5728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.758479057114918e-06,
+      "loss": 3.8573,
+      "step": 5729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.750600307284922e-06,
+      "loss": 4.0445,
+      "step": 5730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.742724350915102e-06,
+      "loss": 4.0714,
+      "step": 5731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.734851189359448e-06,
+      "loss": 3.9334,
+      "step": 5732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.726980823971433e-06,
+      "loss": 3.9108,
+      "step": 5733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.719113256104069e-06,
+      "loss": 3.9096,
+      "step": 5734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.71124848710988e-06,
+      "loss": 3.913,
+      "step": 5735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.703386518340903e-06,
+      "loss": 4.0265,
+      "step": 5736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.695527351148718e-06,
+      "loss": 3.9746,
+      "step": 5737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.687670986884395e-06,
+      "loss": 3.9121,
+      "step": 5738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.67981742689854e-06,
+      "loss": 3.9449,
+      "step": 5739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.671966672541273e-06,
+      "loss": 4.0213,
+      "step": 5740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.664118725162226e-06,
+      "loss": 3.9287,
+      "step": 5741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.65627358611055e-06,
+      "loss": 4.1506,
+      "step": 5742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.648431256734919e-06,
+      "loss": 3.8459,
+      "step": 5743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.640591738383508e-06,
+      "loss": 3.9964,
+      "step": 5744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.632755032404042e-06,
+      "loss": 3.9632,
+      "step": 5745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.62492114014372e-06,
+      "loss": 3.7625,
+      "step": 5746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.617090062949303e-06,
+      "loss": 3.9305,
+      "step": 5747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.609261802167029e-06,
+      "loss": 3.9957,
+      "step": 5748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.60143635914267e-06,
+      "loss": 3.8984,
+      "step": 5749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.593613735221506e-06,
+      "loss": 4.002,
+      "step": 5750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.585793931748342e-06,
+      "loss": 3.8609,
+      "step": 5751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.577976950067484e-06,
+      "loss": 3.9816,
+      "step": 5752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.570162791522776e-06,
+      "loss": 4.097,
+      "step": 5753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.562351457457549e-06,
+      "loss": 3.9534,
+      "step": 5754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.554542949214673e-06,
+      "loss": 3.8881,
+      "step": 5755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.546737268136518e-06,
+      "loss": 4.0254,
+      "step": 5756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.538934415564967e-06,
+      "loss": 3.9177,
+      "step": 5757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.531134392841425e-06,
+      "loss": 4.0912,
+      "step": 5758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.523337201306797e-06,
+      "loss": 3.8973,
+      "step": 5759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.515542842301524e-06,
+      "loss": 3.7266,
+      "step": 5760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.50775131716554e-06,
+      "loss": 3.9852,
+      "step": 5761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.499962627238303e-06,
+      "loss": 3.8885,
+      "step": 5762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.492176773858764e-06,
+      "loss": 3.9062,
+      "step": 5763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.484393758365422e-06,
+      "loss": 3.9734,
+      "step": 5764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.476613582096257e-06,
+      "loss": 3.969,
+      "step": 5765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.468836246388773e-06,
+      "loss": 3.9898,
+      "step": 5766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.461061752579977e-06,
+      "loss": 3.8919,
+      "step": 5767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.45329010200641e-06,
+      "loss": 3.8058,
+      "step": 5768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.4455212960041e-06,
+      "loss": 3.8251,
+      "step": 5769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.437755335908592e-06,
+      "loss": 3.9403,
+      "step": 5770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.429992223054953e-06,
+      "loss": 4.0043,
+      "step": 5771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.422231958777744e-06,
+      "loss": 4.013,
+      "step": 5772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.414474544411038e-06,
+      "loss": 3.9698,
+      "step": 5773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.406719981288436e-06,
+      "loss": 3.9793,
+      "step": 5774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.398968270743041e-06,
+      "loss": 3.9319,
+      "step": 5775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.391219414107456e-06,
+      "loss": 3.8954,
+      "step": 5776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.383473412713801e-06,
+      "loss": 3.9524,
+      "step": 5777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.375730267893703e-06,
+      "loss": 3.7958,
+      "step": 5778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.367989980978294e-06,
+      "loss": 3.9961,
+      "step": 5779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.360252553298215e-06,
+      "loss": 3.841,
+      "step": 5780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.352517986183637e-06,
+      "loss": 3.9963,
+      "step": 5781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.344786280964198e-06,
+      "loss": 3.8791,
+      "step": 5782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.337057438969093e-06,
+      "loss": 3.8629,
+      "step": 5783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.329331461526987e-06,
+      "loss": 3.9758,
+      "step": 5784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.321608349966065e-06,
+      "loss": 3.9307,
+      "step": 5785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.31388810561402e-06,
+      "loss": 3.8871,
+      "step": 5786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.306170729798054e-06,
+      "loss": 3.8844,
+      "step": 5787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.298456223844862e-06,
+      "loss": 3.8698,
+      "step": 5788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.290744589080673e-06,
+      "loss": 3.9942,
+      "step": 5789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.2830358268312e-06,
+      "loss": 3.8833,
+      "step": 5790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.275329938421657e-06,
+      "loss": 3.9957,
+      "step": 5791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.2676269251768e-06,
+      "loss": 3.8467,
+      "step": 5792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.259926788420849e-06,
+      "loss": 3.8559,
+      "step": 5793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.252229529477554e-06,
+      "loss": 3.8062,
+      "step": 5794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.244535149670149e-06,
+      "loss": 3.8081,
+      "step": 5795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.236843650321407e-06,
+      "loss": 3.9806,
+      "step": 5796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.229155032753578e-06,
+      "loss": 3.847,
+      "step": 5797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.22146929828842e-06,
+      "loss": 4.05,
+      "step": 5798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.213786448247204e-06,
+      "loss": 3.9953,
+      "step": 5799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.206106483950693e-06,
+      "loss": 3.9805,
+      "step": 5800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.198429406719177e-06,
+      "loss": 3.9092,
+      "step": 5801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.190755217872425e-06,
+      "loss": 4.1061,
+      "step": 5802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.183083918729714e-06,
+      "loss": 4.0466,
+      "step": 5803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.175415510609844e-06,
+      "loss": 3.8979,
+      "step": 5804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.167749994831091e-06,
+      "loss": 3.8299,
+      "step": 5805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.160087372711256e-06,
+      "loss": 3.8789,
+      "step": 5806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.152427645567623e-06,
+      "loss": 3.9136,
+      "step": 5807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.144770814716992e-06,
+      "loss": 3.8328,
+      "step": 5808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.137116881475654e-06,
+      "loss": 3.9696,
+      "step": 5809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.129465847159414e-06,
+      "loss": 3.9153,
+      "step": 5810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.121817713083584e-06,
+      "loss": 3.9509,
+      "step": 5811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.114172480562956e-06,
+      "loss": 3.9288,
+      "step": 5812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.106530150911837e-06,
+      "loss": 3.8116,
+      "step": 5813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.098890725444033e-06,
+      "loss": 3.8382,
+      "step": 5814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.091254205472845e-06,
+      "loss": 3.7763,
+      "step": 5815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.083620592311075e-06,
+      "loss": 3.9671,
+      "step": 5816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.075989887271043e-06,
+      "loss": 3.9024,
+      "step": 5817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.068362091664552e-06,
+      "loss": 3.9532,
+      "step": 5818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.060737206802896e-06,
+      "loss": 3.8577,
+      "step": 5819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.053115233996899e-06,
+      "loss": 4.0167,
+      "step": 5820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.04549617455686e-06,
+      "loss": 3.8193,
+      "step": 5821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.037880029792583e-06,
+      "loss": 3.9689,
+      "step": 5822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.030266801013366e-06,
+      "loss": 3.9694,
+      "step": 5823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.022656489528013e-06,
+      "loss": 3.9683,
+      "step": 5824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.015049096644834e-06,
+      "loss": 3.9497,
+      "step": 5825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.00744462367162e-06,
+      "loss": 3.9259,
+      "step": 5826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.999843071915672e-06,
+      "loss": 3.9216,
+      "step": 5827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.992244442683772e-06,
+      "loss": 3.9209,
+      "step": 5828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.984648737282232e-06,
+      "loss": 3.995,
+      "step": 5829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.977055957016835e-06,
+      "loss": 3.9491,
+      "step": 5830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.969466103192858e-06,
+      "loss": 3.8575,
+      "step": 5831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.961879177115097e-06,
+      "loss": 3.9144,
+      "step": 5832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.954295180087831e-06,
+      "loss": 3.8368,
+      "step": 5833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.946714113414835e-06,
+      "loss": 3.9072,
+      "step": 5834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.939135978399381e-06,
+      "loss": 3.9792,
+      "step": 5835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.931560776344237e-06,
+      "loss": 3.9311,
+      "step": 5836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.923988508551664e-06,
+      "loss": 4.0661,
+      "step": 5837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.916419176323428e-06,
+      "loss": 3.9619,
+      "step": 5838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.908852780960793e-06,
+      "loss": 3.9241,
+      "step": 5839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.901289323764502e-06,
+      "loss": 3.9116,
+      "step": 5840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.893728806034803e-06,
+      "loss": 3.7319,
+      "step": 5841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.886171229071434e-06,
+      "loss": 3.9393,
+      "step": 5842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.878616594173633e-06,
+      "loss": 3.8983,
+      "step": 5843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.871064902640125e-06,
+      "loss": 3.9159,
+      "step": 5844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.86351615576913e-06,
+      "loss": 3.9213,
+      "step": 5845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.855970354858377e-06,
+      "loss": 3.9176,
+      "step": 5846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.848427501205063e-06,
+      "loss": 3.8465,
+      "step": 5847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.840887596105909e-06,
+      "loss": 3.8453,
+      "step": 5848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.833350640857102e-06,
+      "loss": 3.9207,
+      "step": 5849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.825816636754332e-06,
+      "loss": 4.058,
+      "step": 5850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.818285585092782e-06,
+      "loss": 3.8075,
+      "step": 5851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.810757487167122e-06,
+      "loss": 3.9289,
+      "step": 5852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.803232344271533e-06,
+      "loss": 3.9144,
+      "step": 5853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.795710157699662e-06,
+      "loss": 3.8635,
+      "step": 5854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.788190928744668e-06,
+      "loss": 3.9848,
+      "step": 5855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.78067465869918e-06,
+      "loss": 4.0597,
+      "step": 5856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.773161348855348e-06,
+      "loss": 4.041,
+      "step": 5857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.765651000504795e-06,
+      "loss": 3.7939,
+      "step": 5858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.75814361493862e-06,
+      "loss": 3.9709,
+      "step": 5859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.750639193447453e-06,
+      "loss": 4.0556,
+      "step": 5860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.743137737321383e-06,
+      "loss": 3.8232,
+      "step": 5861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.73563924784999e-06,
+      "loss": 3.9514,
+      "step": 5862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.728143726322359e-06,
+      "loss": 3.9822,
+      "step": 5863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.720651174027052e-06,
+      "loss": 3.8631,
+      "step": 5864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.713161592252122e-06,
+      "loss": 3.9117,
+      "step": 5865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.705674982285127e-06,
+      "loss": 3.8836,
+      "step": 5866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.698191345413086e-06,
+      "loss": 3.8938,
+      "step": 5867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.690710682922541e-06,
+      "loss": 3.9985,
+      "step": 5868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.683232996099498e-06,
+      "loss": 3.8031,
+      "step": 5869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.675758286229457e-06,
+      "loss": 3.9262,
+      "step": 5870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.668286554597403e-06,
+      "loss": 3.921,
+      "step": 5871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.66081780248782e-06,
+      "loss": 3.9918,
+      "step": 5872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.65335203118466e-06,
+      "loss": 3.9945,
+      "step": 5873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.645889241971384e-06,
+      "loss": 3.8261,
+      "step": 5874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.638429436130945e-06,
+      "loss": 3.8941,
+      "step": 5875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.630972614945756e-06,
+      "loss": 3.9387,
+      "step": 5876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.623518779697733e-06,
+      "loss": 4.0205,
+      "step": 5877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.616067931668277e-06,
+      "loss": 3.8918,
+      "step": 5878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.608620072138278e-06,
+      "loss": 3.9953,
+      "step": 5879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.601175202388098e-06,
+      "loss": 3.9988,
+      "step": 5880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5937333236976135e-06,
+      "loss": 3.9463,
+      "step": 5881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.586294437346158e-06,
+      "loss": 3.8908,
+      "step": 5882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.578858544612572e-06,
+      "loss": 3.8436,
+      "step": 5883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.571425646775151e-06,
+      "loss": 3.9146,
+      "step": 5884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.563995745111724e-06,
+      "loss": 4.0075,
+      "step": 5885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.556568840899561e-06,
+      "loss": 3.8827,
+      "step": 5886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5491449354154336e-06,
+      "loss": 4.0059,
+      "step": 5887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.541724029935595e-06,
+      "loss": 3.8823,
+      "step": 5888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5343061257357964e-06,
+      "loss": 3.8912,
+      "step": 5889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.526891224091254e-06,
+      "loss": 3.9144,
+      "step": 5890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.519479326276677e-06,
+      "loss": 4.0533,
+      "step": 5891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.512070433566254e-06,
+      "loss": 3.9798,
+      "step": 5892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5046645472336555e-06,
+      "loss": 3.8609,
+      "step": 5893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.49726166855205e-06,
+      "loss": 3.8349,
+      "step": 5894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.489861798794065e-06,
+      "loss": 3.9532,
+      "step": 5895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.482464939231842e-06,
+      "loss": 3.9825,
+      "step": 5896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.475071091136973e-06,
+      "loss": 3.8662,
+      "step": 5897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.467680255780554e-06,
+      "loss": 3.762,
+      "step": 5898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.460292434433147e-06,
+      "loss": 3.878,
+      "step": 5899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.452907628364811e-06,
+      "loss": 3.9986,
+      "step": 5900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.445525838845077e-06,
+      "loss": 3.847,
+      "step": 5901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.43814706714295e-06,
+      "loss": 3.7526,
+      "step": 5902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.430771314526941e-06,
+      "loss": 3.9903,
+      "step": 5903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.423398582265026e-06,
+      "loss": 3.9268,
+      "step": 5904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.416028871624664e-06,
+      "loss": 3.8674,
+      "step": 5905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.408662183872786e-06,
+      "loss": 4.0106,
+      "step": 5906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.401298520275818e-06,
+      "loss": 4.0133,
+      "step": 5907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.393937882099656e-06,
+      "loss": 3.9362,
+      "step": 5908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.386580270609669e-06,
+      "loss": 3.8405,
+      "step": 5909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.379225687070734e-06,
+      "loss": 4.0144,
+      "step": 5910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.371874132747175e-06,
+      "loss": 3.9337,
+      "step": 5911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.364525608902823e-06,
+      "loss": 3.832,
+      "step": 5912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.357180116800966e-06,
+      "loss": 3.8603,
+      "step": 5913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.349837657704378e-06,
+      "loss": 3.9267,
+      "step": 5914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.3424982328753196e-06,
+      "loss": 4.0268,
+      "step": 5915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.335161843575506e-06,
+      "loss": 3.94,
+      "step": 5916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.327828491066169e-06,
+      "loss": 4.0133,
+      "step": 5917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.32049817660799e-06,
+      "loss": 3.8021,
+      "step": 5918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.313170901461131e-06,
+      "loss": 3.853,
+      "step": 5919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.305846666885235e-06,
+      "loss": 3.9475,
+      "step": 5920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.29852547413942e-06,
+      "loss": 3.9903,
+      "step": 5921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.291207324482297e-06,
+      "loss": 3.8828,
+      "step": 5922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.283892219171934e-06,
+      "loss": 3.8992,
+      "step": 5923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.27658015946587e-06,
+      "loss": 3.887,
+      "step": 5924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.269271146621154e-06,
+      "loss": 3.9188,
+      "step": 5925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.261965181894281e-06,
+      "loss": 4.0118,
+      "step": 5926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.254662266541229e-06,
+      "loss": 3.9021,
+      "step": 5927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.247362401817456e-06,
+      "loss": 3.981,
+      "step": 5928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.24006558897789e-06,
+      "loss": 3.9305,
+      "step": 5929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.2327718292769355e-06,
+      "loss": 3.9636,
+      "step": 5930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.225481123968475e-06,
+      "loss": 3.9885,
+      "step": 5931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.218193474305882e-06,
+      "loss": 3.9346,
+      "step": 5932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.210908881541969e-06,
+      "loss": 3.9252,
+      "step": 5933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.2036273469290535e-06,
+      "loss": 3.9453,
+      "step": 5934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.196348871718908e-06,
+      "loss": 3.9311,
+      "step": 5935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.18907345716279e-06,
+      "loss": 3.7442,
+      "step": 5936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.181801104511418e-06,
+      "loss": 4.0123,
+      "step": 5937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.174531815015009e-06,
+      "loss": 3.8929,
+      "step": 5938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.167265589923227e-06,
+      "loss": 3.9086,
+      "step": 5939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.160002430485229e-06,
+      "loss": 3.9232,
+      "step": 5940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.1527423379496365e-06,
+      "loss": 3.8913,
+      "step": 5941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.145485313564537e-06,
+      "loss": 3.836,
+      "step": 5942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.138231358577499e-06,
+      "loss": 3.7891,
+      "step": 5943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.13098047423556e-06,
+      "loss": 3.8925,
+      "step": 5944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.123732661785229e-06,
+      "loss": 3.9138,
+      "step": 5945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.116487922472498e-06,
+      "loss": 3.921,
+      "step": 5946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.109246257542818e-06,
+      "loss": 4.0055,
+      "step": 5947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.1020076682411065e-06,
+      "loss": 3.9809,
+      "step": 5948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.0947721558117745e-06,
+      "loss": 3.9764,
+      "step": 5949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.087539721498685e-06,
+      "loss": 3.8655,
+      "step": 5950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.080310366545176e-06,
+      "loss": 3.9259,
+      "step": 5951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.073084092194049e-06,
+      "loss": 3.9678,
+      "step": 5952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.0658608996876045e-06,
+      "loss": 3.9932,
+      "step": 5953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.058640790267579e-06,
+      "loss": 3.8973,
+      "step": 5954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.051423765175194e-06,
+      "loss": 3.8456,
+      "step": 5955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.044209825651144e-06,
+      "loss": 4.002,
+      "step": 5956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.036998972935585e-06,
+      "loss": 3.9844,
+      "step": 5957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.029791208268141e-06,
+      "loss": 4.0588,
+      "step": 5958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.022586532887926e-06,
+      "loss": 4.1127,
+      "step": 5959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.015384948033487e-06,
+      "loss": 3.7729,
+      "step": 5960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.00818645494288e-06,
+      "loss": 3.9224,
+      "step": 5961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.0009910548536e-06,
+      "loss": 3.7051,
+      "step": 5962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.993798749002623e-06,
+      "loss": 4.0237,
+      "step": 5963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.986609538626385e-06,
+      "loss": 3.8403,
+      "step": 5964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.979423424960796e-06,
+      "loss": 3.8112,
+      "step": 5965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.972240409241223e-06,
+      "loss": 3.973,
+      "step": 5966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.965060492702524e-06,
+      "loss": 3.9176,
+      "step": 5967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.957883676579011e-06,
+      "loss": 3.9873,
+      "step": 5968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.950709962104454e-06,
+      "loss": 3.923,
+      "step": 5969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.943539350512101e-06,
+      "loss": 3.9089,
+      "step": 5970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.936371843034664e-06,
+      "loss": 3.9452,
+      "step": 5971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.929207440904317e-06,
+      "loss": 3.9274,
+      "step": 5972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.922046145352698e-06,
+      "loss": 3.8369,
+      "step": 5973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.91488795761093e-06,
+      "loss": 3.9397,
+      "step": 5974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.907732878909587e-06,
+      "loss": 3.994,
+      "step": 5975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.900580910478693e-06,
+      "loss": 3.9286,
+      "step": 5976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.89343205354778e-06,
+      "loss": 3.8345,
+      "step": 5977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.886286309345802e-06,
+      "loss": 3.9201,
+      "step": 5978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.879143679101202e-06,
+      "loss": 4.009,
+      "step": 5979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.872004164041879e-06,
+      "loss": 3.8282,
+      "step": 5980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.864867765395189e-06,
+      "loss": 3.9725,
+      "step": 5981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.8577344843879756e-06,
+      "loss": 3.9233,
+      "step": 5982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.850604322246532e-06,
+      "loss": 3.8486,
+      "step": 5983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.843477280196609e-06,
+      "loss": 3.8317,
+      "step": 5984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.836353359463424e-06,
+      "loss": 3.9739,
+      "step": 5985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.829232561271673e-06,
+      "loss": 3.9833,
+      "step": 5986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.822114886845498e-06,
+      "loss": 3.8174,
+      "step": 5987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.815000337408506e-06,
+      "loss": 3.9381,
+      "step": 5988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.80788891418378e-06,
+      "loss": 4.1042,
+      "step": 5989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.800780618393851e-06,
+      "loss": 3.9479,
+      "step": 5990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.7936754512607174e-06,
+      "loss": 3.8721,
+      "step": 5991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.786573414005839e-06,
+      "loss": 3.9536,
+      "step": 5992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.779474507850139e-06,
+      "loss": 3.9525,
+      "step": 5993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.772378734013993e-06,
+      "loss": 3.8525,
+      "step": 5994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.7652860937172655e-06,
+      "loss": 4.0255,
+      "step": 5995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.758196588179244e-06,
+      "loss": 3.9144,
+      "step": 5996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.751110218618714e-06,
+      "loss": 3.8215,
+      "step": 5997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.744026986253896e-06,
+      "loss": 3.8564,
+      "step": 5998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.7369468923024815e-06,
+      "loss": 3.8406,
+      "step": 5999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.72986993798162e-06,
+      "loss": 3.9896,
+      "step": 6000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.722796124507921e-06,
+      "loss": 3.8186,
+      "step": 6001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.715725453097446e-06,
+      "loss": 3.8918,
+      "step": 6002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.708657924965747e-06,
+      "loss": 3.8876,
+      "step": 6003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.701593541327791e-06,
+      "loss": 3.9299,
+      "step": 6004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.694532303398049e-06,
+      "loss": 3.9893,
+      "step": 6005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.687474212390419e-06,
+      "loss": 3.8664,
+      "step": 6006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.680419269518265e-06,
+      "loss": 3.9893,
+      "step": 6007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.673367475994421e-06,
+      "loss": 3.8436,
+      "step": 6008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.66631883303116e-06,
+      "loss": 3.9937,
+      "step": 6009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.659273341840241e-06,
+      "loss": 3.9006,
+      "step": 6010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.6522310036328585e-06,
+      "loss": 3.9709,
+      "step": 6011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.645191819619673e-06,
+      "loss": 3.8366,
+      "step": 6012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.638155791010792e-06,
+      "loss": 3.8362,
+      "step": 6013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.631122919015809e-06,
+      "loss": 3.892,
+      "step": 6014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.624093204843746e-06,
+      "loss": 3.885,
+      "step": 6015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.617066649703091e-06,
+      "loss": 3.9702,
+      "step": 6016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.610043254801784e-06,
+      "loss": 3.8987,
+      "step": 6017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.603023021347246e-06,
+      "loss": 3.8017,
+      "step": 6018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.596005950546328e-06,
+      "loss": 3.8979,
+      "step": 6019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.5889920436053405e-06,
+      "loss": 3.8009,
+      "step": 6020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.581981301730062e-06,
+      "loss": 3.9223,
+      "step": 6021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.57497372612571e-06,
+      "loss": 3.9068,
+      "step": 6022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.567969317996983e-06,
+      "loss": 3.8339,
+      "step": 6023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.560968078548005e-06,
+      "loss": 4.0098,
+      "step": 6024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.553970008982385e-06,
+      "loss": 3.8926,
+      "step": 6025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.5469751105031646e-06,
+      "loss": 3.8462,
+      "step": 6026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.539983384312851e-06,
+      "loss": 3.9723,
+      "step": 6027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.532994831613398e-06,
+      "loss": 4.0213,
+      "step": 6028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.526009453606224e-06,
+      "loss": 3.8847,
+      "step": 6029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.519027251492185e-06,
+      "loss": 3.9685,
+      "step": 6030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.512048226471618e-06,
+      "loss": 3.8903,
+      "step": 6031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.505072379744284e-06,
+      "loss": 3.9103,
+      "step": 6032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.498099712509429e-06,
+      "loss": 4.0047,
+      "step": 6033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.491130225965722e-06,
+      "loss": 3.8871,
+      "step": 6034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.484163921311306e-06,
+      "loss": 3.8179,
+      "step": 6035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.477200799743766e-06,
+      "loss": 3.9047,
+      "step": 6036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.470240862460142e-06,
+      "loss": 3.8272,
+      "step": 6037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4632841106569196e-06,
+      "loss": 3.8658,
+      "step": 6038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.456330545530065e-06,
+      "loss": 4.0536,
+      "step": 6039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.449380168274965e-06,
+      "loss": 3.8174,
+      "step": 6040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.442432980086466e-06,
+      "loss": 3.9326,
+      "step": 6041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.43548898215888e-06,
+      "loss": 3.8083,
+      "step": 6042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.428548175685958e-06,
+      "loss": 4.0016,
+      "step": 6043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.421610561860902e-06,
+      "loss": 3.8839,
+      "step": 6044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4146761418763635e-06,
+      "loss": 3.9435,
+      "step": 6045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.407744916924463e-06,
+      "loss": 3.9428,
+      "step": 6046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4008168881967506e-06,
+      "loss": 3.7757,
+      "step": 6047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3938920568842345e-06,
+      "loss": 3.9116,
+      "step": 6048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.386970424177377e-06,
+      "loss": 3.9593,
+      "step": 6049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.380051991266075e-06,
+      "loss": 4.0028,
+      "step": 6050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.373136759339704e-06,
+      "loss": 3.9684,
+      "step": 6051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.366224729587067e-06,
+      "loss": 3.7177,
+      "step": 6052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.359315903196411e-06,
+      "loss": 3.9443,
+      "step": 6053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3524102813554615e-06,
+      "loss": 3.9051,
+      "step": 6054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.345507865251365e-06,
+      "loss": 3.8806,
+      "step": 6055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.338608656070727e-06,
+      "loss": 3.7598,
+      "step": 6056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.3317126549996025e-06,
+      "loss": 3.9424,
+      "step": 6057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.324819863223497e-06,
+      "loss": 3.9197,
+      "step": 6058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.317930281927348e-06,
+      "loss": 3.9955,
+      "step": 6059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.311043912295564e-06,
+      "loss": 3.8805,
+      "step": 6060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.304160755512003e-06,
+      "loss": 3.9723,
+      "step": 6061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.297280812759945e-06,
+      "loss": 3.9199,
+      "step": 6062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.290404085222141e-06,
+      "loss": 3.961,
+      "step": 6063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.283530574080771e-06,
+      "loss": 3.9441,
+      "step": 6064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.2766602805174765e-06,
+      "loss": 3.8603,
+      "step": 6065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.269793205713332e-06,
+      "loss": 3.9038,
+      "step": 6066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.262929350848881e-06,
+      "loss": 3.8017,
+      "step": 6067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.256068717104094e-06,
+      "loss": 3.9592,
+      "step": 6068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.249211305658384e-06,
+      "loss": 3.7785,
+      "step": 6069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.242357117690639e-06,
+      "loss": 3.8659,
+      "step": 6070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.2355061543791615e-06,
+      "loss": 3.85,
+      "step": 6071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.2286584169017116e-06,
+      "loss": 3.7926,
+      "step": 6072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.221813906435495e-06,
+      "loss": 3.8575,
+      "step": 6073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.214972624157159e-06,
+      "loss": 3.9856,
+      "step": 6074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.208134571242807e-06,
+      "loss": 3.7545,
+      "step": 6075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.201299748867981e-06,
+      "loss": 3.7863,
+      "step": 6076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.194468158207658e-06,
+      "loss": 3.8842,
+      "step": 6077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.187639800436265e-06,
+      "loss": 3.953,
+      "step": 6078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.180814676727689e-06,
+      "loss": 3.8585,
+      "step": 6079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.173992788255239e-06,
+      "loss": 4.0433,
+      "step": 6080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.167174136191675e-06,
+      "loss": 3.9111,
+      "step": 6081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.160358721709211e-06,
+      "loss": 3.8572,
+      "step": 6082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.153546545979491e-06,
+      "loss": 3.9819,
+      "step": 6083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.146737610173606e-06,
+      "loss": 3.8633,
+      "step": 6084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.139931915462091e-06,
+      "loss": 3.8053,
+      "step": 6085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.133129463014925e-06,
+      "loss": 3.9602,
+      "step": 6086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.126330254001522e-06,
+      "loss": 3.9477,
+      "step": 6087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.119534289590748e-06,
+      "loss": 3.8058,
+      "step": 6088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.11274157095092e-06,
+      "loss": 3.892,
+      "step": 6089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.105952099249776e-06,
+      "loss": 3.9238,
+      "step": 6090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.099165875654503e-06,
+      "loss": 3.8961,
+      "step": 6091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.092382901331734e-06,
+      "loss": 3.807,
+      "step": 6092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.085603177447541e-06,
+      "loss": 4.0316,
+      "step": 6093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.078826705167437e-06,
+      "loss": 3.8146,
+      "step": 6094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.0720534856563646e-06,
+      "loss": 3.9702,
+      "step": 6095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.065283520078732e-06,
+      "loss": 3.8256,
+      "step": 6096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.0585168095983805e-06,
+      "loss": 3.9927,
+      "step": 6097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.0517533553785785e-06,
+      "loss": 3.9086,
+      "step": 6098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.044993158582038e-06,
+      "loss": 3.9429,
+      "step": 6099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.038236220370921e-06,
+      "loss": 3.9371,
+      "step": 6100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.0314825419068215e-06,
+      "loss": 4.0385,
+      "step": 6101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.024732124350768e-06,
+      "loss": 3.8816,
+      "step": 6102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.01798496886325e-06,
+      "loss": 3.925,
+      "step": 6103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.011241076604174e-06,
+      "loss": 3.9178,
+      "step": 6104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.004500448732892e-06,
+      "loss": 3.9582,
+      "step": 6105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9977630864081925e-06,
+      "loss": 3.8858,
+      "step": 6106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.991028990788317e-06,
+      "loss": 3.9113,
+      "step": 6107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.98429816303093e-06,
+      "loss": 3.8431,
+      "step": 6108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.977570604293128e-06,
+      "loss": 3.9569,
+      "step": 6109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9708463157314765e-06,
+      "loss": 4.0089,
+      "step": 6110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.964125298501946e-06,
+      "loss": 3.8417,
+      "step": 6111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.957407553759961e-06,
+      "loss": 3.8427,
+      "step": 6112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.950693082660377e-06,
+      "loss": 4.0025,
+      "step": 6113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.943981886357491e-06,
+      "loss": 3.7532,
+      "step": 6114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.937273966005028e-06,
+      "loss": 3.9665,
+      "step": 6115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.930569322756172e-06,
+      "loss": 3.8448,
+      "step": 6116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.923867957763513e-06,
+      "loss": 3.9463,
+      "step": 6117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.91716987217911e-06,
+      "loss": 3.9285,
+      "step": 6118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.910475067154431e-06,
+      "loss": 3.9171,
+      "step": 6119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.903783543840394e-06,
+      "loss": 3.8305,
+      "step": 6120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8970953033873484e-06,
+      "loss": 3.8518,
+      "step": 6121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.89041034694508e-06,
+      "loss": 3.8891,
+      "step": 6122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.883728675662803e-06,
+      "loss": 3.7407,
+      "step": 6123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8770502906891815e-06,
+      "loss": 3.8592,
+      "step": 6124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.870375193172314e-06,
+      "loss": 3.9495,
+      "step": 6125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.863703384259722e-06,
+      "loss": 3.9362,
+      "step": 6126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.857034865098366e-06,
+      "loss": 3.8308,
+      "step": 6127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.850369636834638e-06,
+      "loss": 3.8227,
+      "step": 6128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.84370770061437e-06,
+      "loss": 3.8358,
+      "step": 6129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.837049057582822e-06,
+      "loss": 3.9024,
+      "step": 6130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8303937088847045e-06,
+      "loss": 3.8318,
+      "step": 6131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8237416556641414e-06,
+      "loss": 4.0171,
+      "step": 6132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.817092899064691e-06,
+      "loss": 4.0213,
+      "step": 6133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.810447440229366e-06,
+      "loss": 3.8816,
+      "step": 6134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.803805280300589e-06,
+      "loss": 3.894,
+      "step": 6135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.797166420420227e-06,
+      "loss": 3.8888,
+      "step": 6136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.7905308617295765e-06,
+      "loss": 3.9137,
+      "step": 6137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.783898605369356e-06,
+      "loss": 3.8943,
+      "step": 6138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.777269652479747e-06,
+      "loss": 3.8847,
+      "step": 6139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.770644004200332e-06,
+      "loss": 3.8727,
+      "step": 6140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.764021661670141e-06,
+      "loss": 3.8266,
+      "step": 6141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.757402626027624e-06,
+      "loss": 3.9178,
+      "step": 6142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.7507868984106735e-06,
+      "loss": 3.8507,
+      "step": 6143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.7441744799566156e-06,
+      "loss": 4.0327,
+      "step": 6144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.737565371802189e-06,
+      "loss": 3.849,
+      "step": 6145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.730959575083594e-06,
+      "loss": 3.8633,
+      "step": 6146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.724357090936433e-06,
+      "loss": 3.8058,
+      "step": 6147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.71775792049575e-06,
+      "loss": 3.8442,
+      "step": 6148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.7111620648960195e-06,
+      "loss": 3.8674,
+      "step": 6149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.704569525271147e-06,
+      "loss": 3.7514,
+      "step": 6150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.6979803027544576e-06,
+      "loss": 3.9394,
+      "step": 6151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.691394398478727e-06,
+      "loss": 3.9291,
+      "step": 6152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.684811813576138e-06,
+      "loss": 3.9679,
+      "step": 6153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.678232549178328e-06,
+      "loss": 4.0012,
+      "step": 6154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.671656606416337e-06,
+      "loss": 3.881,
+      "step": 6155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.66508398642065e-06,
+      "loss": 3.9487,
+      "step": 6156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.658514690321176e-06,
+      "loss": 3.9435,
+      "step": 6157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.651948719247252e-06,
+      "loss": 3.739,
+      "step": 6158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.6453860743276386e-06,
+      "loss": 3.9875,
+      "step": 6159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.63882675669054e-06,
+      "loss": 3.8659,
+      "step": 6160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.632270767463574e-06,
+      "loss": 3.9816,
+      "step": 6161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.6257181077737975e-06,
+      "loss": 3.8869,
+      "step": 6162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.619168778747686e-06,
+      "loss": 3.9551,
+      "step": 6163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.612622781511143e-06,
+      "loss": 3.9978,
+      "step": 6164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.606080117189499e-06,
+      "loss": 3.908,
+      "step": 6165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.599540786907512e-06,
+      "loss": 3.9927,
+      "step": 6166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.59300479178938e-06,
+      "loss": 3.8398,
+      "step": 6167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.586472132958709e-06,
+      "loss": 3.834,
+      "step": 6168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.579942811538539e-06,
+      "loss": 3.9355,
+      "step": 6169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.573416828651329e-06,
+      "loss": 3.9891,
+      "step": 6170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.566894185418986e-06,
+      "loss": 3.9658,
+      "step": 6171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.560374882962821e-06,
+      "loss": 3.9289,
+      "step": 6172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.553858922403577e-06,
+      "loss": 3.8536,
+      "step": 6173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.547346304861414e-06,
+      "loss": 3.9598,
+      "step": 6174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.540837031455945e-06,
+      "loss": 3.9472,
+      "step": 6175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.534331103306182e-06,
+      "loss": 4.0652,
+      "step": 6176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.527828521530562e-06,
+      "loss": 3.9244,
+      "step": 6177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.521329287246965e-06,
+      "loss": 3.8241,
+      "step": 6178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.514833401572678e-06,
+      "loss": 3.8443,
+      "step": 6179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.508340865624415e-06,
+      "loss": 3.8501,
+      "step": 6180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.501851680518322e-06,
+      "loss": 3.9749,
+      "step": 6181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.495365847369974e-06,
+      "loss": 4.0739,
+      "step": 6182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.488883367294354e-06,
+      "loss": 3.983,
+      "step": 6183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.482404241405875e-06,
+      "loss": 3.9087,
+      "step": 6184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.475928470818376e-06,
+      "loss": 3.8862,
+      "step": 6185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.469456056645114e-06,
+      "loss": 3.9655,
+      "step": 6186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.462986999998768e-06,
+      "loss": 3.8481,
+      "step": 6187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.456521301991455e-06,
+      "loss": 3.897,
+      "step": 6188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.450058963734692e-06,
+      "loss": 4.0291,
+      "step": 6189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4435999863394396e-06,
+      "loss": 4.0461,
+      "step": 6190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.437144370916069e-06,
+      "loss": 3.989,
+      "step": 6191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.430692118574374e-06,
+      "loss": 3.954,
+      "step": 6192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.424243230423567e-06,
+      "loss": 3.8527,
+      "step": 6193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.417797707572295e-06,
+      "loss": 3.8644,
+      "step": 6194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.411355551128602e-06,
+      "loss": 3.79,
+      "step": 6195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4049167621999906e-06,
+      "loss": 3.9231,
+      "step": 6196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.398481341893355e-06,
+      "loss": 3.9886,
+      "step": 6197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.392049291315007e-06,
+      "loss": 3.8119,
+      "step": 6198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.385620611570707e-06,
+      "loss": 3.771,
+      "step": 6199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.379195303765616e-06,
+      "loss": 3.8249,
+      "step": 6200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.372773369004314e-06,
+      "loss": 3.8352,
+      "step": 6201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.366354808390803e-06,
+      "loss": 4.0063,
+      "step": 6202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.35993962302852e-06,
+      "loss": 3.8353,
+      "step": 6203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.353527814020301e-06,
+      "loss": 3.9424,
+      "step": 6204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.347119382468413e-06,
+      "loss": 3.8239,
+      "step": 6205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.340714329474542e-06,
+      "loss": 3.7842,
+      "step": 6206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.334312656139776e-06,
+      "loss": 3.8655,
+      "step": 6207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.327914363564659e-06,
+      "loss": 4.0003,
+      "step": 6208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.321519452849119e-06,
+      "loss": 3.8515,
+      "step": 6209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.315127925092511e-06,
+      "loss": 3.9491,
+      "step": 6210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.308739781393629e-06,
+      "loss": 3.7491,
+      "step": 6211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.302355022850655e-06,
+      "loss": 3.8996,
+      "step": 6212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.2959736505612115e-06,
+      "loss": 3.9719,
+      "step": 6213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.2895956656223236e-06,
+      "loss": 3.9791,
+      "step": 6214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.283221069130442e-06,
+      "loss": 3.9421,
+      "step": 6215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.276849862181432e-06,
+      "loss": 4.0011,
+      "step": 6216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.2704820458705785e-06,
+      "loss": 3.9094,
+      "step": 6217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.264117621292594e-06,
+      "loss": 3.8124,
+      "step": 6218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.257756589541588e-06,
+      "loss": 3.8398,
+      "step": 6219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.251398951711095e-06,
+      "loss": 4.0544,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.2450447088940676e-06,
+      "loss": 3.8259,
+      "step": 6221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.2386938621828756e-06,
+      "loss": 3.7429,
+      "step": 6222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.2323464126692925e-06,
+      "loss": 3.97,
+      "step": 6223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.226002361444535e-06,
+      "loss": 4.1013,
+      "step": 6224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.219661709599211e-06,
+      "loss": 3.8658,
+      "step": 6225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.213324458223346e-06,
+      "loss": 3.8841,
+      "step": 6226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.206990608406401e-06,
+      "loss": 4.0085,
+      "step": 6227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.200660161237228e-06,
+      "loss": 3.7753,
+      "step": 6228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.19433311780411e-06,
+      "loss": 3.9219,
+      "step": 6229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.188009479194736e-06,
+      "loss": 3.9078,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.181689246496205e-06,
+      "loss": 3.8983,
+      "step": 6231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.175372420795052e-06,
+      "loss": 3.9404,
+      "step": 6232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.169059003177207e-06,
+      "loss": 3.9876,
+      "step": 6233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.16274899472802e-06,
+      "loss": 4.0465,
+      "step": 6234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.156442396532246e-06,
+      "loss": 3.9217,
+      "step": 6235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.150139209674079e-06,
+      "loss": 3.915,
+      "step": 6236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.1438394352370996e-06,
+      "loss": 3.9681,
+      "step": 6237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.137543074304307e-06,
+      "loss": 4.1174,
+      "step": 6238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.131250127958134e-06,
+      "loss": 3.8611,
+      "step": 6239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.124960597280401e-06,
+      "loss": 4.0104,
+      "step": 6240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.1186744833523535e-06,
+      "loss": 4.0394,
+      "step": 6241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.112391787254645e-06,
+      "loss": 3.9248,
+      "step": 6242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.106112510067346e-06,
+      "loss": 3.8533,
+      "step": 6243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.099836652869927e-06,
+      "loss": 3.9847,
+      "step": 6244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.093564216741301e-06,
+      "loss": 4.1432,
+      "step": 6245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.087295202759751e-06,
+      "loss": 3.977,
+      "step": 6246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.081029612003013e-06,
+      "loss": 3.8909,
+      "step": 6247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.074767445548204e-06,
+      "loss": 4.0126,
+      "step": 6248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0685087044718635e-06,
+      "loss": 3.947,
+      "step": 6249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.062253389849942e-06,
+      "loss": 3.7738,
+      "step": 6250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0560015027578015e-06,
+      "loss": 3.9282,
+      "step": 6251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0497530442702084e-06,
+      "loss": 3.8329,
+      "step": 6252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.043508015461354e-06,
+      "loss": 3.9971,
+      "step": 6253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.037266417404824e-06,
+      "loss": 3.8369,
+      "step": 6254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.031028251173628e-06,
+      "loss": 3.7658,
+      "step": 6255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.024793517840176e-06,
+      "loss": 3.8688,
+      "step": 6256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.018562218476294e-06,
+      "loss": 3.932,
+      "step": 6257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.012334354153209e-06,
+      "loss": 3.9138,
+      "step": 6258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.00610992594156e-06,
+      "loss": 3.9846,
+      "step": 6259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.999888934911409e-06,
+      "loss": 3.9703,
+      "step": 6260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9936713821322116e-06,
+      "loss": 3.869,
+      "step": 6261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.987457268672838e-06,
+      "loss": 3.9244,
+      "step": 6262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981246595601555e-06,
+      "loss": 3.8866,
+      "step": 6263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9750393639860696e-06,
+      "loss": 3.9332,
+      "step": 6264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.968835574893466e-06,
+      "loss": 3.9124,
+      "step": 6265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.962635229390247e-06,
+      "loss": 3.9508,
+      "step": 6266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.95643832854232e-06,
+      "loss": 3.9075,
+      "step": 6267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.950244873415017e-06,
+      "loss": 3.8098,
+      "step": 6268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.944054865073056e-06,
+      "loss": 3.9246,
+      "step": 6269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.93786830458057e-06,
+      "loss": 3.9414,
+      "step": 6270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.931685193001107e-06,
+      "loss": 3.9082,
+      "step": 6271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.925505531397601e-06,
+      "loss": 3.9485,
+      "step": 6272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.919329320832425e-06,
+      "loss": 4.0185,
+      "step": 6273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.913156562367327e-06,
+      "loss": 3.9662,
+      "step": 6274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.906987257063489e-06,
+      "loss": 4.022,
+      "step": 6275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.900821405981482e-06,
+      "loss": 3.982,
+      "step": 6276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.894659010181282e-06,
+      "loss": 3.813,
+      "step": 6277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.888500070722282e-06,
+      "loss": 3.9806,
+      "step": 6278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.882344588663271e-06,
+      "loss": 3.8648,
+      "step": 6279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.87619256506244e-06,
+      "loss": 3.9156,
+      "step": 6280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870044000977406e-06,
+      "loss": 3.7972,
+      "step": 6281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.86389889746518e-06,
+      "loss": 3.7163,
+      "step": 6282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.857757255582171e-06,
+      "loss": 3.7339,
+      "step": 6283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.851619076384201e-06,
+      "loss": 3.9779,
+      "step": 6284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8454843609264885e-06,
+      "loss": 3.8451,
+      "step": 6285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.839353110263667e-06,
+      "loss": 3.9035,
+      "step": 6286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.833225325449767e-06,
+      "loss": 3.8286,
+      "step": 6287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.827101007538221e-06,
+      "loss": 3.8167,
+      "step": 6288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.820980157581884e-06,
+      "loss": 3.7827,
+      "step": 6289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.814862776632992e-06,
+      "loss": 4.052,
+      "step": 6290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8087488657431875e-06,
+      "loss": 3.8863,
+      "step": 6291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.802638425963538e-06,
+      "loss": 3.9025,
+      "step": 6292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.796531458344492e-06,
+      "loss": 3.9062,
+      "step": 6293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7904279639359035e-06,
+      "loss": 3.852,
+      "step": 6294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7843279437870345e-06,
+      "loss": 3.9219,
+      "step": 6295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.778231398946556e-06,
+      "loss": 3.9123,
+      "step": 6296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772138330462534e-06,
+      "loss": 3.7207,
+      "step": 6297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.766048739382431e-06,
+      "loss": 3.9471,
+      "step": 6298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.759962626753123e-06,
+      "loss": 3.9188,
+      "step": 6299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.753879993620877e-06,
+      "loss": 3.9994,
+      "step": 6300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.747800841031377e-06,
+      "loss": 3.8872,
+      "step": 6301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.741725170029693e-06,
+      "loss": 3.9319,
+      "step": 6302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.73565298166031e-06,
+      "loss": 3.9664,
+      "step": 6303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7295842769671055e-06,
+      "loss": 3.9358,
+      "step": 6304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.723519056993358e-06,
+      "loss": 3.9213,
+      "step": 6305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.717457322781749e-06,
+      "loss": 3.9183,
+      "step": 6306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.711399075374362e-06,
+      "loss": 3.8577,
+      "step": 6307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.705344315812682e-06,
+      "loss": 3.8103,
+      "step": 6308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.699293045137582e-06,
+      "loss": 3.998,
+      "step": 6309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.693245264389351e-06,
+      "loss": 3.9727,
+      "step": 6310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.687200974607684e-06,
+      "loss": 4.0987,
+      "step": 6311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.681160176831656e-06,
+      "loss": 4.059,
+      "step": 6312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6751228720997495e-06,
+      "loss": 3.8703,
+      "step": 6313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.669089061449846e-06,
+      "loss": 3.7868,
+      "step": 6314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.66305874591923e-06,
+      "loss": 3.968,
+      "step": 6315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.657031926544575e-06,
+      "loss": 3.9132,
+      "step": 6316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.651008604361976e-06,
+      "loss": 3.9455,
+      "step": 6317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.644988780406895e-06,
+      "loss": 3.982,
+      "step": 6318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.638972455714225e-06,
+      "loss": 3.8838,
+      "step": 6319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.632959631318235e-06,
+      "loss": 3.8448,
+      "step": 6320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.626950308252601e-06,
+      "loss": 3.8308,
+      "step": 6321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.620944487550391e-06,
+      "loss": 3.9051,
+      "step": 6322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.614942170244074e-06,
+      "loss": 3.7937,
+      "step": 6323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.608943357365527e-06,
+      "loss": 3.8796,
+      "step": 6324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.60294804994601e-06,
+      "loss": 3.9495,
+      "step": 6325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.596956249016188e-06,
+      "loss": 3.8886,
+      "step": 6326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.590967955606115e-06,
+      "loss": 3.8085,
+      "step": 6327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.584983170745249e-06,
+      "loss": 3.8198,
+      "step": 6328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.579001895462453e-06,
+      "loss": 3.8948,
+      "step": 6329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.573024130785971e-06,
+      "loss": 3.8898,
+      "step": 6330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.567049877743446e-06,
+      "loss": 3.9377,
+      "step": 6331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5610791373619315e-06,
+      "loss": 3.8762,
+      "step": 6332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.555111910667861e-06,
+      "loss": 4.1144,
+      "step": 6333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.549148198687072e-06,
+      "loss": 3.7144,
+      "step": 6334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.543188002444795e-06,
+      "loss": 3.9078,
+      "step": 6335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.537231322965654e-06,
+      "loss": 3.894,
+      "step": 6336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.531278161273666e-06,
+      "loss": 3.8938,
+      "step": 6337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.525328518392258e-06,
+      "loss": 4.0733,
+      "step": 6338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.519382395344246e-06,
+      "loss": 3.9514,
+      "step": 6339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.513439793151833e-06,
+      "loss": 3.9897,
+      "step": 6340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.507500712836621e-06,
+      "loss": 3.8439,
+      "step": 6341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.501565155419602e-06,
+      "loss": 3.9996,
+      "step": 6342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.495633121921175e-06,
+      "loss": 3.896,
+      "step": 6343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.489704613361112e-06,
+      "loss": 4.0324,
+      "step": 6344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.483779630758608e-06,
+      "loss": 4.0365,
+      "step": 6345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4778581751322264e-06,
+      "loss": 3.7718,
+      "step": 6346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.471940247499942e-06,
+      "loss": 3.892,
+      "step": 6347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.466025848879113e-06,
+      "loss": 3.996,
+      "step": 6348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4601149802864864e-06,
+      "loss": 4.0637,
+      "step": 6349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.454207642738217e-06,
+      "loss": 3.912,
+      "step": 6350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.44830383724984e-06,
+      "loss": 3.9541,
+      "step": 6351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.442403564836284e-06,
+      "loss": 3.8568,
+      "step": 6352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.436506826511883e-06,
+      "loss": 3.9187,
+      "step": 6353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.430613623290353e-06,
+      "loss": 3.9433,
+      "step": 6354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424723956184795e-06,
+      "loss": 3.912,
+      "step": 6355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.418837826207725e-06,
+      "loss": 3.9721,
+      "step": 6356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.41295523437103e-06,
+      "loss": 3.9704,
+      "step": 6357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4070761816859955e-06,
+      "loss": 3.8559,
+      "step": 6358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.401200669163291e-06,
+      "loss": 3.9184,
+      "step": 6359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.395328697813e-06,
+      "loss": 3.8736,
+      "step": 6360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.389460268644577e-06,
+      "loss": 3.8583,
+      "step": 6361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.383595382666872e-06,
+      "loss": 3.914,
+      "step": 6362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.377734040888126e-06,
+      "loss": 3.9743,
+      "step": 6363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.371876244315973e-06,
+      "loss": 3.9911,
+      "step": 6364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.366021993957428e-06,
+      "loss": 3.9408,
+      "step": 6365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.360171290818918e-06,
+      "loss": 3.9241,
+      "step": 6366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3543241359062346e-06,
+      "loss": 3.789,
+      "step": 6367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.348480530224583e-06,
+      "loss": 3.9157,
+      "step": 6368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.342640474778542e-06,
+      "loss": 3.8697,
+      "step": 6369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.336803970572084e-06,
+      "loss": 3.9025,
+      "step": 6370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3309710186085705e-06,
+      "loss": 3.9133,
+      "step": 6371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.325141619890757e-06,
+      "loss": 3.9221,
+      "step": 6372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.319315775420777e-06,
+      "loss": 3.9979,
+      "step": 6373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.313493486200162e-06,
+      "loss": 3.7658,
+      "step": 6374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.307674753229845e-06,
+      "loss": 4.0939,
+      "step": 6375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.301859577510123e-06,
+      "loss": 3.8273,
+      "step": 6376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.296047960040692e-06,
+      "loss": 3.8851,
+      "step": 6377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.290239901820639e-06,
+      "loss": 3.8976,
+      "step": 6378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2844354038484354e-06,
+      "loss": 3.863,
+      "step": 6379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2786344671219336e-06,
+      "loss": 3.8685,
+      "step": 6380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.272837092638396e-06,
+      "loss": 3.8374,
+      "step": 6381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.267043281394454e-06,
+      "loss": 3.9731,
+      "step": 6382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.26125303438612e-06,
+      "loss": 3.9743,
+      "step": 6383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.255466352608822e-06,
+      "loss": 3.9149,
+      "step": 6384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.249683237057348e-06,
+      "loss": 3.8828,
+      "step": 6385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2439036887258835e-06,
+      "loss": 3.8878,
+      "step": 6386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.238127708608003e-06,
+      "loss": 3.8638,
+      "step": 6387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.232355297696653e-06,
+      "loss": 3.9337,
+      "step": 6388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.226586456984192e-06,
+      "loss": 3.8367,
+      "step": 6389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2208211874623455e-06,
+      "loss": 3.8385,
+      "step": 6390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.215059490122231e-06,
+      "loss": 3.8155,
+      "step": 6391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.209301365954344e-06,
+      "loss": 3.9088,
+      "step": 6392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2035468159485835e-06,
+      "loss": 3.9168,
+      "step": 6393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.197795841094221e-06,
+      "loss": 3.8877,
+      "step": 6394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.192048442379904e-06,
+      "loss": 3.953,
+      "step": 6395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.186304620793693e-06,
+      "loss": 3.8449,
+      "step": 6396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.180564377323013e-06,
+      "loss": 3.8505,
+      "step": 6397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.174827712954674e-06,
+      "loss": 3.9611,
+      "step": 6398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.169094628674877e-06,
+      "loss": 3.8496,
+      "step": 6399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1633651254692065e-06,
+      "loss": 3.7525,
+      "step": 6400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.157639204322625e-06,
+      "loss": 3.9686,
+      "step": 6401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.151916866219493e-06,
+      "loss": 3.9326,
+      "step": 6402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.14619811214354e-06,
+      "loss": 3.9208,
+      "step": 6403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.140482943077895e-06,
+      "loss": 3.9997,
+      "step": 6404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.134771360005054e-06,
+      "loss": 3.7758,
+      "step": 6405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.12906336390691e-06,
+      "loss": 3.8735,
+      "step": 6406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.123358955764728e-06,
+      "loss": 3.9622,
+      "step": 6407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1176581365591656e-06,
+      "loss": 4.0509,
+      "step": 6408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.111960907270251e-06,
+      "loss": 3.8256,
+      "step": 6409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1062672688774205e-06,
+      "loss": 3.7946,
+      "step": 6410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.100577222359459e-06,
+      "loss": 3.9001,
+      "step": 6411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.094890768694568e-06,
+      "loss": 3.9919,
+      "step": 6412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.089207908860304e-06,
+      "loss": 3.9566,
+      "step": 6413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.08352864383362e-06,
+      "loss": 3.8633,
+      "step": 6414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0778529745908455e-06,
+      "loss": 4.0036,
+      "step": 6415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.072180902107692e-06,
+      "loss": 3.969,
+      "step": 6416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.066512427359262e-06,
+      "loss": 3.9234,
+      "step": 6417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.060847551320026e-06,
+      "loss": 3.9041,
+      "step": 6418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.055186274963846e-06,
+      "loss": 3.8818,
+      "step": 6419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.049528599263949e-06,
+      "loss": 3.9475,
+      "step": 6420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0438745251929755e-06,
+      "loss": 3.9004,
+      "step": 6421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.038224053722911e-06,
+      "loss": 3.8949,
+      "step": 6422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.032577185825143e-06,
+      "loss": 3.8645,
+      "step": 6423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.026933922470427e-06,
+      "loss": 3.8758,
+      "step": 6424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.021294264628914e-06,
+      "loss": 3.9012,
+      "step": 6425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.015658213270127e-06,
+      "loss": 3.9901,
+      "step": 6426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.01002576936296e-06,
+      "loss": 3.8426,
+      "step": 6427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.004396933875701e-06,
+      "loss": 3.782,
+      "step": 6428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9987717077760055e-06,
+      "loss": 3.9514,
+      "step": 6429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9931500920309285e-06,
+      "loss": 3.8837,
+      "step": 6430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.987532087606874e-06,
+      "loss": 4.0189,
+      "step": 6431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.981917695469659e-06,
+      "loss": 3.9183,
+      "step": 6432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.976306916584455e-06,
+      "loss": 3.8766,
+      "step": 6433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.970699751915818e-06,
+      "loss": 3.8301,
+      "step": 6434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.965096202427687e-06,
+      "loss": 3.9092,
+      "step": 6435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.959496269083377e-06,
+      "loss": 4.0536,
+      "step": 6436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.953899952845572e-06,
+      "loss": 3.8826,
+      "step": 6437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.948307254676362e-06,
+      "loss": 3.9206,
+      "step": 6438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.942718175537177e-06,
+      "loss": 3.8923,
+      "step": 6439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.937132716388864e-06,
+      "loss": 3.8382,
+      "step": 6440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.931550878191617e-06,
+      "loss": 3.9422,
+      "step": 6441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9259726619050225e-06,
+      "loss": 3.8709,
+      "step": 6442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9203980684880375e-06,
+      "loss": 3.9311,
+      "step": 6443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9148270988989984e-06,
+      "loss": 3.9051,
+      "step": 6444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.909259754095618e-06,
+      "loss": 3.9634,
+      "step": 6445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.903696035034996e-06,
+      "loss": 3.9623,
+      "step": 6446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.898135942673595e-06,
+      "loss": 3.9307,
+      "step": 6447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.892579477967254e-06,
+      "loss": 3.982,
+      "step": 6448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.887026641871203e-06,
+      "loss": 3.8237,
+      "step": 6449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.881477435340033e-06,
+      "loss": 4.0021,
+      "step": 6450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.875931859327722e-06,
+      "loss": 3.9655,
+      "step": 6451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.870389914787609e-06,
+      "loss": 3.9298,
+      "step": 6452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.864851602672431e-06,
+      "loss": 3.9617,
+      "step": 6453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.859316923934283e-06,
+      "loss": 3.9286,
+      "step": 6454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.853785879524641e-06,
+      "loss": 3.9476,
+      "step": 6455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.84825847039435e-06,
+      "loss": 3.8622,
+      "step": 6456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.842734697493636e-06,
+      "loss": 3.928,
+      "step": 6457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8372145617721096e-06,
+      "loss": 3.8575,
+      "step": 6458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.831698064178741e-06,
+      "loss": 3.9282,
+      "step": 6459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.826185205661873e-06,
+      "loss": 3.8816,
+      "step": 6460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.82067598716924e-06,
+      "loss": 3.6912,
+      "step": 6461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.815170409647939e-06,
+      "loss": 4.0485,
+      "step": 6462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8096684740444395e-06,
+      "loss": 3.9527,
+      "step": 6463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8041701813045868e-06,
+      "loss": 3.8512,
+      "step": 6464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.798675532373605e-06,
+      "loss": 3.8844,
+      "step": 6465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.793184528196081e-06,
+      "loss": 3.7459,
+      "step": 6466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.787697169715984e-06,
+      "loss": 3.9895,
+      "step": 6467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7822134578766655e-06,
+      "loss": 3.9661,
+      "step": 6468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.776733393620832e-06,
+      "loss": 3.9122,
+      "step": 6469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7712569778905695e-06,
+      "loss": 3.9291,
+      "step": 6470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.765784211627335e-06,
+      "loss": 3.9088,
+      "step": 6471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7603150957719655e-06,
+      "loss": 3.9438,
+      "step": 6472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7548496312646535e-06,
+      "loss": 3.9136,
+      "step": 6473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7493878190449944e-06,
+      "loss": 3.9621,
+      "step": 6474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.743929660051923e-06,
+      "loss": 3.8299,
+      "step": 6475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.738475155223761e-06,
+      "loss": 3.9409,
+      "step": 6476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7330243054982117e-06,
+      "loss": 3.8799,
+      "step": 6477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.72757711181233e-06,
+      "loss": 3.7882,
+      "step": 6478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7221335751025543e-06,
+      "loss": 3.9532,
+      "step": 6479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7166936963046933e-06,
+      "loss": 3.8359,
+      "step": 6480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.711257476353916e-06,
+      "loss": 3.9124,
+      "step": 6481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.705824916184783e-06,
+      "loss": 3.8922,
+      "step": 6482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.700396016731214e-06,
+      "loss": 3.7438,
+      "step": 6483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.694970778926493e-06,
+      "loss": 3.8712,
+      "step": 6484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.689549203703277e-06,
+      "loss": 3.8383,
+      "step": 6485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6841312919936142e-06,
+      "loss": 3.8766,
+      "step": 6486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.678717044728894e-06,
+      "loss": 3.9231,
+      "step": 6487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6733064628398874e-06,
+      "loss": 3.8729,
+      "step": 6488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6678995472567452e-06,
+      "loss": 3.9079,
+      "step": 6489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.662496298908974e-06,
+      "loss": 4.0134,
+      "step": 6490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6570967187254556e-06,
+      "loss": 3.9523,
+      "step": 6491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6517008076344393e-06,
+      "loss": 4.0698,
+      "step": 6492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6463085665635417e-06,
+      "loss": 3.9479,
+      "step": 6493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.640919996439751e-06,
+      "loss": 3.9445,
+      "step": 6494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.63553509818943e-06,
+      "loss": 3.7827,
+      "step": 6495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6301538727383094e-06,
+      "loss": 4.025,
+      "step": 6496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6247763210114777e-06,
+      "loss": 3.8492,
+      "step": 6497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619402443933398e-06,
+      "loss": 3.7889,
+      "step": 6498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614032242427903e-06,
+      "loss": 3.9052,
+      "step": 6499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.608665717418197e-06,
+      "loss": 3.9774,
+      "step": 6500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6033028698268414e-06,
+      "loss": 3.8438,
+      "step": 6501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5979437005757677e-06,
+      "loss": 3.8414,
+      "step": 6502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.592588210586287e-06,
+      "loss": 4.0667,
+      "step": 6503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.587236400779073e-06,
+      "loss": 3.7636,
+      "step": 6504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5818882720741605e-06,
+      "loss": 3.9037,
+      "step": 6505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5765438253909544e-06,
+      "loss": 4.035,
+      "step": 6506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.571203061648226e-06,
+      "loss": 4.0324,
+      "step": 6507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.565865981764116e-06,
+      "loss": 3.8228,
+      "step": 6508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5605325866561264e-06,
+      "loss": 4.0,
+      "step": 6509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5552028772411346e-06,
+      "loss": 3.8848,
+      "step": 6510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.54987685443538e-06,
+      "loss": 3.9558,
+      "step": 6511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.544554519154464e-06,
+      "loss": 3.8686,
+      "step": 6512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.539235872313354e-06,
+      "loss": 3.9379,
+      "step": 6513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5339209148263967e-06,
+      "loss": 3.9351,
+      "step": 6514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5286096476072895e-06,
+      "loss": 3.8627,
+      "step": 6515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5233020715690984e-06,
+      "loss": 3.895,
+      "step": 6516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5179981876242636e-06,
+      "loss": 3.9199,
+      "step": 6517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.512697996684583e-06,
+      "loss": 3.9913,
+      "step": 6518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.50740149966122e-06,
+      "loss": 3.9868,
+      "step": 6519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.502108697464701e-06,
+      "loss": 3.9198,
+      "step": 6520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.496819591004921e-06,
+      "loss": 3.8993,
+      "step": 6521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491534181191136e-06,
+      "loss": 3.8158,
+      "step": 6522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4862524689319776e-06,
+      "loss": 3.9453,
+      "step": 6523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.480974455135422e-06,
+      "loss": 3.7864,
+      "step": 6524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4757001407088346e-06,
+      "loss": 3.8354,
+      "step": 6525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4704295265589215e-06,
+      "loss": 3.8359,
+      "step": 6526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4651626135917687e-06,
+      "loss": 3.9082,
+      "step": 6527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.459899402712813e-06,
+      "loss": 3.8026,
+      "step": 6528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4546398948268664e-06,
+      "loss": 3.9824,
+      "step": 6529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4493840908380896e-06,
+      "loss": 3.945,
+      "step": 6530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4441319916500236e-06,
+      "loss": 3.8558,
+      "step": 6531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4388835981655736e-06,
+      "loss": 3.8581,
+      "step": 6532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4336389112869877e-06,
+      "loss": 3.8587,
+      "step": 6533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4283979319158945e-06,
+      "loss": 3.8042,
+      "step": 6534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4231606609532756e-06,
+      "loss": 3.8726,
+      "step": 6535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4179270992994783e-06,
+      "loss": 3.8754,
+      "step": 6536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4126972478542076e-06,
+      "loss": 4.0065,
+      "step": 6537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4074711075165494e-06,
+      "loss": 3.9273,
+      "step": 6538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.402248679184927e-06,
+      "loss": 3.8238,
+      "step": 6539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.397029963757134e-06,
+      "loss": 3.8303,
+      "step": 6540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3918149621303407e-06,
+      "loss": 3.7277,
+      "step": 6541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3866036752010583e-06,
+      "loss": 3.9996,
+      "step": 6542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.381396103865167e-06,
+      "loss": 3.7276,
+      "step": 6543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.376192249017912e-06,
+      "loss": 3.9861,
+      "step": 6544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.370992111553886e-06,
+      "loss": 3.795,
+      "step": 6545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3657956923670686e-06,
+      "loss": 3.7652,
+      "step": 6546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3606029923507753e-06,
+      "loss": 3.9758,
+      "step": 6547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.355414012397695e-06,
+      "loss": 4.0365,
+      "step": 6548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35022875339987e-06,
+      "loss": 3.8911,
+      "step": 6549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3450472162487036e-06,
+      "loss": 3.801,
+      "step": 6550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.339869401834972e-06,
+      "loss": 3.8716,
+      "step": 6551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3346953110487883e-06,
+      "loss": 3.9959,
+      "step": 6552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.329524944779655e-06,
+      "loss": 3.9312,
+      "step": 6553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3243583039164057e-06,
+      "loss": 3.8973,
+      "step": 6554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3191953893472518e-06,
+      "loss": 3.8991,
+      "step": 6555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3140362019597555e-06,
+      "loss": 3.9004,
+      "step": 6556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3088807426408435e-06,
+      "loss": 3.7527,
+      "step": 6557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.303729012276788e-06,
+      "loss": 3.8427,
+      "step": 6558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2985810117532464e-06,
+      "loss": 3.9598,
+      "step": 6559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2934367419552082e-06,
+      "loss": 3.8478,
+      "step": 6560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.288296203767044e-06,
+      "loss": 3.9328,
+      "step": 6561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2831593980724663e-06,
+      "loss": 4.0574,
+      "step": 6562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2780263257545517e-06,
+      "loss": 3.8727,
+      "step": 6563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.272896987695734e-06,
+      "loss": 3.9737,
+      "step": 6564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.267771384777804e-06,
+      "loss": 3.9806,
+      "step": 6565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.262649517881913e-06,
+      "loss": 3.8673,
+      "step": 6566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2575313878885737e-06,
+      "loss": 3.8955,
+      "step": 6567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2524169956776457e-06,
+      "loss": 3.983,
+      "step": 6568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.247306342128359e-06,
+      "loss": 3.7906,
+      "step": 6569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.242199428119291e-06,
+      "loss": 3.9184,
+      "step": 6570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2370962545283796e-06,
+      "loss": 3.8523,
+      "step": 6571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2319968222329216e-06,
+      "loss": 3.7745,
+      "step": 6572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.226901132109558e-06,
+      "loss": 3.8206,
+      "step": 6573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2218091850343107e-06,
+      "loss": 3.8693,
+      "step": 6574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.21672098188254e-06,
+      "loss": 3.8653,
+      "step": 6575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2116365235289656e-06,
+      "loss": 3.8232,
+      "step": 6576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2065558108476614e-06,
+      "loss": 4.0586,
+      "step": 6577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2014788447120694e-06,
+      "loss": 3.9366,
+      "step": 6578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.196405625994972e-06,
+      "loss": 3.8244,
+      "step": 6579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1913361555685194e-06,
+      "loss": 3.8274,
+      "step": 6580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1862704343042004e-06,
+      "loss": 3.9237,
+      "step": 6581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1812084630728876e-06,
+      "loss": 3.8515,
+      "step": 6582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1761502427447854e-06,
+      "loss": 3.9669,
+      "step": 6583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1710957741894615e-06,
+      "loss": 3.8681,
+      "step": 6584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.166045058275835e-06,
+      "loss": 3.8987,
+      "step": 6585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1609980958721833e-06,
+      "loss": 4.0152,
+      "step": 6586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1559548878461324e-06,
+      "loss": 3.7575,
+      "step": 6587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1509154350646747e-06,
+      "loss": 3.9587,
+      "step": 6588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1458797383941556e-06,
+      "loss": 3.9947,
+      "step": 6589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1408477987002667e-06,
+      "loss": 3.7312,
+      "step": 6590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.135819616848051e-06,
+      "loss": 3.9812,
+      "step": 6591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.130795193701916e-06,
+      "loss": 3.9131,
+      "step": 6592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.125774530125616e-06,
+      "loss": 3.9017,
+      "step": 6593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1207576269822567e-06,
+      "loss": 3.9075,
+      "step": 6594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1157444851343144e-06,
+      "loss": 3.8129,
+      "step": 6595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1107351054435906e-06,
+      "loss": 3.9878,
+      "step": 6596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.105729488771272e-06,
+      "loss": 3.9386,
+      "step": 6597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1007276359778735e-06,
+      "loss": 4.0162,
+      "step": 6598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0957295479232735e-06,
+      "loss": 3.871,
+      "step": 6599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0907352254666983e-06,
+      "loss": 3.8998,
+      "step": 6600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0857446694667335e-06,
+      "loss": 3.8553,
+      "step": 6601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.080757880781307e-06,
+      "loss": 3.9696,
+      "step": 6602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.075774860267716e-06,
+      "loss": 3.8783,
+      "step": 6603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0707956087825924e-06,
+      "loss": 3.938,
+      "step": 6604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.065820127181923e-06,
+      "loss": 3.8563,
+      "step": 6605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.060848416321063e-06,
+      "loss": 4.003,
+      "step": 6606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.055880477054701e-06,
+      "loss": 3.8544,
+      "step": 6607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0509163102368816e-06,
+      "loss": 3.9963,
+      "step": 6608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0459559167209994e-06,
+      "loss": 4.0254,
+      "step": 6609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0409992973598146e-06,
+      "loss": 3.9751,
+      "step": 6610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.03604645300542e-06,
+      "loss": 3.9298,
+      "step": 6611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.031097384509271e-06,
+      "loss": 3.9955,
+      "step": 6612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.026152092722165e-06,
+      "loss": 4.007,
+      "step": 6613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0212105784942492e-06,
+      "loss": 4.0,
+      "step": 6614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0162728426750444e-06,
+      "loss": 3.9205,
+      "step": 6615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0113388861133907e-06,
+      "loss": 3.8782,
+      "step": 6616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.006408709657496e-06,
+      "loss": 3.8292,
+      "step": 6617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0014823141549185e-06,
+      "loss": 3.9104,
+      "step": 6618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9965597004525614e-06,
+      "loss": 3.9293,
+      "step": 6619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.991640869396675e-06,
+      "loss": 3.9154,
+      "step": 6620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.986725821832867e-06,
+      "loss": 3.8942,
+      "step": 6621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9818145586060915e-06,
+      "loss": 3.9561,
+      "step": 6622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9769070805606453e-06,
+      "loss": 3.7231,
+      "step": 6623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9720033885401814e-06,
+      "loss": 3.8995,
+      "step": 6624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.967103483387715e-06,
+      "loss": 3.8576,
+      "step": 6625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.962207365945585e-06,
+      "loss": 3.8763,
+      "step": 6626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.957315037055494e-06,
+      "loss": 3.8198,
+      "step": 6627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.952426497558489e-06,
+      "loss": 3.8854,
+      "step": 6628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9475417482949656e-06,
+      "loss": 3.8275,
+      "step": 6629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.942660790104662e-06,
+      "loss": 4.0276,
+      "step": 6630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.937783623826687e-06,
+      "loss": 3.8768,
+      "step": 6631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9329102502994753e-06,
+      "loss": 3.9825,
+      "step": 6632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.92804067036081e-06,
+      "loss": 3.9723,
+      "step": 6633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9231748848478374e-06,
+      "loss": 3.6692,
+      "step": 6634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.918312894597039e-06,
+      "loss": 3.9173,
+      "step": 6635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.913454700444246e-06,
+      "loss": 3.926,
+      "step": 6636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9086003032246403e-06,
+      "loss": 3.9831,
+      "step": 6637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.903749703772743e-06,
+      "loss": 3.8382,
+      "step": 6638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8989029029224376e-06,
+      "loss": 3.9878,
+      "step": 6639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8940599015069404e-06,
+      "loss": 3.8527,
+      "step": 6640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8892207003588217e-06,
+      "loss": 3.7888,
+      "step": 6641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8843853003099885e-06,
+      "loss": 3.9291,
+      "step": 6642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.879553702191715e-06,
+      "loss": 3.8243,
+      "step": 6643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8747259068346022e-06,
+      "loss": 3.925,
+      "step": 6644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.869901915068596e-06,
+      "loss": 3.8895,
+      "step": 6645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8650817277230123e-06,
+      "loss": 3.9008,
+      "step": 6646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.860265345626489e-06,
+      "loss": 3.9012,
+      "step": 6647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.85545276960702e-06,
+      "loss": 3.8293,
+      "step": 6648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.850644000491942e-06,
+      "loss": 3.8695,
+      "step": 6649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8458390391079363e-06,
+      "loss": 3.9972,
+      "step": 6650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8410378862810256e-06,
+      "loss": 3.8986,
+      "step": 6651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.836240542836599e-06,
+      "loss": 4.022,
+      "step": 6652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8314470095993612e-06,
+      "loss": 3.8873,
+      "step": 6653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.826657287393389e-06,
+      "loss": 3.8352,
+      "step": 6654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8218713770420817e-06,
+      "loss": 3.9207,
+      "step": 6655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8170892793681987e-06,
+      "loss": 3.9063,
+      "step": 6656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.812310995193834e-06,
+      "loss": 3.894,
+      "step": 6657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.807536525340429e-06,
+      "loss": 3.8835,
+      "step": 6658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.802765870628768e-06,
+      "loss": 4.0454,
+      "step": 6659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.797999031878992e-06,
+      "loss": 3.7625,
+      "step": 6660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7932360099105658e-06,
+      "loss": 3.9717,
+      "step": 6661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.788476805542317e-06,
+      "loss": 3.76,
+      "step": 6662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7837214195924026e-06,
+      "loss": 3.9536,
+      "step": 6663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7789698528783324e-06,
+      "loss": 3.9077,
+      "step": 6664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7742221062169505e-06,
+      "loss": 4.0505,
+      "step": 6665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.769478180424445e-06,
+      "loss": 3.824,
+      "step": 6666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7647380763163673e-06,
+      "loss": 3.9056,
+      "step": 6667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.760001794707584e-06,
+      "loss": 3.7411,
+      "step": 6668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7552693364123224e-06,
+      "loss": 3.8783,
+      "step": 6669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7505407022441393e-06,
+      "loss": 3.8249,
+      "step": 6670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7458158930159517e-06,
+      "loss": 3.9149,
+      "step": 6671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7410949095400068e-06,
+      "loss": 3.8104,
+      "step": 6672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7363777526278915e-06,
+      "loss": 3.884,
+      "step": 6673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.731664423090541e-06,
+      "loss": 3.8641,
+      "step": 6674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.726954921738234e-06,
+      "loss": 3.8845,
+      "step": 6675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7222492493805903e-06,
+      "loss": 3.8156,
+      "step": 6676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.717547406826565e-06,
+      "loss": 3.8363,
+      "step": 6677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.712849394884462e-06,
+      "loss": 4.0145,
+      "step": 6678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7081552143619183e-06,
+      "loss": 3.8139,
+      "step": 6679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7034648660659244e-06,
+      "loss": 3.9587,
+      "step": 6680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6987783508028025e-06,
+      "loss": 3.83,
+      "step": 6681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6940956693782214e-06,
+      "loss": 3.9638,
+      "step": 6682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6894168225971877e-06,
+      "loss": 3.8699,
+      "step": 6683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.684741811264052e-06,
+      "loss": 3.8034,
+      "step": 6684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.680070636182497e-06,
+      "loss": 3.8429,
+      "step": 6685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6754032981555544e-06,
+      "loss": 3.7917,
+      "step": 6686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6707397979855884e-06,
+      "loss": 3.9028,
+      "step": 6687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6660801364743136e-06,
+      "loss": 3.9449,
+      "step": 6688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6614243144227892e-06,
+      "loss": 3.8017,
+      "step": 6689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6567723326313908e-06,
+      "loss": 4.0278,
+      "step": 6690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6521241918998575e-06,
+      "loss": 3.8989,
+      "step": 6691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.647479893027252e-06,
+      "loss": 3.811,
+      "step": 6692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6428394368119864e-06,
+      "loss": 3.8543,
+      "step": 6693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.638202824051808e-06,
+      "loss": 3.9338,
+      "step": 6694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.633570055543802e-06,
+      "loss": 3.9993,
+      "step": 6695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6289411320843976e-06,
+      "loss": 3.9072,
+      "step": 6696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.624316054469364e-06,
+      "loss": 3.8799,
+      "step": 6697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6196948234937957e-06,
+      "loss": 3.8623,
+      "step": 6698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.615077439952146e-06,
+      "loss": 3.8738,
+      "step": 6699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6104639046381912e-06,
+      "loss": 3.8929,
+      "step": 6700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.605854218345055e-06,
+      "loss": 3.859,
+      "step": 6701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.60124838186519e-06,
+      "loss": 3.7593,
+      "step": 6702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5966463959904014e-06,
+      "loss": 3.9918,
+      "step": 6703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5920482615118175e-06,
+      "loss": 3.9464,
+      "step": 6704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.587453979219914e-06,
+      "loss": 3.9946,
+      "step": 6705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.582863549904499e-06,
+      "loss": 3.8376,
+      "step": 6706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.578276974354718e-06,
+      "loss": 3.7936,
+      "step": 6707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.573694253359063e-06,
+      "loss": 3.7127,
+      "step": 6708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5691153877053562e-06,
+      "loss": 3.8899,
+      "step": 6709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.564540378180749e-06,
+      "loss": 3.9115,
+      "step": 6710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.559969225571751e-06,
+      "loss": 3.7541,
+      "step": 6711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.555401930664189e-06,
+      "loss": 3.9343,
+      "step": 6712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.550838494243235e-06,
+      "loss": 3.906,
+      "step": 6713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.546278917093398e-06,
+      "loss": 3.86,
+      "step": 6714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.541723199998522e-06,
+      "loss": 4.0122,
+      "step": 6715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.537171343741779e-06,
+      "loss": 3.8017,
+      "step": 6716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.532623349105695e-06,
+      "loss": 3.937,
+      "step": 6717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.528079216872126e-06,
+      "loss": 3.999,
+      "step": 6718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5235389478222572e-06,
+      "loss": 4.029,
+      "step": 6719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.519002542736612e-06,
+      "loss": 3.8052,
+      "step": 6720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.514470002395053e-06,
+      "loss": 3.7919,
+      "step": 6721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.509941327576773e-06,
+      "loss": 3.8992,
+      "step": 6722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5054165190603025e-06,
+      "loss": 3.8547,
+      "step": 6723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.500895577623516e-06,
+      "loss": 3.7717,
+      "step": 6724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.496378504043609e-06,
+      "loss": 3.9936,
+      "step": 6725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4918652990971236e-06,
+      "loss": 3.9017,
+      "step": 6726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4873559635599313e-06,
+      "loss": 3.856,
+      "step": 6727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.48285049820724e-06,
+      "loss": 3.8156,
+      "step": 6728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.478348903813585e-06,
+      "loss": 3.8552,
+      "step": 6729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4738511811528505e-06,
+      "loss": 3.9266,
+      "step": 6730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4693573309982343e-06,
+      "loss": 3.9519,
+      "step": 6731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4648673541222995e-06,
+      "loss": 3.9717,
+      "step": 6732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4603812512969143e-06,
+      "loss": 3.8698,
+      "step": 6733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4558990232932953e-06,
+      "loss": 3.9125,
+      "step": 6734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.451420670881982e-06,
+      "loss": 3.927,
+      "step": 6735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4469461948328637e-06,
+      "loss": 3.795,
+      "step": 6736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4424755959151556e-06,
+      "loss": 3.8764,
+      "step": 6737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4380088748973933e-06,
+      "loss": 3.8842,
+      "step": 6738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4335460325474736e-06,
+      "loss": 3.9026,
+      "step": 6739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4290870696326046e-06,
+      "loss": 4.0167,
+      "step": 6740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4246319869193317e-06,
+      "loss": 4.1043,
+      "step": 6741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4201807851735344e-06,
+      "loss": 4.0417,
+      "step": 6742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4157334651604303e-06,
+      "loss": 3.9314,
+      "step": 6743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4112900276445582e-06,
+      "loss": 3.9507,
+      "step": 6744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.406850473389799e-06,
+      "loss": 3.895,
+      "step": 6745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4024148031593723e-06,
+      "loss": 3.8877,
+      "step": 6746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.397983017715813e-06,
+      "loss": 3.8494,
+      "step": 6747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.393555117821e-06,
+      "loss": 3.8096,
+      "step": 6748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3891311042361366e-06,
+      "loss": 3.8765,
+      "step": 6749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.384710977721766e-06,
+      "loss": 3.9076,
+      "step": 6750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3802947390377556e-06,
+      "loss": 3.7307,
+      "step": 6751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3758823889433068e-06,
+      "loss": 3.8214,
+      "step": 6752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3714739281969546e-06,
+      "loss": 4.0425,
+      "step": 6753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.367069357556573e-06,
+      "loss": 3.7521,
+      "step": 6754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3626686777793504e-06,
+      "loss": 3.8522,
+      "step": 6755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3582718896218187e-06,
+      "loss": 3.9689,
+      "step": 6756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3538789938398335e-06,
+      "loss": 4.0187,
+      "step": 6757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.349489991188586e-06,
+      "loss": 3.8854,
+      "step": 6758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3451048824225914e-06,
+      "loss": 3.8173,
+      "step": 6759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.340723668295711e-06,
+      "loss": 3.8677,
+      "step": 6760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3363463495611194e-06,
+      "loss": 3.8191,
+      "step": 6761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3319729269713263e-06,
+      "loss": 3.8082,
+      "step": 6762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3276034012781803e-06,
+      "loss": 3.9091,
+      "step": 6763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.323237773232853e-06,
+      "loss": 3.8444,
+      "step": 6764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3188760435858435e-06,
+      "loss": 4.0054,
+      "step": 6765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3145182130869775e-06,
+      "loss": 3.9753,
+      "step": 6766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.31016428248543e-06,
+      "loss": 4.0466,
+      "step": 6767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3058142525296866e-06,
+      "loss": 3.7657,
+      "step": 6768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.301468123967565e-06,
+      "loss": 3.9582,
+      "step": 6769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.297125897546215e-06,
+      "loss": 3.9195,
+      "step": 6770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.29278757401212e-06,
+      "loss": 3.8629,
+      "step": 6771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.288453154111081e-06,
+      "loss": 3.8401,
+      "step": 6772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2841226385882432e-06,
+      "loss": 4.0229,
+      "step": 6773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2797960281880665e-06,
+      "loss": 3.9991,
+      "step": 6774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2754733236543507e-06,
+      "loss": 3.9062,
+      "step": 6775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2711545257302153e-06,
+      "loss": 3.8502,
+      "step": 6776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2668396351581135e-06,
+      "loss": 3.9413,
+      "step": 6777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.262528652679824e-06,
+      "loss": 3.8511,
+      "step": 6778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.258221579036454e-06,
+      "loss": 4.0381,
+      "step": 6779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2539184149684338e-06,
+      "loss": 3.8676,
+      "step": 6780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2496191612155356e-06,
+      "loss": 3.9564,
+      "step": 6781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2453238185168507e-06,
+      "loss": 3.9132,
+      "step": 6782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2410323876107977e-06,
+      "loss": 3.8786,
+      "step": 6783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2367448692351217e-06,
+      "loss": 3.8414,
+      "step": 6784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2324612641268977e-06,
+      "loss": 3.7843,
+      "step": 6785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2281815730225252e-06,
+      "loss": 4.0126,
+      "step": 6786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.22390579665773e-06,
+      "loss": 3.9064,
+      "step": 6787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.219633935767576e-06,
+      "loss": 3.9621,
+      "step": 6788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.215365991086443e-06,
+      "loss": 3.8643,
+      "step": 6789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2111019633480305e-06,
+      "loss": 3.9769,
+      "step": 6790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.206841853285388e-06,
+      "loss": 3.8491,
+      "step": 6791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.202585661630871e-06,
+      "loss": 3.7593,
+      "step": 6792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1983333891161723e-06,
+      "loss": 3.9124,
+      "step": 6793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1940850364723016e-06,
+      "loss": 3.9699,
+      "step": 6794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1898406044295966e-06,
+      "loss": 3.8202,
+      "step": 6795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.185600093717735e-06,
+      "loss": 4.0394,
+      "step": 6796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.181363505065703e-06,
+      "loss": 3.8408,
+      "step": 6797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1771308392018213e-06,
+      "loss": 3.874,
+      "step": 6798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1729020968537296e-06,
+      "loss": 3.89,
+      "step": 6799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.168677278748407e-06,
+      "loss": 3.9233,
+      "step": 6800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.164456385612143e-06,
+      "loss": 3.8842,
+      "step": 6801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1602394181705566e-06,
+      "loss": 3.8797,
+      "step": 6802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1560263771485984e-06,
+      "loss": 3.9786,
+      "step": 6803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1518172632705338e-06,
+      "loss": 3.9951,
+      "step": 6804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1476120772599613e-06,
+      "loss": 3.8811,
+      "step": 6805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1434108198398027e-06,
+      "loss": 3.9729,
+      "step": 6806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1392134917322974e-06,
+      "loss": 3.7738,
+      "step": 6807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.135020093659015e-06,
+      "loss": 3.8534,
+      "step": 6808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1308306263408574e-06,
+      "loss": 3.8848,
+      "step": 6809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1266450904980333e-06,
+      "loss": 3.8708,
+      "step": 6810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1224634868500902e-06,
+      "loss": 3.8469,
+      "step": 6811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1182858161158948e-06,
+      "loss": 3.9786,
+      "step": 6812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.114112079013636e-06,
+      "loss": 3.9529,
+      "step": 6813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1099422762608264e-06,
+      "loss": 3.9611,
+      "step": 6814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.105776408574303e-06,
+      "loss": 3.8036,
+      "step": 6815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1016144766702215e-06,
+      "loss": 3.951,
+      "step": 6816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.097456481264079e-06,
+      "loss": 3.8675,
+      "step": 6817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0933024230706725e-06,
+      "loss": 3.9386,
+      "step": 6818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.08915230280414e-06,
+      "loss": 3.848,
+      "step": 6819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.085006121177932e-06,
+      "loss": 3.9239,
+      "step": 6820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0808638789048263e-06,
+      "loss": 3.7964,
+      "step": 6821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0767255766969196e-06,
+      "loss": 3.7968,
+      "step": 6822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.072591215265632e-06,
+      "loss": 3.8528,
+      "step": 6823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0684607953217166e-06,
+      "loss": 4.0395,
+      "step": 6824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0643343175752367e-06,
+      "loss": 3.9256,
+      "step": 6825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0602117827355773e-06,
+      "loss": 3.948,
+      "step": 6826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0560931915114517e-06,
+      "loss": 3.9658,
+      "step": 6827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0519785446108997e-06,
+      "loss": 3.919,
+      "step": 6828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.047867842741272e-06,
+      "loss": 4.0258,
+      "step": 6829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0437610866092444e-06,
+      "loss": 3.9861,
+      "step": 6830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0396582769208137e-06,
+      "loss": 3.9416,
+      "step": 6831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.035559414381308e-06,
+      "loss": 3.8621,
+      "step": 6832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.031464499695368e-06,
+      "loss": 3.7982,
+      "step": 6833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0273735335669538e-06,
+      "loss": 3.8946,
+      "step": 6834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0232865166993493e-06,
+      "loss": 3.8911,
+      "step": 6835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0192034497951566e-06,
+      "loss": 3.8872,
+      "step": 6836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0151243335563148e-06,
+      "loss": 3.9888,
+      "step": 6837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0110491686840566e-06,
+      "loss": 3.9734,
+      "step": 6838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0069779558789608e-06,
+      "loss": 3.8692,
+      "step": 6839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.002910695840915e-06,
+      "loss": 3.8365,
+      "step": 6840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9988473892691232e-06,
+      "loss": 3.8991,
+      "step": 6841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.994788036862119e-06,
+      "loss": 3.8719,
+      "step": 6842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9907326393177523e-06,
+      "loss": 3.9319,
+      "step": 6843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9866811973331845e-06,
+      "loss": 3.9605,
+      "step": 6844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.982633711604917e-06,
+      "loss": 4.0459,
+      "step": 6845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.978590182828749e-06,
+      "loss": 3.9644,
+      "step": 6846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9745506116998214e-06,
+      "loss": 3.8021,
+      "step": 6847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970514998912576e-06,
+      "loss": 3.8356,
+      "step": 6848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9664833451607857e-06,
+      "loss": 3.9148,
+      "step": 6849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.962455651137535e-06,
+      "loss": 3.8902,
+      "step": 6850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9584319175352306e-06,
+      "loss": 3.9366,
+      "step": 6851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9544121450455977e-06,
+      "loss": 3.8295,
+      "step": 6852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.950396334359686e-06,
+      "loss": 3.9215,
+      "step": 6853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.946384486167863e-06,
+      "loss": 3.8785,
+      "step": 6854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9423766011597982e-06,
+      "loss": 3.8519,
+      "step": 6855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9383726800245084e-06,
+      "loss": 3.8182,
+      "step": 6856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.934372723450309e-06,
+      "loss": 3.9573,
+      "step": 6857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9303767321248387e-06,
+      "loss": 3.8204,
+      "step": 6858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.926384706735049e-06,
+      "loss": 3.9405,
+      "step": 6859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9223966479672255e-06,
+      "loss": 4.0302,
+      "step": 6860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9184125565069542e-06,
+      "loss": 3.7753,
+      "step": 6861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.914432433039151e-06,
+      "loss": 3.8241,
+      "step": 6862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9104562782480436e-06,
+      "loss": 3.9709,
+      "step": 6863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9064840928171763e-06,
+      "loss": 3.8888,
+      "step": 6864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9025158774294172e-06,
+      "loss": 3.8622,
+      "step": 6865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.898551632766951e-06,
+      "loss": 3.9378,
+      "step": 6866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.894591359511269e-06,
+      "loss": 3.9436,
+      "step": 6867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906350583431958e-06,
+      "loss": 3.7543,
+      "step": 6868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8866827299428652e-06,
+      "loss": 3.9615,
+      "step": 6869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8827343749897226e-06,
+      "loss": 3.9634,
+      "step": 6870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8787899941625414e-06,
+      "loss": 3.8615,
+      "step": 6871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8748495881394046e-06,
+      "loss": 3.847,
+      "step": 6872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.870913157597709e-06,
+      "loss": 3.8468,
+      "step": 6873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8669807032141772e-06,
+      "loss": 3.8654,
+      "step": 6874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8630522256648463e-06,
+      "loss": 3.6911,
+      "step": 6875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.859127725625065e-06,
+      "loss": 3.7465,
+      "step": 6876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855207203769499e-06,
+      "loss": 3.9312,
+      "step": 6877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8512906607721342e-06,
+      "loss": 3.9333,
+      "step": 6878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8473780973062655e-06,
+      "loss": 3.8729,
+      "step": 6879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8434695140445073e-06,
+      "loss": 3.7304,
+      "step": 6880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8395649116587977e-06,
+      "loss": 3.9021,
+      "step": 6881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.835664290820377e-06,
+      "loss": 3.938,
+      "step": 6882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8317676521998034e-06,
+      "loss": 3.9981,
+      "step": 6883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8278749964669661e-06,
+      "loss": 3.8874,
+      "step": 6884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8239863242910521e-06,
+      "loss": 3.9027,
+      "step": 6885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8201016363405654e-06,
+      "loss": 3.9549,
+      "step": 6886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8162209332833358e-06,
+      "loss": 4.0155,
+      "step": 6887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8123442157864906e-06,
+      "loss": 3.808,
+      "step": 6888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8084714845164913e-06,
+      "loss": 3.9247,
+      "step": 6889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8046027401391053e-06,
+      "loss": 3.8655,
+      "step": 6890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8007379833194144e-06,
+      "loss": 3.8249,
+      "step": 6891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7968772147218066e-06,
+      "loss": 3.8821,
+      "step": 6892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7930204350100038e-06,
+      "loss": 4.0085,
+      "step": 6893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7891676448470256e-06,
+      "loss": 3.9741,
+      "step": 6894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.785318844895209e-06,
+      "loss": 3.9573,
+      "step": 6895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7814740358162136e-06,
+      "loss": 3.9272,
+      "step": 6896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7776332182710047e-06,
+      "loss": 3.9543,
+      "step": 6897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.773796392919863e-06,
+      "loss": 3.9909,
+      "step": 6898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7699635604223829e-06,
+      "loss": 3.9331,
+      "step": 6899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7661347214374707e-06,
+      "loss": 3.8804,
+      "step": 6900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.762309876623347e-06,
+      "loss": 3.8995,
+      "step": 6901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7584890266375553e-06,
+      "loss": 3.9827,
+      "step": 6902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7546721721369313e-06,
+      "loss": 4.0306,
+      "step": 6903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7508593137776503e-06,
+      "loss": 3.9854,
+      "step": 6904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7470504522151792e-06,
+      "loss": 4.0119,
+      "step": 6905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7432455881043086e-06,
+      "loss": 3.9895,
+      "step": 6906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7394447220991345e-06,
+      "loss": 4.0353,
+      "step": 6907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.735647854853073e-06,
+      "loss": 3.8317,
+      "step": 6908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7318549870188467e-06,
+      "loss": 3.8176,
+      "step": 6909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.728066119248492e-06,
+      "loss": 3.8042,
+      "step": 6910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7242812521933689e-06,
+      "loss": 3.9204,
+      "step": 6911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7205003865041341e-06,
+      "loss": 3.7962,
+      "step": 6912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7167235228307627e-06,
+      "loss": 3.7999,
+      "step": 6913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7129506618225376e-06,
+      "loss": 3.9134,
+      "step": 6914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7091818041280626e-06,
+      "loss": 3.9693,
+      "step": 6915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7054169503952415e-06,
+      "loss": 3.9023,
+      "step": 6916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7016561012713039e-06,
+      "loss": 3.9523,
+      "step": 6917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6978992574027824e-06,
+      "loss": 3.9715,
+      "step": 6918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6941464194355189e-06,
+      "loss": 3.9366,
+      "step": 6919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.690397588014664e-06,
+      "loss": 3.9423,
+      "step": 6920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.686652763784699e-06,
+      "loss": 3.8165,
+      "step": 6921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6829119473893927e-06,
+      "loss": 3.909,
+      "step": 6922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6791751394718391e-06,
+      "loss": 3.9019,
+      "step": 6923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6754423406744323e-06,
+      "loss": 3.9273,
+      "step": 6924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6717135516388928e-06,
+      "loss": 3.9138,
+      "step": 6925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6679887730062405e-06,
+      "loss": 3.847,
+      "step": 6926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6642680054168024e-06,
+      "loss": 3.968,
+      "step": 6927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6605512495102282e-06,
+      "loss": 3.881,
+      "step": 6928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.656838505925462e-06,
+      "loss": 3.8506,
+      "step": 6929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531297753007797e-06,
+      "loss": 3.951,
+      "step": 6930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6494250582737431e-06,
+      "loss": 3.8247,
+      "step": 6931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6457243554812485e-06,
+      "loss": 3.8566,
+      "step": 6932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6420276675594814e-06,
+      "loss": 4.0136,
+      "step": 6933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6383349951439476e-06,
+      "loss": 3.8994,
+      "step": 6934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.634646338869461e-06,
+      "loss": 3.9817,
+      "step": 6935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6309616993701426e-06,
+      "loss": 3.9567,
+      "step": 6936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6272810772794217e-06,
+      "loss": 3.9023,
+      "step": 6937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6236044732300421e-06,
+      "loss": 3.9363,
+      "step": 6938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6199318878540592e-06,
+      "loss": 3.9383,
+      "step": 6939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6162633217828322e-06,
+      "loss": 3.8599,
+      "step": 6940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6125987756470257e-06,
+      "loss": 3.9138,
+      "step": 6941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6089382500766192e-06,
+      "loss": 3.8723,
+      "step": 6942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.605281745700904e-06,
+      "loss": 3.8469,
+      "step": 6943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6016292631484686e-06,
+      "loss": 3.9948,
+      "step": 6944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5979808030472165e-06,
+      "loss": 3.7765,
+      "step": 6945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5943363660243655e-06,
+      "loss": 3.9023,
+      "step": 6946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5906959527064335e-06,
+      "loss": 3.8764,
+      "step": 6947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5870595637192538e-06,
+      "loss": 3.9386,
+      "step": 6948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5834271996879646e-06,
+      "loss": 3.8408,
+      "step": 6949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5797988612370029e-06,
+      "loss": 3.7758,
+      "step": 6950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5761745489901307e-06,
+      "loss": 3.8515,
+      "step": 6951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5725542635704028e-06,
+      "loss": 3.8018,
+      "step": 6952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.568938005600193e-06,
+      "loss": 3.8943,
+      "step": 6953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5653257757011763e-06,
+      "loss": 4.0168,
+      "step": 6954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.561717574494337e-06,
+      "loss": 3.9624,
+      "step": 6955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5581134025999643e-06,
+      "loss": 3.8784,
+      "step": 6956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5545132606376605e-06,
+      "loss": 3.9429,
+      "step": 6957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5509171492263303e-06,
+      "loss": 3.8587,
+      "step": 6958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5473250689841846e-06,
+      "loss": 3.937,
+      "step": 6959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5437370205287516e-06,
+      "loss": 3.988,
+      "step": 6960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5401530044768519e-06,
+      "loss": 3.8194,
+      "step": 6961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5365730214446206e-06,
+      "loss": 3.7869,
+      "step": 6962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5329970720474984e-06,
+      "loss": 3.9499,
+      "step": 6963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.52942515690023e-06,
+      "loss": 3.9345,
+      "step": 6964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5258572766168739e-06,
+      "loss": 4.0671,
+      "step": 6965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.522293431810784e-06,
+      "loss": 3.9974,
+      "step": 6966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5187336230946287e-06,
+      "loss": 3.958,
+      "step": 6967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5151778510803876e-06,
+      "loss": 3.8538,
+      "step": 6968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5116261163793333e-06,
+      "loss": 3.6903,
+      "step": 6969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5080784196020492e-06,
+      "loss": 3.7758,
+      "step": 6970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5045347613584253e-06,
+      "loss": 3.9536,
+      "step": 6971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5009951422576607e-06,
+      "loss": 3.908,
+      "step": 6972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4974595629082488e-06,
+      "loss": 3.902,
+      "step": 6973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4939280239180091e-06,
+      "loss": 3.9166,
+      "step": 6974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4904005258940423e-06,
+      "loss": 3.9532,
+      "step": 6975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.486877069442777e-06,
+      "loss": 3.8909,
+      "step": 6976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4833576551699284e-06,
+      "loss": 3.8856,
+      "step": 6977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4798422836805297e-06,
+      "loss": 3.9595,
+      "step": 6978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4763309555789111e-06,
+      "loss": 3.9483,
+      "step": 6979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4728236714687066e-06,
+      "loss": 3.9691,
+      "step": 6980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4693204319528697e-06,
+      "loss": 3.8386,
+      "step": 6981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4658212376336383e-06,
+      "loss": 3.8774,
+      "step": 6982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.46232608911257e-06,
+      "loss": 4.0523,
+      "step": 6983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.458834986990515e-06,
+      "loss": 4.0196,
+      "step": 6984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.45534793186764e-06,
+      "loss": 3.8718,
+      "step": 6985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.45186492434341e-06,
+      "loss": 3.9503,
+      "step": 6986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4483859650165938e-06,
+      "loss": 3.9629,
+      "step": 6987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4449110544852595e-06,
+      "loss": 3.9296,
+      "step": 6988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4414401933467908e-06,
+      "loss": 3.9354,
+      "step": 6989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4379733821978686e-06,
+      "loss": 3.9624,
+      "step": 6990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4345106216344773e-06,
+      "loss": 3.8285,
+      "step": 6991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4310519122519044e-06,
+      "loss": 4.0201,
+      "step": 6992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4275972546447415e-06,
+      "loss": 4.0397,
+      "step": 6993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4241466494068822e-06,
+      "loss": 3.8366,
+      "step": 6994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4207000971315275e-06,
+      "loss": 3.9048,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.417257598411187e-06,
+      "loss": 3.7825,
+      "step": 6996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4138191538376589e-06,
+      "loss": 3.9387,
+      "step": 6997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4103847640020512e-06,
+      "loss": 3.8334,
+      "step": 6998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4069544294947778e-06,
+      "loss": 3.8144,
+      "step": 6999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4035281509055532e-06,
+      "loss": 3.8828,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4001059288233891e-06,
+      "loss": 3.881,
+      "step": 7001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3966877638366126e-06,
+      "loss": 4.0889,
+      "step": 7002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3932736565328396e-06,
+      "loss": 3.8372,
+      "step": 7003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.389863607498998e-06,
+      "loss": 3.9789,
+      "step": 7004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3864576173213184e-06,
+      "loss": 4.006,
+      "step": 7005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3830556865853245e-06,
+      "loss": 3.9259,
+      "step": 7006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3796578158758482e-06,
+      "loss": 3.8232,
+      "step": 7007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3762640057770255e-06,
+      "loss": 3.9099,
+      "step": 7008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3728742568722864e-06,
+      "loss": 3.8187,
+      "step": 7009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.369488569744376e-06,
+      "loss": 3.807,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3661069449753262e-06,
+      "loss": 3.8387,
+      "step": 7011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.362729383146477e-06,
+      "loss": 3.9643,
+      "step": 7012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3593558848384785e-06,
+      "loss": 3.9678,
+      "step": 7013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.355986450631269e-06,
+      "loss": 3.7777,
+      "step": 7014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.352621081104094e-06,
+      "loss": 3.9079,
+      "step": 7015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3492597768354959e-06,
+      "loss": 3.9437,
+      "step": 7016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3459025384033264e-06,
+      "loss": 3.8641,
+      "step": 7017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3425493663847348e-06,
+      "loss": 3.8373,
+      "step": 7018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3392002613561662e-06,
+      "loss": 3.9177,
+      "step": 7019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.335855223893373e-06,
+      "loss": 3.8848,
+      "step": 7020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3325142545714014e-06,
+      "loss": 3.8805,
+      "step": 7021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3291773539646113e-06,
+      "loss": 3.7515,
+      "step": 7022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3258445226466466e-06,
+      "loss": 3.7867,
+      "step": 7023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3225157611904627e-06,
+      "loss": 4.0778,
+      "step": 7024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.319191070168313e-06,
+      "loss": 3.9377,
+      "step": 7025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3158704501517516e-06,
+      "loss": 3.981,
+      "step": 7026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.31255390171163e-06,
+      "loss": 3.9611,
+      "step": 7027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3092414254181007e-06,
+      "loss": 3.861,
+      "step": 7028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3059330218406164e-06,
+      "loss": 3.9327,
+      "step": 7029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3026286915479274e-06,
+      "loss": 3.9048,
+      "step": 7030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2993284351080908e-06,
+      "loss": 3.8539,
+      "step": 7031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2960322530884611e-06,
+      "loss": 3.8378,
+      "step": 7032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2927401460556876e-06,
+      "loss": 3.8127,
+      "step": 7033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2894521145757205e-06,
+      "loss": 3.9781,
+      "step": 7034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2861681592138103e-06,
+      "loss": 3.7505,
+      "step": 7035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2828882805345111e-06,
+      "loss": 3.9471,
+      "step": 7036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2796124791016605e-06,
+      "loss": 3.7917,
+      "step": 7037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2763407554784223e-06,
+      "loss": 3.8569,
+      "step": 7038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2730731102272352e-06,
+      "loss": 3.9376,
+      "step": 7039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2698095439098446e-06,
+      "loss": 3.9188,
+      "step": 7040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2665500570872985e-06,
+      "loss": 3.8557,
+      "step": 7041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2632946503199406e-06,
+      "loss": 3.905,
+      "step": 7042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.260043324167412e-06,
+      "loss": 4.02,
+      "step": 7043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2567960791886519e-06,
+      "loss": 3.9981,
+      "step": 7044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2535529159418968e-06,
+      "loss": 3.93,
+      "step": 7045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2503138349846927e-06,
+      "loss": 3.9408,
+      "step": 7046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2470788368738718e-06,
+      "loss": 3.7428,
+      "step": 7047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.243847922165564e-06,
+      "loss": 3.8944,
+      "step": 7048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2406210914152005e-06,
+      "loss": 3.8416,
+      "step": 7049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2373983451775178e-06,
+      "loss": 3.7388,
+      "step": 7050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2341796840065367e-06,
+      "loss": 3.9478,
+      "step": 7051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.230965108455584e-06,
+      "loss": 4.006,
+      "step": 7052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.227754619077287e-06,
+      "loss": 3.9345,
+      "step": 7053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2245482164235627e-06,
+      "loss": 3.9181,
+      "step": 7054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2213459010456284e-06,
+      "loss": 3.7974,
+      "step": 7055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2181476734939967e-06,
+      "loss": 3.9498,
+      "step": 7056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2149535343184859e-06,
+      "loss": 3.8433,
+      "step": 7057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2117634840681985e-06,
+      "loss": 3.9293,
+      "step": 7058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2085775232915486e-06,
+      "loss": 4.02,
+      "step": 7059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2053956525362314e-06,
+      "loss": 3.8964,
+      "step": 7060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2022178723492566e-06,
+      "loss": 3.9415,
+      "step": 7061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1990441832769179e-06,
+      "loss": 3.9592,
+      "step": 7062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.195874585864809e-06,
+      "loss": 3.8768,
+      "step": 7063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1927090806578195e-06,
+      "loss": 3.8663,
+      "step": 7064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.189547668200136e-06,
+      "loss": 3.7983,
+      "step": 7065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1863903490352379e-06,
+      "loss": 4.0385,
+      "step": 7066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1832371237059158e-06,
+      "loss": 3.9445,
+      "step": 7067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1800879927542335e-06,
+      "loss": 3.8416,
+      "step": 7068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1769429567215773e-06,
+      "loss": 3.8475,
+      "step": 7069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1738020161486035e-06,
+      "loss": 3.8372,
+      "step": 7070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1706651715752803e-06,
+      "loss": 3.8883,
+      "step": 7071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1675324235408708e-06,
+      "loss": 3.8028,
+      "step": 7072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.164403772583919e-06,
+      "loss": 3.937,
+      "step": 7073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1612792192422922e-06,
+      "loss": 4.0065,
+      "step": 7074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.158158764053127e-06,
+      "loss": 3.986,
+      "step": 7075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1550424075528698e-06,
+      "loss": 3.9182,
+      "step": 7076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1519301502772555e-06,
+      "loss": 4.0117,
+      "step": 7077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1488219927613203e-06,
+      "loss": 3.9314,
+      "step": 7078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1457179355393921e-06,
+      "loss": 3.8372,
+      "step": 7079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1426179791450914e-06,
+      "loss": 3.8294,
+      "step": 7080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1395221241113364e-06,
+      "loss": 3.8788,
+      "step": 7081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1364303709703483e-06,
+      "loss": 4.0469,
+      "step": 7082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1333427202536273e-06,
+      "loss": 3.7715,
+      "step": 7083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1302591724919793e-06,
+      "loss": 3.9263,
+      "step": 7084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1271797282154995e-06,
+      "loss": 3.8519,
+      "step": 7085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1241043879535811e-06,
+      "loss": 3.8941,
+      "step": 7086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1210331522349126e-06,
+      "loss": 3.9354,
+      "step": 7087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1179660215874716e-06,
+      "loss": 3.817,
+      "step": 7088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1149029965385416e-06,
+      "loss": 3.7832,
+      "step": 7089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.111844077614682e-06,
+      "loss": 3.857,
+      "step": 7090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1087892653417642e-06,
+      "loss": 3.8658,
+      "step": 7091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1057385602449399e-06,
+      "loss": 3.8204,
+      "step": 7092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1026919628486649e-06,
+      "loss": 3.915,
+      "step": 7093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0996494736766783e-06,
+      "loss": 3.8663,
+      "step": 7094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0966110932520284e-06,
+      "loss": 3.8122,
+      "step": 7095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0935768220970395e-06,
+      "loss": 3.848,
+      "step": 7096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0905466607333463e-06,
+      "loss": 3.949,
+      "step": 7097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0875206096818608e-06,
+      "loss": 3.8045,
+      "step": 7098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0844986694628024e-06,
+      "loss": 3.9244,
+      "step": 7099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.081480840595675e-06,
+      "loss": 3.8672,
+      "step": 7100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0784671235992777e-06,
+      "loss": 3.9483,
+      "step": 7101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0754575189917015e-06,
+      "loss": 3.9342,
+      "step": 7102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0724520272903382e-06,
+      "loss": 3.9987,
+      "step": 7103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0694506490118632e-06,
+      "loss": 3.8735,
+      "step": 7104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0664533846722447e-06,
+      "loss": 3.7964,
+      "step": 7105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0634602347867534e-06,
+      "loss": 3.8231,
+      "step": 7106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0604711998699446e-06,
+      "loss": 3.8285,
+      "step": 7107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0574862804356683e-06,
+      "loss": 3.9455,
+      "step": 7108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0545054769970613e-06,
+      "loss": 4.0767,
+      "step": 7109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0515287900665667e-06,
+      "loss": 3.8958,
+      "step": 7110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0485562201559079e-06,
+      "loss": 3.9104,
+      "step": 7111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0455877677761045e-06,
+      "loss": 3.8434,
+      "step": 7112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0426234334374645e-06,
+      "loss": 3.9043,
+      "step": 7113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0396632176495947e-06,
+      "loss": 3.8497,
+      "step": 7114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0367071209213903e-06,
+      "loss": 3.9975,
+      "step": 7115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0337551437610366e-06,
+      "loss": 3.9728,
+      "step": 7116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0308072866760137e-06,
+      "loss": 3.9647,
+      "step": 7117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0278635501730972e-06,
+      "loss": 3.9525,
+      "step": 7118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0249239347583428e-06,
+      "loss": 3.8987,
+      "step": 7119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0219884409371078e-06,
+      "loss": 3.8867,
+      "step": 7120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0190570692140356e-06,
+      "loss": 3.964,
+      "step": 7121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0161298200930646e-06,
+      "loss": 3.9412,
+      "step": 7122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0132066940774203e-06,
+      "loss": 3.8217,
+      "step": 7123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.010287691669623e-06,
+      "loss": 3.8811,
+      "step": 7124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0073728133714877e-06,
+      "loss": 3.9262,
+      "step": 7125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0044620596841138e-06,
+      "loss": 3.8087,
+      "step": 7126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0015554311078894e-06,
+      "loss": 4.001,
+      "step": 7127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.986529281425016e-07,
+      "loss": 3.9666,
+      "step": 7128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.957545512869231e-07,
+      "loss": 3.8803,
+      "step": 7129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.928603010394139e-07,
+      "loss": 3.9343,
+      "step": 7130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.899701778975396e-07,
+      "loss": 3.7865,
+      "step": 7131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.870841823581362e-07,
+      "loss": 3.8169,
+      "step": 7132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.84202314917343e-07,
+      "loss": 3.9661,
+      "step": 7133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.813245760705886e-07,
+      "loss": 3.9516,
+      "step": 7134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.784509663125884e-07,
+      "loss": 4.0017,
+      "step": 7135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.755814861373502e-07,
+      "loss": 3.7825,
+      "step": 7136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.72716136038168e-07,
+      "loss": 3.976,
+      "step": 7137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.69854916507626e-07,
+      "loss": 3.9212,
+      "step": 7138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.669978280376108e-07,
+      "loss": 3.8287,
+      "step": 7139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.641448711192796e-07,
+      "loss": 3.8015,
+      "step": 7140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.61296046243096e-07,
+      "loss": 3.782,
+      "step": 7141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.584513538987984e-07,
+      "loss": 3.9095,
+      "step": 7142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.556107945754316e-07,
+      "loss": 3.8907,
+      "step": 7143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.527743687613138e-07,
+      "loss": 3.9667,
+      "step": 7144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.499420769440576e-07,
+      "loss": 3.8261,
+      "step": 7145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.471139196105733e-07,
+      "loss": 4.0266,
+      "step": 7146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.442898972470526e-07,
+      "loss": 3.8105,
+      "step": 7147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.414700103389767e-07,
+      "loss": 3.8972,
+      "step": 7148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.386542593711162e-07,
+      "loss": 3.9215,
+      "step": 7149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.35842644827531e-07,
+      "loss": 3.7623,
+      "step": 7150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.330351671915677e-07,
+      "loss": 3.9639,
+      "step": 7151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.302318269458682e-07,
+      "loss": 3.8076,
+      "step": 7152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.274326245723608e-07,
+      "loss": 3.8446,
+      "step": 7153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.246375605522578e-07,
+      "loss": 4.0279,
+      "step": 7154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.218466353660638e-07,
+      "loss": 3.8858,
+      "step": 7155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.190598494935726e-07,
+      "loss": 3.7554,
+      "step": 7156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.162772034138622e-07,
+      "loss": 3.9297,
+      "step": 7157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.134986976053028e-07,
+      "loss": 3.8246,
+      "step": 7158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.107243325455511e-07,
+      "loss": 3.8747,
+      "step": 7159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.079541087115507e-07,
+      "loss": 3.9523,
+      "step": 7160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.051880265795426e-07,
+      "loss": 3.8949,
+      "step": 7161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.024260866250439e-07,
+      "loss": 3.9663,
+      "step": 7162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.996682893228609e-07,
+      "loss": 3.78,
+      "step": 7163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.969146351470975e-07,
+      "loss": 3.9307,
+      "step": 7164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.941651245711335e-07,
+      "loss": 3.8137,
+      "step": 7165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.914197580676409e-07,
+      "loss": 3.8707,
+      "step": 7166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.886785361085865e-07,
+      "loss": 3.8471,
+      "step": 7167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.859414591652126e-07,
+      "loss": 3.7841,
+      "step": 7168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.832085277080571e-07,
+      "loss": 3.961,
+      "step": 7169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.804797422069383e-07,
+      "loss": 3.9535,
+      "step": 7170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.777551031309728e-07,
+      "loss": 3.7297,
+      "step": 7171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.750346109485525e-07,
+      "loss": 3.7958,
+      "step": 7172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.723182661273615e-07,
+      "loss": 3.9753,
+      "step": 7173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.696060691343766e-07,
+      "loss": 3.8972,
+      "step": 7174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.668980204358495e-07,
+      "loss": 4.1029,
+      "step": 7175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.641941204973275e-07,
+      "loss": 3.8595,
+      "step": 7176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.614943697836442e-07,
+      "loss": 3.8939,
+      "step": 7177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.587987687589172e-07,
+      "loss": 3.9838,
+      "step": 7178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.561073178865453e-07,
+      "loss": 3.9527,
+      "step": 7179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.534200176292306e-07,
+      "loss": 3.9852,
+      "step": 7180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.507368684489397e-07,
+      "loss": 3.8757,
+      "step": 7181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.480578708069509e-07,
+      "loss": 3.9719,
+      "step": 7182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.453830251638039e-07,
+      "loss": 4.0356,
+      "step": 7183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.427123319793395e-07,
+      "loss": 4.0213,
+      "step": 7184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.40045791712682e-07,
+      "loss": 3.8476,
+      "step": 7185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.373834048222395e-07,
+      "loss": 3.9199,
+      "step": 7186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.347251717657018e-07,
+      "loss": 4.0114,
+      "step": 7187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.320710930000586e-07,
+      "loss": 3.8714,
+      "step": 7188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.294211689815729e-07,
+      "loss": 3.928,
+      "step": 7189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.267754001657968e-07,
+      "loss": 3.8006,
+      "step": 7190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.241337870075722e-07,
+      "loss": 3.9859,
+      "step": 7191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.21496329961019e-07,
+      "loss": 3.9887,
+      "step": 7192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.188630294795469e-07,
+      "loss": 3.8583,
+      "step": 7193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.162338860158519e-07,
+      "loss": 3.8305,
+      "step": 7194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.136089000219144e-07,
+      "loss": 3.9239,
+      "step": 7195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.10988071949001e-07,
+      "loss": 3.9165,
+      "step": 7196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.083714022476569e-07,
+      "loss": 3.8727,
+      "step": 7197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.057588913677277e-07,
+      "loss": 3.7768,
+      "step": 7198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.031505397583266e-07,
+      "loss": 3.8965,
+      "step": 7199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.005463478678616e-07,
+      "loss": 3.8315,
+      "step": 7200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.979463161440243e-07,
+      "loss": 3.7676,
+      "step": 7201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.953504450337879e-07,
+      "loss": 3.9079,
+      "step": 7202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.927587349834148e-07,
+      "loss": 3.9457,
+      "step": 7203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.901711864384514e-07,
+      "loss": 3.799,
+      "step": 7204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.875877998437225e-07,
+      "loss": 3.9081,
+      "step": 7205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.850085756433479e-07,
+      "loss": 3.8892,
+      "step": 7206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.824335142807199e-07,
+      "loss": 3.8992,
+      "step": 7207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.798626161985289e-07,
+      "loss": 3.9535,
+      "step": 7208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.772958818387327e-07,
+      "loss": 3.9611,
+      "step": 7209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.747333116425948e-07,
+      "loss": 3.9359,
+      "step": 7210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.721749060506406e-07,
+      "loss": 3.8043,
+      "step": 7211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.696206655026933e-07,
+      "loss": 3.9883,
+      "step": 7212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.670705904378572e-07,
+      "loss": 3.8336,
+      "step": 7213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.645246812945206e-07,
+      "loss": 3.9173,
+      "step": 7214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.619829385103472e-07,
+      "loss": 3.8423,
+      "step": 7215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.594453625223014e-07,
+      "loss": 3.7612,
+      "step": 7216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.569119537666175e-07,
+      "loss": 3.8456,
+      "step": 7217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.543827126788194e-07,
+      "loss": 3.8553,
+      "step": 7218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.518576396937122e-07,
+      "loss": 3.7555,
+      "step": 7219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.493367352453873e-07,
+      "loss": 3.8914,
+      "step": 7220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.468199997672148e-07,
+      "loss": 3.8599,
+      "step": 7221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.443074336918488e-07,
+      "loss": 3.9899,
+      "step": 7222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.417990374512296e-07,
+      "loss": 3.9377,
+      "step": 7223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.392948114765847e-07,
+      "loss": 3.8678,
+      "step": 7224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.36794756198414e-07,
+      "loss": 3.9349,
+      "step": 7225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.3429887204651e-07,
+      "loss": 3.8585,
+      "step": 7226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.318071594499404e-07,
+      "loss": 3.9784,
+      "step": 7227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.293196188370627e-07,
+      "loss": 3.8847,
+      "step": 7228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.268362506355125e-07,
+      "loss": 3.9851,
+      "step": 7229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.243570552722068e-07,
+      "loss": 3.8187,
+      "step": 7230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.218820331733517e-07,
+      "loss": 3.8707,
+      "step": 7231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.194111847644347e-07,
+      "loss": 3.8412,
+      "step": 7232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.16944510470216e-07,
+      "loss": 3.8375,
+      "step": 7233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.144820107147482e-07,
+      "loss": 3.9587,
+      "step": 7234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.120236859213675e-07,
+      "loss": 3.9208,
+      "step": 7235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.095695365126858e-07,
+      "loss": 3.8292,
+      "step": 7236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.071195629105993e-07,
+      "loss": 3.8742,
+      "step": 7237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.046737655362845e-07,
+      "loss": 4.0122,
+      "step": 7238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.022321448102054e-07,
+      "loss": 3.8898,
+      "step": 7239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.997947011521067e-07,
+      "loss": 3.9183,
+      "step": 7240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.973614349810115e-07,
+      "loss": 3.7678,
+      "step": 7241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.949323467152269e-07,
+      "loss": 3.8086,
+      "step": 7242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.925074367723383e-07,
+      "loss": 3.8701,
+      "step": 7243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.900867055692151e-07,
+      "loss": 3.916,
+      "step": 7244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.876701535220131e-07,
+      "loss": 3.7453,
+      "step": 7245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.852577810461669e-07,
+      "loss": 3.9998,
+      "step": 7246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.82849588556389e-07,
+      "loss": 3.8347,
+      "step": 7247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.804455764666734e-07,
+      "loss": 4.0412,
+      "step": 7248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.780457451903032e-07,
+      "loss": 3.7639,
+      "step": 7249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.75650095139832e-07,
+      "loss": 3.9686,
+      "step": 7250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.732586267270968e-07,
+      "loss": 3.998,
+      "step": 7251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.7087134036323e-07,
+      "loss": 3.9428,
+      "step": 7252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.684882364586225e-07,
+      "loss": 3.9508,
+      "step": 7253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.661093154229637e-07,
+      "loss": 3.8494,
+      "step": 7254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.637345776652176e-07,
+      "loss": 3.9747,
+      "step": 7255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.613640235936274e-07,
+      "loss": 4.0649,
+      "step": 7256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.589976536157199e-07,
+      "loss": 3.8327,
+      "step": 7257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.566354681383003e-07,
+      "loss": 3.9527,
+      "step": 7258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.542774675674546e-07,
+      "loss": 3.7638,
+      "step": 7259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.519236523085531e-07,
+      "loss": 3.958,
+      "step": 7260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.495740227662444e-07,
+      "loss": 3.9502,
+      "step": 7261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.472285793444499e-07,
+      "loss": 3.8332,
+      "step": 7262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.448873224463914e-07,
+      "loss": 3.8939,
+      "step": 7263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.425502524745469e-07,
+      "loss": 3.8129,
+      "step": 7264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.402173698306923e-07,
+      "loss": 4.012,
+      "step": 7265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.378886749158707e-07,
+      "loss": 3.8228,
+      "step": 7266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.355641681304175e-07,
+      "loss": 4.0094,
+      "step": 7267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.332438498739435e-07,
+      "loss": 3.9384,
+      "step": 7268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.309277205453323e-07,
+      "loss": 3.867,
+      "step": 7269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.2861578054276e-07,
+      "loss": 3.9248,
+      "step": 7270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.26308030263667e-07,
+      "loss": 3.8699,
+      "step": 7271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.240044701047914e-07,
+      "loss": 3.9587,
+      "step": 7272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.217051004621361e-07,
+      "loss": 3.9939,
+      "step": 7273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.194099217309901e-07,
+      "loss": 3.8468,
+      "step": 7274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.17118934305927e-07,
+      "loss": 3.9703,
+      "step": 7275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.14832138580787e-07,
+      "loss": 4.0079,
+      "step": 7276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.125495349487004e-07,
+      "loss": 3.8875,
+      "step": 7277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.102711238020725e-07,
+      "loss": 3.7578,
+      "step": 7278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.079969055325901e-07,
+      "loss": 3.8723,
+      "step": 7279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.057268805312128e-07,
+      "loss": 3.7668,
+      "step": 7280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.034610491881864e-07,
+      "loss": 3.929,
+      "step": 7281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.011994118930359e-07,
+      "loss": 3.8725,
+      "step": 7282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.989419690345638e-07,
+      "loss": 3.8916,
+      "step": 7283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.966887210008487e-07,
+      "loss": 3.8525,
+      "step": 7284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.944396681792475e-07,
+      "loss": 3.9167,
+      "step": 7285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.921948109564035e-07,
+      "loss": 3.9547,
+      "step": 7286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.899541497182276e-07,
+      "loss": 3.8231,
+      "step": 7287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.877176848499199e-07,
+      "loss": 3.9106,
+      "step": 7288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.854854167359564e-07,
+      "loss": 3.9423,
+      "step": 7289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.832573457600826e-07,
+      "loss": 3.8643,
+      "step": 7290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.810334723053367e-07,
+      "loss": 3.9538,
+      "step": 7291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.78813796754027e-07,
+      "loss": 3.8052,
+      "step": 7292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.765983194877394e-07,
+      "loss": 3.8753,
+      "step": 7293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.743870408873419e-07,
+      "loss": 3.8515,
+      "step": 7294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.721799613329748e-07,
+      "loss": 3.7696,
+      "step": 7295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.699770812040678e-07,
+      "loss": 3.9292,
+      "step": 7296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.677784008793153e-07,
+      "loss": 3.9468,
+      "step": 7297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.655839207367008e-07,
+      "loss": 3.866,
+      "step": 7298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.633936411534729e-07,
+      "loss": 3.8073,
+      "step": 7299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.612075625061775e-07,
+      "loss": 3.7732,
+      "step": 7300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.590256851706194e-07,
+      "loss": 3.9482,
+      "step": 7301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.568480095218876e-07,
+      "loss": 3.8873,
+      "step": 7302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.546745359343546e-07,
+      "loss": 3.8466,
+      "step": 7303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.525052647816636e-07,
+      "loss": 4.0889,
+      "step": 7304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.503401964367355e-07,
+      "loss": 4.0327,
+      "step": 7305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.481793312717754e-07,
+      "loss": 3.9105,
+      "step": 7306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.460226696582555e-07,
+      "loss": 3.9098,
+      "step": 7307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.43870211966932e-07,
+      "loss": 3.8403,
+      "step": 7308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.417219585678424e-07,
+      "loss": 3.9855,
+      "step": 7309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.395779098302911e-07,
+      "loss": 4.0299,
+      "step": 7310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.374380661228667e-07,
+      "loss": 3.8305,
+      "step": 7311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.353024278134333e-07,
+      "loss": 3.9378,
+      "step": 7312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.331709952691332e-07,
+      "loss": 3.7858,
+      "step": 7313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.310437688563819e-07,
+      "loss": 4.0714,
+      "step": 7314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.289207489408754e-07,
+      "loss": 3.8442,
+      "step": 7315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.268019358875803e-07,
+      "loss": 3.9826,
+      "step": 7316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.246873300607553e-07,
+      "loss": 3.9004,
+      "step": 7317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.225769318239177e-07,
+      "loss": 3.8403,
+      "step": 7318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.204707415398746e-07,
+      "loss": 3.8914,
+      "step": 7319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.183687595707032e-07,
+      "loss": 3.8765,
+      "step": 7320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.162709862777559e-07,
+      "loss": 3.8569,
+      "step": 7321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.141774220216666e-07,
+      "loss": 4.015,
+      "step": 7322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.12088067162339e-07,
+      "loss": 3.8554,
+      "step": 7323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.100029220589637e-07,
+      "loss": 4.0474,
+      "step": 7324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.07921987069998e-07,
+      "loss": 3.9468,
+      "step": 7325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.058452625531812e-07,
+      "loss": 3.9349,
+      "step": 7326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.037727488655192e-07,
+      "loss": 4.0214,
+      "step": 7327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.017044463633103e-07,
+      "loss": 3.9505,
+      "step": 7328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.996403554021145e-07,
+      "loss": 3.89,
+      "step": 7329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.975804763367759e-07,
+      "loss": 3.9582,
+      "step": 7330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.955248095214082e-07,
+      "loss": 3.872,
+      "step": 7331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.934733553094068e-07,
+      "loss": 3.9649,
+      "step": 7332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.914261140534393e-07,
+      "loss": 3.9795,
+      "step": 7333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.893830861054493e-07,
+      "loss": 4.0352,
+      "step": 7334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.873442718166616e-07,
+      "loss": 3.8025,
+      "step": 7335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.853096715375649e-07,
+      "loss": 3.9189,
+      "step": 7336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.832792856179352e-07,
+      "loss": 4.0754,
+      "step": 7337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.812531144068183e-07,
+      "loss": 3.886,
+      "step": 7338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.792311582525383e-07,
+      "loss": 3.8867,
+      "step": 7339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.772134175026921e-07,
+      "loss": 3.9256,
+      "step": 7340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.751998925041495e-07,
+      "loss": 3.8195,
+      "step": 7341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7319058360306423e-07,
+      "loss": 3.8706,
+      "step": 7342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.711854911448543e-07,
+      "loss": 3.9434,
+      "step": 7343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.691846154742191e-07,
+      "loss": 3.9907,
+      "step": 7344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6718795693513617e-07,
+      "loss": 3.7988,
+      "step": 7345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.651955158708532e-07,
+      "loss": 3.8482,
+      "step": 7346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.632072926238934e-07,
+      "loss": 3.916,
+      "step": 7347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.612232875360528e-07,
+      "loss": 3.8023,
+      "step": 7348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.592435009484086e-07,
+      "loss": 4.0656,
+      "step": 7349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.57267933201308e-07,
+      "loss": 4.0697,
+      "step": 7350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.55296584634371e-07,
+      "loss": 3.9895,
+      "step": 7351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5332945558649867e-07,
+      "loss": 4.0113,
+      "step": 7352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.513665463958622e-07,
+      "loss": 3.921,
+      "step": 7353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.494078573999111e-07,
+      "loss": 3.9389,
+      "step": 7354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.474533889353594e-07,
+      "loss": 3.8323,
+      "step": 7355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.455031413382105e-07,
+      "loss": 3.913,
+      "step": 7356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.435571149437323e-07,
+      "loss": 3.9462,
+      "step": 7357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.416153100864684e-07,
+      "loss": 3.9225,
+      "step": 7358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.396777271002378e-07,
+      "loss": 3.8649,
+      "step": 7359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.377443663181324e-07,
+      "loss": 4.0757,
+      "step": 7360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3581522807252253e-07,
+      "loss": 3.9291,
+      "step": 7361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3389031269504834e-07,
+      "loss": 3.8336,
+      "step": 7362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3196962051662013e-07,
+      "loss": 3.8041,
+      "step": 7363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.300531518674322e-07,
+      "loss": 3.8824,
+      "step": 7364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2814090707694874e-07,
+      "loss": 3.9387,
+      "step": 7365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2623288647390404e-07,
+      "loss": 3.9344,
+      "step": 7366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.243290903863079e-07,
+      "loss": 3.8433,
+      "step": 7367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2242951914145123e-07,
+      "loss": 3.9096,
+      "step": 7368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2053417306588404e-07,
+      "loss": 3.9173,
+      "step": 7369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.186430524854429e-07,
+      "loss": 3.9334,
+      "step": 7370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1675615772523434e-07,
+      "loss": 3.9252,
+      "step": 7371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.148734891096351e-07,
+      "loss": 3.9455,
+      "step": 7372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.129950469622945e-07,
+      "loss": 4.0527,
+      "step": 7373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.111208316061432e-07,
+      "loss": 3.9286,
+      "step": 7374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.092508433633846e-07,
+      "loss": 3.79,
+      "step": 7375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.073850825554837e-07,
+      "loss": 3.9506,
+      "step": 7376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.055235495031895e-07,
+      "loss": 3.768,
+      "step": 7377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.03666244526521e-07,
+      "loss": 3.99,
+      "step": 7378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.018131679447701e-07,
+      "loss": 3.8339,
+      "step": 7379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9996432007649855e-07,
+      "loss": 4.0372,
+      "step": 7380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9811970123955214e-07,
+      "loss": 3.8994,
+      "step": 7381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9627931175103827e-07,
+      "loss": 3.8827,
+      "step": 7382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9444315192734006e-07,
+      "loss": 3.9418,
+      "step": 7383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9261122208411885e-07,
+      "loss": 3.8904,
+      "step": 7384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9078352253630047e-07,
+      "loss": 3.7947,
+      "step": 7385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8896005359808916e-07,
+      "loss": 3.9814,
+      "step": 7386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.871408155829592e-07,
+      "loss": 3.923,
+      "step": 7387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.853258088036604e-07,
+      "loss": 4.0586,
+      "step": 7388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.835150335722154e-07,
+      "loss": 3.7674,
+      "step": 7389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.817084901999113e-07,
+      "loss": 3.9135,
+      "step": 7390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7990617899731906e-07,
+      "loss": 3.8629,
+      "step": 7391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.781081002742742e-07,
+      "loss": 3.9758,
+      "step": 7392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7631425433989064e-07,
+      "loss": 3.9208,
+      "step": 7393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.745246415025466e-07,
+      "loss": 4.0307,
+      "step": 7394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.727392620699016e-07,
+      "loss": 3.7739,
+      "step": 7395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7095811634887956e-07,
+      "loss": 3.8212,
+      "step": 7396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6918120464568273e-07,
+      "loss": 3.7894,
+      "step": 7397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6740852726578066e-07,
+      "loss": 3.8711,
+      "step": 7398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.656400845139213e-07,
+      "loss": 3.884,
+      "step": 7399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6387587669411414e-07,
+      "loss": 3.8769,
+      "step": 7400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6211590410965004e-07,
+      "loss": 3.9566,
+      "step": 7401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.603601670630896e-07,
+      "loss": 3.9514,
+      "step": 7402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5860866585626373e-07,
+      "loss": 3.9879,
+      "step": 7403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5686140079027595e-07,
+      "loss": 3.8706,
+      "step": 7404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.551183721655027e-07,
+      "loss": 3.8787,
+      "step": 7405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.533795802815876e-07,
+      "loss": 3.8064,
+      "step": 7406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.516450254374526e-07,
+      "loss": 3.7525,
+      "step": 7407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.499147079312842e-07,
+      "loss": 3.8951,
+      "step": 7408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4818862806054453e-07,
+      "loss": 3.8747,
+      "step": 7409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4646678612196834e-07,
+      "loss": 3.9375,
+      "step": 7410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.447491824115606e-07,
+      "loss": 3.9967,
+      "step": 7411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.43035817224599e-07,
+      "loss": 4.0224,
+      "step": 7412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4132669085562563e-07,
+      "loss": 3.9911,
+      "step": 7413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3962180359846387e-07,
+      "loss": 3.8775,
+      "step": 7414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.379211557462014e-07,
+      "loss": 3.9359,
+      "step": 7415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.362247475911989e-07,
+      "loss": 3.8293,
+      "step": 7416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3453257942508973e-07,
+      "loss": 3.9758,
+      "step": 7417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.328446515387773e-07,
+      "loss": 3.8843,
+      "step": 7418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3116096422243224e-07,
+      "loss": 3.8161,
+      "step": 7419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.294815177655064e-07,
+      "loss": 3.7834,
+      "step": 7420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2780631245671046e-07,
+      "loss": 3.8477,
+      "step": 7421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2613534858403625e-07,
+      "loss": 4.0115,
+      "step": 7422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.244686264347374e-07,
+      "loss": 3.9292,
+      "step": 7423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2280614629534854e-07,
+      "loss": 4.0526,
+      "step": 7424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.211479084516633e-07,
+      "loss": 4.051,
+      "step": 7425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1949391318875644e-07,
+      "loss": 3.7212,
+      "step": 7426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1784416079096434e-07,
+      "loss": 3.8548,
+      "step": 7427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.161986515419019e-07,
+      "loss": 3.8882,
+      "step": 7428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1455738572444824e-07,
+      "loss": 4.0286,
+      "step": 7429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1292036362076114e-07,
+      "loss": 3.8745,
+      "step": 7430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.112875855122599e-07,
+      "loss": 3.8745,
+      "step": 7431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0965905167963936e-07,
+      "loss": 3.8918,
+      "step": 7432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.080347624028618e-07,
+      "loss": 3.7241,
+      "step": 7433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0641471796116483e-07,
+      "loss": 3.8736,
+      "step": 7434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.04798918633048e-07,
+      "loss": 3.9025,
+      "step": 7435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0318736469628906e-07,
+      "loss": 3.8796,
+      "step": 7436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.015800564279303e-07,
+      "loss": 3.8174,
+      "step": 7437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9997699410428957e-07,
+      "loss": 3.7704,
+      "step": 7438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.98378178000952e-07,
+      "loss": 4.0832,
+      "step": 7439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.967836083927727e-07,
+      "loss": 3.8638,
+      "step": 7440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9519328555387415e-07,
+      "loss": 4.0445,
+      "step": 7441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.936072097576514e-07,
+      "loss": 3.9411,
+      "step": 7442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.920253812767698e-07,
+      "loss": 3.9829,
+      "step": 7443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9044780038316454e-07,
+      "loss": 3.8282,
+      "step": 7444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.888744673480437e-07,
+      "loss": 3.8826,
+      "step": 7445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.873053824418742e-07,
+      "loss": 3.806,
+      "step": 7446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8574054593440427e-07,
+      "loss": 3.8678,
+      "step": 7447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8417995809464635e-07,
+      "loss": 3.83,
+      "step": 7448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.82623619190886e-07,
+      "loss": 3.8387,
+      "step": 7449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8107152949067295e-07,
+      "loss": 4.0065,
+      "step": 7450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.795236892608327e-07,
+      "loss": 3.817,
+      "step": 7451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.779800987674497e-07,
+      "loss": 3.858,
+      "step": 7452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7644075827589213e-07,
+      "loss": 3.8632,
+      "step": 7453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7490566805079e-07,
+      "loss": 3.9142,
+      "step": 7454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7337482835604057e-07,
+      "loss": 3.8909,
+      "step": 7455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7184823945481107e-07,
+      "loss": 3.9792,
+      "step": 7456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7032590160954716e-07,
+      "loss": 3.8623,
+      "step": 7457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6880781508194784e-07,
+      "loss": 3.8133,
+      "step": 7458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.67293980132996e-07,
+      "loss": 3.9847,
+      "step": 7459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.657843970229335e-07,
+      "loss": 3.8938,
+      "step": 7460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.642790660112776e-07,
+      "loss": 3.8276,
+      "step": 7461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6277798735681314e-07,
+      "loss": 3.9505,
+      "step": 7462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.612811613175892e-07,
+      "loss": 3.928,
+      "step": 7463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5978858815093046e-07,
+      "loss": 3.8668,
+      "step": 7464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.583002681134233e-07,
+      "loss": 3.8671,
+      "step": 7465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5681620146093265e-07,
+      "loss": 3.9173,
+      "step": 7466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5533638844858487e-07,
+      "loss": 3.8611,
+      "step": 7467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.538608293307765e-07,
+      "loss": 3.8919,
+      "step": 7468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.52389524361174e-07,
+      "loss": 4.0068,
+      "step": 7469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.509224737927113e-07,
+      "loss": 4.0145,
+      "step": 7470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.494596778775893e-07,
+      "loss": 3.832,
+      "step": 7471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4800113686728465e-07,
+      "loss": 3.9516,
+      "step": 7472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4654685101253016e-07,
+      "loss": 3.9215,
+      "step": 7473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.450968205633425e-07,
+      "loss": 3.9025,
+      "step": 7474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.436510457689917e-07,
+      "loss": 3.8668,
+      "step": 7475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4220952687802887e-07,
+      "loss": 4.031,
+      "step": 7476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4077226413826427e-07,
+      "loss": 3.9121,
+      "step": 7477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3933925779678343e-07,
+      "loss": 3.8515,
+      "step": 7478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.379105080999311e-07,
+      "loss": 3.9684,
+      "step": 7479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3648601529333015e-07,
+      "loss": 3.8193,
+      "step": 7480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3506577962186805e-07,
+      "loss": 3.9333,
+      "step": 7481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3364980132969672e-07,
+      "loss": 3.857,
+      "step": 7482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.322380806602409e-07,
+      "loss": 3.8016,
+      "step": 7483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3083061785618977e-07,
+      "loss": 3.9255,
+      "step": 7484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2942741315950533e-07,
+      "loss": 3.8736,
+      "step": 7485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2802846681141688e-07,
+      "loss": 3.8135,
+      "step": 7486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2663377905241255e-07,
+      "loss": 3.9706,
+      "step": 7487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2524335012225894e-07,
+      "loss": 3.802,
+      "step": 7488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.238571802599898e-07,
+      "loss": 3.9845,
+      "step": 7489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2247526970390064e-07,
+      "loss": 3.9303,
+      "step": 7490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2109761869155698e-07,
+      "loss": 3.8672,
+      "step": 7491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1972422745979438e-07,
+      "loss": 3.9997,
+      "step": 7492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1835509624471563e-07,
+      "loss": 4.0392,
+      "step": 7493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.16990225281688e-07,
+      "loss": 3.834,
+      "step": 7494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1562961480534882e-07,
+      "loss": 3.9918,
+      "step": 7495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1427326504960822e-07,
+      "loss": 3.88,
+      "step": 7496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1292117624763242e-07,
+      "loss": 3.8605,
+      "step": 7497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.115733486318605e-07,
+      "loss": 3.8422,
+      "step": 7498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1022978243400438e-07,
+      "loss": 3.8433,
+      "step": 7499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0889047788503756e-07,
+      "loss": 3.8527,
+      "step": 7500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0755543521519817e-07,
+      "loss": 3.8779,
+      "step": 7501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06224654653997e-07,
+      "loss": 4.0056,
+      "step": 7502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0489813643021495e-07,
+      "loss": 3.8843,
+      "step": 7503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0357588077189182e-07,
+      "loss": 3.9728,
+      "step": 7504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0225788790633748e-07,
+      "loss": 3.8437,
+      "step": 7505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0094415806013457e-07,
+      "loss": 4.1151,
+      "step": 7506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.996346914591274e-07,
+      "loss": 3.9174,
+      "step": 7507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.983294883284248e-07,
+      "loss": 3.9564,
+      "step": 7508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.970285488924084e-07,
+      "loss": 4.0676,
+      "step": 7509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.957318733747271e-07,
+      "loss": 3.9541,
+      "step": 7510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.944394619982942e-07,
+      "loss": 3.8856,
+      "step": 7511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9315131498528483e-07,
+      "loss": 3.7798,
+      "step": 7512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9186743255714957e-07,
+      "loss": 3.9078,
+      "step": 7513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.905878149346063e-07,
+      "loss": 3.9113,
+      "step": 7514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.893124623376319e-07,
+      "loss": 3.8912,
+      "step": 7515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8804137498547592e-07,
+      "loss": 3.7834,
+      "step": 7516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.867745530966497e-07,
+      "loss": 3.8407,
+      "step": 7517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8551199688894015e-07,
+      "loss": 3.8902,
+      "step": 7518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.842537065793931e-07,
+      "loss": 3.8851,
+      "step": 7519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8299968238432163e-07,
+      "loss": 3.9059,
+      "step": 7520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8174992451930606e-07,
+      "loss": 3.9404,
+      "step": 7521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.805044331991995e-07,
+      "loss": 3.9008,
+      "step": 7522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7926320863811131e-07,
+      "loss": 3.8762,
+      "step": 7523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7802625104942627e-07,
+      "loss": 3.8631,
+      "step": 7524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7679356064578822e-07,
+      "loss": 3.9311,
+      "step": 7525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7556513763911097e-07,
+      "loss": 3.8675,
+      "step": 7526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7434098224057837e-07,
+      "loss": 3.833,
+      "step": 7527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7312109466063596e-07,
+      "loss": 3.8537,
+      "step": 7528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7190547510899102e-07,
+      "loss": 3.9725,
+      "step": 7529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.706941237946291e-07,
+      "loss": 3.7896,
+      "step": 7530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6948704092579203e-07,
+      "loss": 3.9095,
+      "step": 7531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6828422670999432e-07,
+      "loss": 3.8728,
+      "step": 7532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6708568135401226e-07,
+      "loss": 3.8633,
+      "step": 7533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6589140506388934e-07,
+      "loss": 3.9162,
+      "step": 7534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6470139804493356e-07,
+      "loss": 3.988,
+      "step": 7535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6351566050172573e-07,
+      "loss": 3.7971,
+      "step": 7536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6233419263810278e-07,
+      "loss": 3.9613,
+      "step": 7537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6115699465717448e-07,
+      "loss": 3.9173,
+      "step": 7538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5998406676131783e-07,
+      "loss": 4.0073,
+      "step": 7539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5881540915216874e-07,
+      "loss": 3.7275,
+      "step": 7540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5765102203063598e-07,
+      "loss": 3.863,
+      "step": 7541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5649090559688717e-07,
+      "loss": 3.8215,
+      "step": 7542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5533506005036557e-07,
+      "loss": 4.0019,
+      "step": 7543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5418348558977058e-07,
+      "loss": 3.9161,
+      "step": 7544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5303618241306882e-07,
+      "loss": 3.8174,
+      "step": 7545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5189315071749976e-07,
+      "loss": 3.7701,
+      "step": 7546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5075439069956455e-07,
+      "loss": 3.7792,
+      "step": 7547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4961990255502324e-07,
+      "loss": 3.8325,
+      "step": 7548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4848968647891148e-07,
+      "loss": 3.8757,
+      "step": 7549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4736374266552944e-07,
+      "loss": 3.8373,
+      "step": 7550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4624207130843338e-07,
+      "loss": 3.8772,
+      "step": 7551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4512467260045514e-07,
+      "loss": 3.8759,
+      "step": 7552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4401154673368832e-07,
+      "loss": 3.9824,
+      "step": 7553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4290269389949096e-07,
+      "loss": 3.8278,
+      "step": 7554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.417981142884911e-07,
+      "loss": 3.9073,
+      "step": 7555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4069780809057575e-07,
+      "loss": 3.8046,
+      "step": 7556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3960177549489916e-07,
+      "loss": 4.0558,
+      "step": 7557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3851001668988562e-07,
+      "loss": 3.9131,
+      "step": 7558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3742253186321828e-07,
+      "loss": 3.8902,
+      "step": 7559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3633932120184767e-07,
+      "loss": 3.969,
+      "step": 7560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3526038489199422e-07,
+      "loss": 3.8648,
+      "step": 7561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3418572311913736e-07,
+      "loss": 3.9861,
+      "step": 7562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.331153360680265e-07,
+      "loss": 3.8606,
+      "step": 7563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3204922392266728e-07,
+      "loss": 3.8864,
+      "step": 7564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.309873868663436e-07,
+      "loss": 3.8827,
+      "step": 7565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2992982508159003e-07,
+      "loss": 3.8264,
+      "step": 7566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2887653875021943e-07,
+      "loss": 4.0357,
+      "step": 7567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2782752805330366e-07,
+      "loss": 3.8824,
+      "step": 7568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2678279317117903e-07,
+      "loss": 3.8515,
+      "step": 7569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2574233428344905e-07,
+      "loss": 4.054,
+      "step": 7570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2470615156897625e-07,
+      "loss": 4.0116,
+      "step": 7571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2367424520589588e-07,
+      "loss": 4.0042,
+      "step": 7572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2264661537160493e-07,
+      "loss": 3.8969,
+      "step": 7573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.216232622427621e-07,
+      "loss": 3.9466,
+      "step": 7574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.206041859952961e-07,
+      "loss": 3.8922,
+      "step": 7575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1958938680439736e-07,
+      "loss": 3.8028,
+      "step": 7576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1857886484452074e-07,
+      "loss": 3.749,
+      "step": 7577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1757262028938842e-07,
+      "loss": 4.0325,
+      "step": 7578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1657065331198425e-07,
+      "loss": 3.7782,
+      "step": 7579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1557296408455931e-07,
+      "loss": 3.9352,
+      "step": 7580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1457955277862642e-07,
+      "loss": 3.9799,
+      "step": 7581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1359041956496286e-07,
+      "loss": 3.8789,
+      "step": 7582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1260556461361593e-07,
+      "loss": 3.9247,
+      "step": 7583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1162498809389189e-07,
+      "loss": 3.8943,
+      "step": 7584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.106486901743642e-07,
+      "loss": 3.865,
+      "step": 7585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0967667102286528e-07,
+      "loss": 3.8657,
+      "step": 7586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0870893080650313e-07,
+      "loss": 3.7862,
+      "step": 7587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0774546969163913e-07,
+      "loss": 3.8338,
+      "step": 7588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0678628784390465e-07,
+      "loss": 3.9608,
+      "step": 7589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.058313854281956e-07,
+      "loss": 3.8893,
+      "step": 7590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0488076260866952e-07,
+      "loss": 3.8759,
+      "step": 7591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.039344195487485e-07,
+      "loss": 3.9532,
+      "step": 7592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0299235641111904e-07,
+      "loss": 3.9081,
+      "step": 7593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0205457335773494e-07,
+      "loss": 3.8162,
+      "step": 7594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0112107054981168e-07,
+      "loss": 3.9036,
+      "step": 7595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0019184814782923e-07,
+      "loss": 3.9478,
+      "step": 7596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.926690631153201e-08,
+      "loss": 4.0985,
+      "step": 7597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.834624519992896e-08,
+      "loss": 3.9151,
+      "step": 7598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.742986497128792e-08,
+      "loss": 3.8764,
+      "step": 7599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.65177657831523e-08,
+      "loss": 3.8717,
+      "step": 7600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.560994779231614e-08,
+      "loss": 3.9503,
+      "step": 7601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.470641115484625e-08,
+      "loss": 3.9658,
+      "step": 7602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.380715602607116e-08,
+      "loss": 3.9473,
+      "step": 7603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.291218256058386e-08,
+      "loss": 3.9654,
+      "step": 7604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.202149091223911e-08,
+      "loss": 3.6743,
+      "step": 7605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.113508123415881e-08,
+      "loss": 3.8215,
+      "step": 7606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.025295367872666e-08,
+      "loss": 3.9402,
+      "step": 7607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.937510839759078e-08,
+      "loss": 3.8534,
+      "step": 7608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.850154554166101e-08,
+      "loss": 3.8453,
+      "step": 7609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.763226526111722e-08,
+      "loss": 3.9108,
+      "step": 7610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.676726770539267e-08,
+      "loss": 3.9961,
+      "step": 7611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.590655302319617e-08,
+      "loss": 3.9537,
+      "step": 7612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.505012136249269e-08,
+      "loss": 3.9758,
+      "step": 7613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.41979728705089e-08,
+      "loss": 3.8345,
+      "step": 7614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.335010769374429e-08,
+      "loss": 4.0146,
+      "step": 7615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.25065259779545e-08,
+      "loss": 3.9455,
+      "step": 7616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.166722786816238e-08,
+      "loss": 3.9062,
+      "step": 7617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.083221350865256e-08,
+      "loss": 3.8486,
+      "step": 7618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.000148304297129e-08,
+      "loss": 3.8788,
+      "step": 7619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.91750366139321e-08,
+      "loss": 3.7701,
+      "step": 7620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.835287436361305e-08,
+      "loss": 3.9148,
+      "step": 7621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.753499643334827e-08,
+      "loss": 3.8198,
+      "step": 7622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.672140296374475e-08,
+      "loss": 3.8837,
+      "step": 7623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.591209409466837e-08,
+      "loss": 3.9286,
+      "step": 7624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.510706996524674e-08,
+      "loss": 3.9377,
+      "step": 7625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.43063307138775e-08,
+      "loss": 4.0004,
+      "step": 7626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.350987647820884e-08,
+      "loss": 3.9183,
+      "step": 7627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.271770739516737e-08,
+      "loss": 3.869,
+      "step": 7628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.192982360093581e-08,
+      "loss": 3.8769,
+      "step": 7629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.114622523095305e-08,
+      "loss": 3.7905,
+      "step": 7630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.036691241993909e-08,
+      "loss": 3.9351,
+      "step": 7631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.959188530185901e-08,
+      "loss": 3.9669,
+      "step": 7632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.882114400995343e-08,
+      "loss": 3.8672,
+      "step": 7633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.805468867672193e-08,
+      "loss": 3.7378,
+      "step": 7634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.729251943392301e-08,
+      "loss": 3.9698,
+      "step": 7635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.653463641258517e-08,
+      "loss": 3.8399,
+      "step": 7636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.578103974299588e-08,
+      "loss": 3.8838,
+      "step": 7637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.503172955470981e-08,
+      "loss": 3.9506,
+      "step": 7638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.428670597654063e-08,
+      "loss": 3.8714,
+      "step": 7639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.354596913656363e-08,
+      "loss": 3.8571,
+      "step": 7640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.280951916212419e-08,
+      "loss": 3.8716,
+      "step": 7641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.207735617982658e-08,
+      "loss": 3.9561,
+      "step": 7642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.134948031553678e-08,
+      "loss": 3.8928,
+      "step": 7643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.062589169438249e-08,
+      "loss": 4.0654,
+      "step": 7644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.99065904407614e-08,
+      "loss": 3.831,
+      "step": 7645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9191576678324646e-08,
+      "loss": 3.8943,
+      "step": 7646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8480850529998856e-08,
+      "loss": 3.8998,
+      "step": 7647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.7774412117958534e-08,
+      "loss": 3.9314,
+      "step": 7648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.707226156365375e-08,
+      "loss": 3.8798,
+      "step": 7649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.637439898779073e-08,
+      "loss": 3.8894,
+      "step": 7650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.56808245103374e-08,
+      "loss": 4.0096,
+      "step": 7651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.499153825053172e-08,
+      "loss": 4.0174,
+      "step": 7652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4306540326867774e-08,
+      "loss": 3.9342,
+      "step": 7653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3625830857104154e-08,
+      "loss": 3.8649,
+      "step": 7654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.294940995826392e-08,
+      "loss": 3.9927,
+      "step": 7655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.227727774663183e-08,
+      "loss": 3.8256,
+      "step": 7656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1609434337754337e-08,
+      "loss": 3.9493,
+      "step": 7657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.094587984643962e-08,
+      "loss": 3.89,
+      "step": 7658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0286614386763074e-08,
+      "loss": 4.0601,
+      "step": 7659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.963163807205906e-08,
+      "loss": 3.9295,
+      "step": 7660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.898095101492917e-08,
+      "loss": 3.9333,
+      "step": 7661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.833455332722836e-08,
+      "loss": 3.9424,
+      "step": 7662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.769244512008164e-08,
+      "loss": 3.8401,
+      "step": 7663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7054626503878486e-08,
+      "loss": 4.083,
+      "step": 7664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.642109758826174e-08,
+      "loss": 3.8756,
+      "step": 7665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.579185848214429e-08,
+      "loss": 3.7777,
+      "step": 7666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5166909293703484e-08,
+      "loss": 3.8752,
+      "step": 7667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.454625013037006e-08,
+      "loss": 3.6842,
+      "step": 7668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.392988109884477e-08,
+      "loss": 3.9635,
+      "step": 7669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3317802305090085e-08,
+      "loss": 3.9794,
+      "step": 7670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.271001385432738e-08,
+      "loss": 4.0577,
+      "step": 7671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2106515851042525e-08,
+      "loss": 3.7526,
+      "step": 7672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.150730839898309e-08,
+      "loss": 3.8728,
+      "step": 7673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.091239160116112e-08,
+      "loss": 3.9084,
+      "step": 7674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.032176555985035e-08,
+      "loss": 3.9531,
+      "step": 7675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9735430376586226e-08,
+      "loss": 3.8591,
+      "step": 7676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9153386152165906e-08,
+      "loss": 3.7733,
+      "step": 7677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.857563298664823e-08,
+      "loss": 3.7583,
+      "step": 7678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.800217097935932e-08,
+      "loss": 3.9136,
+      "step": 7679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7433000228878634e-08,
+      "loss": 3.8873,
+      "step": 7680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.686812083305569e-08,
+      "loss": 3.9165,
+      "step": 7681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.630753288900446e-08,
+      "loss": 3.9944,
+      "step": 7682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.575123649308953e-08,
+      "loss": 4.0359,
+      "step": 7683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5199231740945525e-08,
+      "loss": 3.9917,
+      "step": 7684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4651518727474296e-08,
+      "loss": 4.001,
+      "step": 7685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.410809754682831e-08,
+      "loss": 3.9152,
+      "step": 7686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3568968292430056e-08,
+      "loss": 3.7749,
+      "step": 7687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.303413105696096e-08,
+      "loss": 3.9655,
+      "step": 7688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.250358593236691e-08,
+      "loss": 3.9224,
+      "step": 7689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1977333009855506e-08,
+      "loss": 3.9201,
+      "step": 7690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.145537237989326e-08,
+      "loss": 3.7562,
+      "step": 7691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0937704132213975e-08,
+      "loss": 3.786,
+      "step": 7692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0424328355810325e-08,
+      "loss": 3.8673,
+      "step": 7693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.991524513893396e-08,
+      "loss": 3.9844,
+      "step": 7694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9410454569106517e-08,
+      "loss": 3.9177,
+      "step": 7695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8909956733105813e-08,
+      "loss": 3.9787,
+      "step": 7696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8413751716974134e-08,
+      "loss": 3.8872,
+      "step": 7697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.792183960601269e-08,
+      "loss": 3.7289,
+      "step": 7698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.743422048478994e-08,
+      "loss": 3.97,
+      "step": 7699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6950894437130504e-08,
+      "loss": 3.897,
+      "step": 7700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6471861546123466e-08,
+      "loss": 3.8018,
+      "step": 7701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.59971218941224e-08,
+      "loss": 3.9316,
+      "step": 7702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5526675562739798e-08,
+      "loss": 3.8451,
+      "step": 7703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.506052263284986e-08,
+      "loss": 4.0354,
+      "step": 7704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4598663184591252e-08,
+      "loss": 3.9058,
+      "step": 7705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4141097297361582e-08,
+      "loss": 3.9961,
+      "step": 7706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3687825049822922e-08,
+      "loss": 3.9333,
+      "step": 7707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3238846519896274e-08,
+      "loss": 3.9405,
+      "step": 7708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2794161784769895e-08,
+      "loss": 3.9435,
+      "step": 7709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2353770920885418e-08,
+      "loss": 3.9908,
+      "step": 7710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1917674003954502e-08,
+      "loss": 4.0245,
+      "step": 7711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1485871108944954e-08,
+      "loss": 3.8364,
+      "step": 7712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.105836231009184e-08,
+      "loss": 3.8709,
+      "step": 7713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.063514768088637e-08,
+      "loss": 3.8617,
+      "step": 7714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0216227294084234e-08,
+      "loss": 3.945,
+      "step": 7715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9801601221702827e-08,
+      "loss": 3.9245,
+      "step": 7716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.939126953502124e-08,
+      "loss": 3.87,
+      "step": 7717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8985232304580268e-08,
+      "loss": 3.9823,
+      "step": 7718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8583489600182412e-08,
+      "loss": 3.9046,
+      "step": 7719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8186041490894644e-08,
+      "loss": 3.8326,
+      "step": 7720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7792888045037314e-08,
+      "loss": 4.0203,
+      "step": 7721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7404029330203576e-08,
+      "loss": 3.8299,
+      "step": 7722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7019465413239953e-08,
+      "loss": 3.9595,
+      "step": 7723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.663919636025746e-08,
+      "loss": 3.9721,
+      "step": 7724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6263222236628794e-08,
+      "loss": 3.8459,
+      "step": 7725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5891543106988372e-08,
+      "loss": 3.93,
+      "step": 7726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.55241590352323e-08,
+      "loss": 3.9596,
+      "step": 7727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5161070084518392e-08,
+      "loss": 3.9491,
+      "step": 7728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4802276317266162e-08,
+      "loss": 3.9528,
+      "step": 7729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.444777779515405e-08,
+      "loss": 3.8366,
+      "step": 7730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4097574579127748e-08,
+      "loss": 3.8386,
+      "step": 7731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3751666729389101e-08,
+      "loss": 3.9004,
+      "step": 7732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3410054305404429e-08,
+      "loss": 3.9864,
+      "step": 7733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3072737365901755e-08,
+      "loss": 3.9122,
+      "step": 7734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2739715968868026e-08,
+      "loss": 3.9112,
+      "step": 7735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2410990171554671e-08,
+      "loss": 3.828,
+      "step": 7736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2086560030474814e-08,
+      "loss": 3.8903,
+      "step": 7737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1766425601397734e-08,
+      "loss": 3.9114,
+      "step": 7738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.145058693936274e-08,
+      "loss": 3.8945,
+      "step": 7739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.113904409866251e-08,
+      "loss": 3.8951,
+      "step": 7740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0831797132854205e-08,
+      "loss": 3.9393,
+      "step": 7741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0528846094762235e-08,
+      "loss": 3.8868,
+      "step": 7742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0230191036464387e-08,
+      "loss": 3.8705,
+      "step": 7743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.93583200930015e-09,
+      "loss": 3.8998,
+      "step": 7744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.64576906387904e-09,
+      "loss": 3.9092,
+      "step": 7745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.360002250061172e-09,
+      "loss": 3.8235,
+      "step": 7746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.078531616976694e-09,
+      "loss": 3.9732,
+      "step": 7747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.801357213011895e-09,
+      "loss": 3.8496,
+      "step": 7748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.528479085817554e-09,
+      "loss": 3.9064,
+      "step": 7749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.259897282303364e-09,
+      "loss": 3.9475,
+      "step": 7750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.995611848640728e-09,
+      "loss": 3.8904,
+      "step": 7751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.73562283026552e-09,
+      "loss": 3.893,
+      "step": 7752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.479930271869774e-09,
+      "loss": 3.8378,
+      "step": 7753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.228534217415539e-09,
+      "loss": 3.9358,
+      "step": 7754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.981434710115475e-09,
+      "loss": 3.9106,
+      "step": 7755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.738631792452266e-09,
+      "loss": 3.8043,
+      "step": 7756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.500125506161969e-09,
+      "loss": 3.9652,
+      "step": 7757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.265915892253449e-09,
+      "loss": 3.8252,
+      "step": 7758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.036002990983392e-09,
+      "loss": 3.8128,
+      "step": 7759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.810386841878512e-09,
+      "loss": 3.8777,
+      "step": 7760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.589067483727229e-09,
+      "loss": 3.9905,
+      "step": 7761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3720449545768825e-09,
+      "loss": 4.0063,
+      "step": 7762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.159319291733744e-09,
+      "loss": 3.8579,
+      "step": 7763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.950890531765784e-09,
+      "loss": 3.9357,
+      "step": 7764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.746758710511001e-09,
+      "loss": 3.9725,
+      "step": 7765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.546923863055219e-09,
+      "loss": 3.8536,
+      "step": 7766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.351386023757065e-09,
+      "loss": 3.91,
+      "step": 7767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1601452262313155e-09,
+      "loss": 3.8857,
+      "step": 7768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.973201503351675e-09,
+      "loss": 3.9298,
+      "step": 7769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.790554887256326e-09,
+      "loss": 3.8274,
+      "step": 7770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.612205409347924e-09,
+      "loss": 3.8866,
+      "step": 7771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.438153100282504e-09,
+      "loss": 3.8667,
+      "step": 7772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2683979899833517e-09,
+      "loss": 3.7872,
+      "step": 7773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.102940107635455e-09,
+      "loss": 3.8644,
+      "step": 7774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9417794816799516e-09,
+      "loss": 3.8236,
+      "step": 7775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7849161398224578e-09,
+      "loss": 3.8072,
+      "step": 7776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.632350109033066e-09,
+      "loss": 3.8454,
+      "step": 7777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4840814155352443e-09,
+      "loss": 3.7741,
+      "step": 7778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3401100848197135e-09,
+      "loss": 3.8391,
+      "step": 7779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2004361416361195e-09,
+      "loss": 3.9066,
+      "step": 7780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0650596099985876e-09,
+      "loss": 3.883,
+      "step": 7781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9339805131773914e-09,
+      "loss": 3.8049,
+      "step": 7782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8071988737100588e-09,
+      "loss": 3.8152,
+      "step": 7783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6847147133847163e-09,
+      "loss": 3.9064,
+      "step": 7784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5665280532650706e-09,
+      "loss": 3.8532,
+      "step": 7785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4526389136654273e-09,
+      "loss": 3.9709,
+      "step": 7786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3430473141645695e-09,
+      "loss": 3.897,
+      "step": 7787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2377532736057573e-09,
+      "loss": 3.8906,
+      "step": 7788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1367568100856262e-09,
+      "loss": 3.8726,
+      "step": 7789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0400579409680645e-09,
+      "loss": 3.8072,
+      "step": 7790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.476566828786614e-10,
+      "loss": 3.836,
+      "step": 7791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.595530516991578e-10,
+      "loss": 3.8117,
+      "step": 7792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.757470625785468e-10,
+      "loss": 3.8902,
+      "step": 7793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.962387299219719e-10,
+      "loss": 4.0,
+      "step": 7794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.210280674018298e-10,
+      "loss": 3.8189,
+      "step": 7795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.501150879411165e-10,
+      "loss": 4.0116,
+      "step": 7796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.83499803735632e-10,
+      "loss": 3.8731,
+      "step": 7797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2118222623455106e-10,
+      "loss": 3.8599,
+      "step": 7798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6316236615430155e-10,
+      "loss": 3.865,
+      "step": 7799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.094402334674618e-10,
+      "loss": 3.7706,
+      "step": 7800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6001583740553614e-10,
+      "loss": 3.9518,
+      "step": 7801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1488918647283306e-10,
+      "loss": 3.9183,
+      "step": 7802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7406028842148482e-10,
+      "loss": 3.9306,
+      "step": 7803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3752915027087642e-10,
+      "loss": 3.9142,
+      "step": 7804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0529577830209469e-10,
+      "loss": 3.8663,
+      "step": 7805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.736017805792806e-11,
+      "loss": 3.9293,
+      "step": 7806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3722354340091094e-11,
+      "loss": 3.9096,
+      "step": 7807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.438231120922453e-11,
+      "loss": 3.9686,
+      "step": 7808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.934005199599742e-11,
+      "loss": 3.7662,
+      "step": 7809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.595579281678312e-12,
+      "loss": 3.8652,
+      "step": 7810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1488949120129774e-12,
+      "loss": 3.8479,
+      "step": 7811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 4.0256,
+      "step": 7812
+    },
+    {
+      "epoch": 1.0,
+      "step": 7812,
+      "total_flos": 1.4870537719250944e+18,
+      "train_loss": 4.018061516410683,
+      "train_runtime": 34130.3967,
+      "train_samples_per_second": 58.599,
+      "train_steps_per_second": 0.229
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7812,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 24000,
+  "total_flos": 1.4870537719250944e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}