diff --git "a/checkpoint-9472/trainer_state.json" "b/checkpoint-9472/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-9472/trainer_state.json"
@@ -0,0 +1,56923 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9700499168053245,
+  "eval_steps": 1000,
+  "global_step": 9472,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.859375e-09,
+      "loss": 1.296,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.34375e-08,
+      "loss": 1.3068,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2734375e-08,
+      "loss": 1.1979,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.375e-08,
+      "loss": 1.3552,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4648437500000001e-07,
+      "loss": 1.2051,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.109375e-07,
+      "loss": 1.7044,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.87109375e-07,
+      "loss": 1.4791,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.75e-07,
+      "loss": 1.5419,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.74609375e-07,
+      "loss": 1.332,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.859375000000001e-07,
+      "loss": 1.3585,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.08984375e-07,
+      "loss": 1.3662,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4375e-07,
+      "loss": 1.1231,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.90234375e-07,
+      "loss": 1.5108,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1484375e-06,
+      "loss": 1.3973,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.318359375e-06,
+      "loss": 1.2765,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2937,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.693359375e-06,
+      "loss": 1.4439,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8984375e-06,
+      "loss": 1.4299,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.115234375e-06,
+      "loss": 1.474,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3437500000000002e-06,
+      "loss": 1.3522,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.583984375e-06,
+      "loss": 1.5762,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8359375e-06,
+      "loss": 1.5249,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.099609375e-06,
+      "loss": 1.2834,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.375e-06,
+      "loss": 1.2947,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6621093750000003e-06,
+      "loss": 1.2467,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9609375e-06,
+      "loss": 1.3112,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.271484375e-06,
+      "loss": 1.1706,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.59375e-06,
+      "loss": 1.2307,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927734375e-06,
+      "loss": 1.2279,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2734375e-06,
+      "loss": 1.4434,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.630859375e-06,
+      "loss": 1.3737,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 1.4432,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999998388251064e-06,
+      "loss": 1.2401,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999999355300446e-06,
+      "loss": 1.5202,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999998549426068e-06,
+      "loss": 1.3696,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999997421202059e-06,
+      "loss": 1.2896,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999995970628542e-06,
+      "loss": 1.0738,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999994197705672e-06,
+      "loss": 1.2536,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999992102433639e-06,
+      "loss": 1.0722,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999989684812669e-06,
+      "loss": 1.268,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999986944843022e-06,
+      "loss": 1.0615,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999983882524992e-06,
+      "loss": 1.0649,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999980497858907e-06,
+      "loss": 1.3919,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999976790845131e-06,
+      "loss": 1.4423,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999972761484064e-06,
+      "loss": 1.1149,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999968409776138e-06,
+      "loss": 1.2653,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99996373572182e-06,
+      "loss": 1.1168,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999958739321613e-06,
+      "loss": 1.2264,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999953420576053e-06,
+      "loss": 1.2457,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999947779485713e-06,
+      "loss": 0.9746,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999941816051197e-06,
+      "loss": 1.1715,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999935530273147e-06,
+      "loss": 1.2051,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999928922152239e-06,
+      "loss": 1.2213,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999921991689182e-06,
+      "loss": 1.2407,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999147388847215e-06,
+      "loss": 1.2428,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999907163739636e-06,
+      "loss": 0.8133,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998992662547405e-06,
+      "loss": 0.9955,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999891046430881e-06,
+      "loss": 1.2974,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999882504268945e-06,
+      "loss": 0.8606,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999873639769846e-06,
+      "loss": 0.9606,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99986445293454e-06,
+      "loss": 0.9696,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999854943764012e-06,
+      "loss": 1.5109,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999845112259284e-06,
+      "loss": 1.1779,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999834958421414e-06,
+      "loss": 0.9999,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99982448225149e-06,
+      "loss": 1.1015,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999813683750641e-06,
+      "loss": 0.911,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999802562920025e-06,
+      "loss": 1.2481,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999791119760837e-06,
+      "loss": 1.0154,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999779354274308e-06,
+      "loss": 1.2935,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999767266461702e-06,
+      "loss": 0.9441,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9997548563243165e-06,
+      "loss": 1.1343,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9997421238634865e-06,
+      "loss": 1.1293,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999729069080578e-06,
+      "loss": 1.2322,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999715691976996e-06,
+      "loss": 1.1863,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999701992554177e-06,
+      "loss": 1.3767,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999687970813593e-06,
+      "loss": 1.085,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99967362675675e-06,
+      "loss": 1.3077,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996589603851905e-06,
+      "loss": 1.1452,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999643971700489e-06,
+      "loss": 1.0257,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999628660704258e-06,
+      "loss": 1.1751,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99961302739814e-06,
+      "loss": 0.9148,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999597071783817e-06,
+      "loss": 0.9614,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999580793863003e-06,
+      "loss": 1.1894,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999564193637447e-06,
+      "loss": 1.241,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999547271108932e-06,
+      "loss": 1.0488,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999530026279275e-06,
+      "loss": 0.9997,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999512459150333e-06,
+      "loss": 1.1317,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99949456972399e-06,
+      "loss": 1.1307,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999476358002171e-06,
+      "loss": 1.0631,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9994578239868306e-06,
+      "loss": 1.0541,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999438967679961e-06,
+      "loss": 1.1123,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999419789083588e-06,
+      "loss": 1.1307,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9994002881997734e-06,
+      "loss": 1.2111,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999380465030611e-06,
+      "loss": 1.2605,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999360319578232e-06,
+      "loss": 0.9863,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999339851844801e-06,
+      "loss": 1.0087,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999319061832517e-06,
+      "loss": 1.2027,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9992979495436134e-06,
+      "loss": 1.1981,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99927651498036e-06,
+      "loss": 0.8458,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9992547581450585e-06,
+      "loss": 0.9559,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999232679040047e-06,
+      "loss": 1.3397,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999210277667698e-06,
+      "loss": 1.2488,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99918755403042e-06,
+      "loss": 0.9681,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999164508130651e-06,
+      "loss": 1.0477,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999141139970872e-06,
+      "loss": 1.0544,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999117449553591e-06,
+      "loss": 0.94,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999093436881353e-06,
+      "loss": 0.9462,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999069101956739e-06,
+      "loss": 1.0561,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999044444782366e-06,
+      "loss": 1.1276,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99901946536088e-06,
+      "loss": 0.9206,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998994163694967e-06,
+      "loss": 0.939,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998968539787344e-06,
+      "loss": 1.1145,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998942593640767e-06,
+      "loss": 1.0664,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998916325258021e-06,
+      "loss": 0.9519,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99888973464193e-06,
+      "loss": 1.0119,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998862821795352e-06,
+      "loss": 0.981,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998835586721177e-06,
+      "loss": 1.1971,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998808029422333e-06,
+      "loss": 1.0213,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998780149901779e-06,
+      "loss": 1.1281,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998751948162512e-06,
+      "loss": 0.8975,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998723424207563e-06,
+      "loss": 1.0096,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998694578039996e-06,
+      "loss": 1.2597,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99866540966291e-06,
+      "loss": 1.1371,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998635919079441e-06,
+      "loss": 1.1777,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9986061062927545e-06,
+      "loss": 0.9861,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9985759713060565e-06,
+      "loss": 1.1509,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998545514122585e-06,
+      "loss": 1.0427,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9985147347456115e-06,
+      "loss": 1.0939,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998483633178444e-06,
+      "loss": 1.0729,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9984522094244245e-06,
+      "loss": 1.0709,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998420463486928e-06,
+      "loss": 1.103,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998388395369367e-06,
+      "loss": 1.0118,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998356005075187e-06,
+      "loss": 1.1575,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998323292607869e-06,
+      "loss": 1.0233,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9982902579709254e-06,
+      "loss": 0.8519,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998256901167909e-06,
+      "loss": 0.9166,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998223222202402e-06,
+      "loss": 1.1034,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998189221078024e-06,
+      "loss": 0.9666,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998154897798428e-06,
+      "loss": 1.165,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998120252367302e-06,
+      "loss": 1.2268,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998085284788369e-06,
+      "loss": 0.9355,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998049995065386e-06,
+      "loss": 1.1064,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998014383202145e-06,
+      "loss": 1.1342,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997978449202472e-06,
+      "loss": 0.9912,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997942193070229e-06,
+      "loss": 0.998,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997905614809312e-06,
+      "loss": 1.0592,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99786871442365e-06,
+      "loss": 0.8856,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9978314919172075e-06,
+      "loss": 1.0244,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997793947293987e-06,
+      "loss": 1.1642,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99775608055802e-06,
+      "loss": 0.9795,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997717891713375e-06,
+      "loss": 0.9266,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997679380764158e-06,
+      "loss": 1.1468,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997640547714504e-06,
+      "loss": 0.9199,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997601392568589e-06,
+      "loss": 1.0031,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997561915330617e-06,
+      "loss": 1.0035,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9975221160048306e-06,
+      "loss": 0.9766,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997481994595507e-06,
+      "loss": 1.1334,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997441551106958e-06,
+      "loss": 1.2685,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997400785543528e-06,
+      "loss": 1.0696,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997359697909597e-06,
+      "loss": 0.9403,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99731828820958e-06,
+      "loss": 0.9769,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9972765564479265e-06,
+      "loss": 0.8709,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997234502629122e-06,
+      "loss": 1.0661,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997192126757684e-06,
+      "loss": 1.0048,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997149428838165e-06,
+      "loss": 1.1654,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9971064088751545e-06,
+      "loss": 1.0636,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997063066873274e-06,
+      "loss": 1.0468,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997019402837181e-06,
+      "loss": 1.071,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996975416771567e-06,
+      "loss": 0.9035,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996931108681158e-06,
+      "loss": 0.9098,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996886478570716e-06,
+      "loss": 0.9487,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996841526445035e-06,
+      "loss": 0.8831,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996796252308945e-06,
+      "loss": 0.9892,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996750656167313e-06,
+      "loss": 1.0303,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996704738025036e-06,
+      "loss": 1.1696,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9966584978870496e-06,
+      "loss": 1.145,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996611935758321e-06,
+      "loss": 1.0817,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996565051643854e-06,
+      "loss": 0.9447,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9965178455486854e-06,
+      "loss": 0.9365,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996470317477888e-06,
+      "loss": 0.9432,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996422467436569e-06,
+      "loss": 1.0671,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99637429542987e-06,
+      "loss": 1.1596,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9963258014629674e-06,
+      "loss": 1.0035,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996276985541071e-06,
+      "loss": 0.9413,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9962278476694256e-06,
+      "loss": 1.0505,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996178387853312e-06,
+      "loss": 1.0839,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996128606098044e-06,
+      "loss": 0.9701,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996078502408972e-06,
+      "loss": 1.0174,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.996028076791477e-06,
+      "loss": 0.9293,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995977329250981e-06,
+      "loss": 1.0603,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995926259792934e-06,
+      "loss": 1.2609,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995874868422825e-06,
+      "loss": 1.027,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995823155146174e-06,
+      "loss": 1.2603,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995771119968538e-06,
+      "loss": 0.9458,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9957187628955105e-06,
+      "loss": 0.9053,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995666083932715e-06,
+      "loss": 1.0194,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995613083085813e-06,
+      "loss": 1.1203,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9955597603604985e-06,
+      "loss": 1.1192,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9955061157625015e-06,
+      "loss": 1.2204,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995452149297587e-06,
+      "loss": 0.9118,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9953978609715515e-06,
+      "loss": 1.412,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995343250790231e-06,
+      "loss": 0.9113,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995288318759491e-06,
+      "loss": 0.9874,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9952330648852344e-06,
+      "loss": 0.9065,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9951774891734e-06,
+      "loss": 1.2576,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9951215916299575e-06,
+      "loss": 1.0187,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995065372260913e-06,
+      "loss": 0.8289,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995008831072308e-06,
+      "loss": 0.9299,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9949519680702195e-06,
+      "loss": 0.8268,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994894783260755e-06,
+      "loss": 0.8399,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9948372766500595e-06,
+      "loss": 0.977,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994779448244313e-06,
+      "loss": 1.0439,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994721298049727e-06,
+      "loss": 1.0908,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994662826072552e-06,
+      "loss": 1.0506,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99460403231907e-06,
+      "loss": 0.7117,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994544916795599e-06,
+      "loss": 0.9917,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9944854795084894e-06,
+      "loss": 1.1558,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994425720464129e-06,
+      "loss": 1.0204,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994365639668939e-06,
+      "loss": 1.0093,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994305237129374e-06,
+      "loss": 1.1288,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994244512851926e-06,
+      "loss": 0.8111,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994183466843119e-06,
+      "loss": 0.7688,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99412209910951e-06,
+      "loss": 1.0126,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994060409657697e-06,
+      "loss": 0.9432,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9939983984943064e-06,
+      "loss": 0.8748,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993936065626001e-06,
+      "loss": 1.1009,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993873411059479e-06,
+      "loss": 1.335,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993810434801473e-06,
+      "loss": 0.9907,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993747136858748e-06,
+      "loss": 1.145,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993683517238109e-06,
+      "loss": 0.8521,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993619575946389e-06,
+      "loss": 1.1498,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993555312990459e-06,
+      "loss": 0.9589,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993490728377223e-06,
+      "loss": 0.937,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9934258221136235e-06,
+      "loss": 1.0393,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993360594206633e-06,
+      "loss": 0.8115,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993295044663259e-06,
+      "loss": 1.1074,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993229173490547e-06,
+      "loss": 0.9184,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993162980695574e-06,
+      "loss": 1.2034,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993096466285451e-06,
+      "loss": 1.1001,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993029630267327e-06,
+      "loss": 0.8281,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992962472648384e-06,
+      "loss": 1.1165,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9928949934358355e-06,
+      "loss": 0.9084,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992827192636934e-06,
+      "loss": 0.8897,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992759070258965e-06,
+      "loss": 1.0191,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992690626309246e-06,
+      "loss": 1.1065,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992621860795134e-06,
+      "loss": 0.9703,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9925527737240155e-06,
+      "loss": 1.307,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9924833651033155e-06,
+      "loss": 0.8289,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992413634940492e-06,
+      "loss": 0.7635,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992343583243036e-06,
+      "loss": 1.096,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992273210018476e-06,
+      "loss": 0.9323,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9922025152743725e-06,
+      "loss": 0.8956,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992131499018324e-06,
+      "loss": 1.0958,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9920601612579575e-06,
+      "loss": 1.0515,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991988502000941e-06,
+      "loss": 1.1646,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991916521254974e-06,
+      "loss": 0.8565,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991844219027789e-06,
+      "loss": 1.2754,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991771595327157e-06,
+      "loss": 0.9573,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99169865016088e-06,
+      "loss": 0.9665,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991625383536798e-06,
+      "loss": 0.8779,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991551795462781e-06,
+      "loss": 0.9983,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991477885946736e-06,
+      "loss": 0.9575,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991403654996606e-06,
+      "loss": 0.9767,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991329102620368e-06,
+      "loss": 0.9856,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.991254228826031e-06,
+      "loss": 1.097,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99117903362164e-06,
+      "loss": 0.9823,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9911035170152755e-06,
+      "loss": 1.1262,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.991027679015052e-06,
+      "loss": 1.0386,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9909515196291175e-06,
+      "loss": 0.7662,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990875038865656e-06,
+      "loss": 0.977,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990798236732884e-06,
+      "loss": 1.2878,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990721113239055e-06,
+      "loss": 0.8911,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990643668392457e-06,
+      "loss": 1.0702,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990565902201409e-06,
+      "loss": 1.0518,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990487814674268e-06,
+      "loss": 0.8105,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990409405819426e-06,
+      "loss": 1.0531,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990330675645307e-06,
+      "loss": 1.0297,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9902516241603695e-06,
+      "loss": 0.9146,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990172251373108e-06,
+      "loss": 1.0208,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990092557292052e-06,
+      "loss": 0.9104,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.990012541925763e-06,
+      "loss": 1.019,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989932205282841e-06,
+      "loss": 1.0193,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989851547371916e-06,
+      "loss": 1.1017,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989770568201656e-06,
+      "loss": 0.8855,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989689267780762e-06,
+      "loss": 1.3386,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989607646117969e-06,
+      "loss": 0.7632,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989525703222049e-06,
+      "loss": 0.9788,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989443439101805e-06,
+      "loss": 0.9542,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989360853766076e-06,
+      "loss": 1.0081,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989277947223737e-06,
+      "loss": 1.0064,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989194719483696e-06,
+      "loss": 1.3006,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989111170554895e-06,
+      "loss": 0.9354,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.989027300446313e-06,
+      "loss": 1.0887,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988943109166961e-06,
+      "loss": 1.2155,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988858596725884e-06,
+      "loss": 1.0584,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988773763132165e-06,
+      "loss": 1.1149,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988688608394918e-06,
+      "loss": 1.0847,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988603132523294e-06,
+      "loss": 0.9848,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988517335526476e-06,
+      "loss": 0.9482,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9884312174136835e-06,
+      "loss": 1.031,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988344778194171e-06,
+      "loss": 1.134,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988258017877224e-06,
+      "loss": 1.0821,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988170936472167e-06,
+      "loss": 0.8503,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.988083533988355e-06,
+      "loss": 1.1176,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987995810435181e-06,
+      "loss": 1.185,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987907765822072e-06,
+      "loss": 1.1113,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987819400158485e-06,
+      "loss": 1.1201,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987730713453917e-06,
+      "loss": 1.0803,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987641705717897e-06,
+      "loss": 1.0283,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987552376959988e-06,
+      "loss": 0.9846,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987462727189791e-06,
+      "loss": 0.8748,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.987372756416937e-06,
+      "loss": 1.15,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.987282464651092e-06,
+      "loss": 1.0994,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9871918519019616e-06,
+      "loss": 0.8744,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.987100918179278e-06,
+      "loss": 1.1972,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.987009663492814e-06,
+      "loss": 1.0936,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9869180878523755e-06,
+      "loss": 0.9805,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986826191267801e-06,
+      "loss": 0.9992,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986733973748966e-06,
+      "loss": 0.843,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986641435305779e-06,
+      "loss": 0.9656,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986548575948182e-06,
+      "loss": 0.9131,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9864553956861546e-06,
+      "loss": 1.287,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986361894529707e-06,
+      "loss": 0.8363,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986268072488887e-06,
+      "loss": 1.0932,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.986173929573777e-06,
+      "loss": 0.7928,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9860794657944895e-06,
+      "loss": 1.1357,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.985984681161177e-06,
+      "loss": 1.0201,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9858895756840246e-06,
+      "loss": 1.2461,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9857941493732495e-06,
+      "loss": 0.9724,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9856984022391064e-06,
+      "loss": 0.9171,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.985602334291883e-06,
+      "loss": 0.9148,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.985505945541902e-06,
+      "loss": 0.9829,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.98540923599952e-06,
+      "loss": 1.0627,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.985312205675129e-06,
+      "loss": 1.0944,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9852148545791535e-06,
+      "loss": 0.9828,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.985117182722055e-06,
+      "loss": 0.8804,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.98501919011433e-06,
+      "loss": 1.1282,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984920876766505e-06,
+      "loss": 1.2349,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984822242689144e-06,
+      "loss": 1.044,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984723287892846e-06,
+      "loss": 1.3231,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984624012388245e-06,
+      "loss": 0.9244,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984524416186006e-06,
+      "loss": 0.8762,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9844244992968315e-06,
+      "loss": 0.7774,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984324261731458e-06,
+      "loss": 1.0715,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984223703500654e-06,
+      "loss": 1.1075,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984122824615229e-06,
+      "loss": 0.8161,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.984021625086017e-06,
+      "loss": 0.9664,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983920104923895e-06,
+      "loss": 0.8845,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.98381826413977e-06,
+      "loss": 1.02,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983716102744585e-06,
+      "loss": 1.0522,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.98361362074932e-06,
+      "loss": 1.1609,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983510818164982e-06,
+      "loss": 1.0494,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983407695002621e-06,
+      "loss": 1.0643,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983304251273316e-06,
+      "loss": 1.0307,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983200486988182e-06,
+      "loss": 1.0185,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.983096402158368e-06,
+      "loss": 0.9799,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9829919967950585e-06,
+      "loss": 1.1228,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.982887270909472e-06,
+      "loss": 0.8885,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.982782224512861e-06,
+      "loss": 0.9031,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9826768576165135e-06,
+      "loss": 1.0953,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.982571170231749e-06,
+      "loss": 1.0339,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.982465162369926e-06,
+      "loss": 0.9442,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.982358834042435e-06,
+      "loss": 0.8183,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.982252185260699e-06,
+      "loss": 0.9198,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.982145216036178e-06,
+      "loss": 1.1153,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.982037926380368e-06,
+      "loss": 1.1304,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981930316304794e-06,
+      "loss": 0.9464,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981822385821021e-06,
+      "loss": 1.2954,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981714134940646e-06,
+      "loss": 1.0039,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981605563675298e-06,
+      "loss": 1.1989,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981496672036647e-06,
+      "loss": 1.0865,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981387460036391e-06,
+      "loss": 0.8881,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981277927686265e-06,
+      "loss": 1.0311,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981168074998037e-06,
+      "loss": 0.8663,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.981057901983514e-06,
+      "loss": 0.9217,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.980947408654532e-06,
+      "loss": 1.1235,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9808365950229625e-06,
+      "loss": 1.1253,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.980725461100715e-06,
+      "loss": 1.0498,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.980614006899728e-06,
+      "loss": 1.001,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9805022324319786e-06,
+      "loss": 1.1628,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.980390137709478e-06,
+      "loss": 1.1906,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.98027772274427e-06,
+      "loss": 0.911,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.980164987548433e-06,
+      "loss": 0.8285,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.98005193213408e-06,
+      "loss": 1.0446,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979938556513359e-06,
+      "loss": 1.0896,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979824860698454e-06,
+      "loss": 1.1504,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979710844701581e-06,
+      "loss": 1.2828,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979596508534989e-06,
+      "loss": 0.9782,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979481852210965e-06,
+      "loss": 0.9754,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979366875741829e-06,
+      "loss": 1.1549,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979251579139935e-06,
+      "loss": 0.9424,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.979135962417671e-06,
+      "loss": 1.0477,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9790200255874605e-06,
+      "loss": 0.7897,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978903768661761e-06,
+      "loss": 0.8454,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978787191653064e-06,
+      "loss": 0.9438,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978670294573896e-06,
+      "loss": 1.0034,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978553077436817e-06,
+      "loss": 0.9945,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978435540254422e-06,
+      "loss": 0.999,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978317683039342e-06,
+      "loss": 0.8327,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978199505804239e-06,
+      "loss": 0.8786,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.978081008561811e-06,
+      "loss": 1.1112,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.977962191324792e-06,
+      "loss": 0.9308,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.977843054105947e-06,
+      "loss": 0.975,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9777235969180795e-06,
+      "loss": 0.979,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9776038197740235e-06,
+      "loss": 0.8599,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.97748372268665e-06,
+      "loss": 0.9853,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.977363305668862e-06,
+      "loss": 0.8872,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9772425687335995e-06,
+      "loss": 0.8615,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9771215118938355e-06,
+      "loss": 0.9343,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.977000135162578e-06,
+      "loss": 1.0547,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.976878438552868e-06,
+      "loss": 1.1056,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.976756422077782e-06,
+      "loss": 0.9628,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.976634085750432e-06,
+      "loss": 1.3367,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.97651142958396e-06,
+      "loss": 1.096,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.976388453591548e-06,
+      "loss": 0.8851,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9762651577864095e-06,
+      "loss": 1.003,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.976141542181792e-06,
+      "loss": 0.6459,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.976017606790979e-06,
+      "loss": 1.1279,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9758933516272854e-06,
+      "loss": 1.3032,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975768776704064e-06,
+      "loss": 0.8337,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975643882034699e-06,
+      "loss": 1.1416,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975518667632612e-06,
+      "loss": 1.1516,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975393133511257e-06,
+      "loss": 0.7993,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975267279684121e-06,
+      "loss": 1.057,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975141106164729e-06,
+      "loss": 1.0412,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.975014612966637e-06,
+      "loss": 1.2189,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.974887800103437e-06,
+      "loss": 1.2125,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.974760667588756e-06,
+      "loss": 1.025,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.974633215436253e-06,
+      "loss": 1.0234,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.974505443659622e-06,
+      "loss": 1.0045,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9743773522725955e-06,
+      "loss": 0.9311,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.974248941288934e-06,
+      "loss": 1.062,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.974120210722436e-06,
+      "loss": 0.9071,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.973991160586933e-06,
+      "loss": 0.9888,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.973861790896292e-06,
+      "loss": 0.9163,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.973732101664415e-06,
+      "loss": 0.9302,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9736020929052345e-06,
+      "loss": 1.1172,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.973471764632722e-06,
+      "loss": 0.9883,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.97334111686088e-06,
+      "loss": 0.8618,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.973210149603748e-06,
+      "loss": 1.0525,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.973078862875397e-06,
+      "loss": 0.9544,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972947256689934e-06,
+      "loss": 1.1121,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972815331061502e-06,
+      "loss": 0.9561,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9726830860042736e-06,
+      "loss": 0.8641,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972550521532459e-06,
+      "loss": 1.0043,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972417637660304e-06,
+      "loss": 0.9515,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972284434402086e-06,
+      "loss": 1.0395,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972150911772118e-06,
+      "loss": 0.9791,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.972017069784745e-06,
+      "loss": 1.0428,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971882908454352e-06,
+      "loss": 0.829,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971748427795351e-06,
+      "loss": 0.8857,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971613627822194e-06,
+      "loss": 1.2787,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971478508549365e-06,
+      "loss": 0.964,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971343069991381e-06,
+      "loss": 0.8449,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971207312162798e-06,
+      "loss": 0.988,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.971071235078201e-06,
+      "loss": 0.8777,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.970934838752211e-06,
+      "loss": 0.963,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9707981231994845e-06,
+      "loss": 1.0039,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.970661088434712e-06,
+      "loss": 0.9266,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.970523734472618e-06,
+      "loss": 1.1094,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.97038606132796e-06,
+      "loss": 1.0275,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.970248069015532e-06,
+      "loss": 1.0736,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.970109757550161e-06,
+      "loss": 1.0078,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9699711269467085e-06,
+      "loss": 0.9505,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.96983217722007e-06,
+      "loss": 0.7334,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.969692908385177e-06,
+      "loss": 1.1101,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.969553320456991e-06,
+      "loss": 1.0431,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.969413413450514e-06,
+      "loss": 0.9238,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.969273187380778e-06,
+      "loss": 0.979,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.969132642262849e-06,
+      "loss": 1.2238,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968991778111829e-06,
+      "loss": 1.0796,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968850594942856e-06,
+      "loss": 0.9291,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968709092771097e-06,
+      "loss": 0.8568,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968567271611759e-06,
+      "loss": 0.8486,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968425131480078e-06,
+      "loss": 1.1037,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968282672391329e-06,
+      "loss": 1.1142,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.968139894360819e-06,
+      "loss": 1.0361,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.967996797403889e-06,
+      "loss": 1.0687,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9678533815359155e-06,
+      "loss": 0.9581,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.967709646772307e-06,
+      "loss": 1.1905,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.967565593128509e-06,
+      "loss": 0.9304,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9674212206199995e-06,
+      "loss": 0.9152,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9672765292622926e-06,
+      "loss": 0.9298,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.967131519070934e-06,
+      "loss": 1.2012,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.966986190061505e-06,
+      "loss": 1.1008,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9668405422496215e-06,
+      "loss": 0.8475,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.966694575650933e-06,
+      "loss": 1.2126,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.966548290281125e-06,
+      "loss": 0.9447,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.966401686155915e-06,
+      "loss": 1.0535,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.966254763291055e-06,
+      "loss": 0.9571,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.966107521702333e-06,
+      "loss": 0.9626,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.965959961405569e-06,
+      "loss": 0.9257,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.965812082416618e-06,
+      "loss": 1.1311,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.965663884751372e-06,
+      "loss": 1.1409,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9655153684257515e-06,
+      "loss": 1.1109,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.965366533455717e-06,
+      "loss": 1.0314,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.96521737985726e-06,
+      "loss": 1.0284,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.965067907646408e-06,
+      "loss": 0.9187,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.96491811683922e-06,
+      "loss": 0.9242,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.964768007451792e-06,
+      "loss": 0.8744,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.964617579500253e-06,
+      "loss": 1.0825,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.964466833000767e-06,
+      "loss": 0.8537,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.964315767969532e-06,
+      "loss": 1.0924,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.964164384422778e-06,
+      "loss": 1.2262,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.964012682376773e-06,
+      "loss": 0.9773,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.963860661847817e-06,
+      "loss": 1.1186,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.963708322852244e-06,
+      "loss": 1.021,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.963555665406423e-06,
+      "loss": 0.9731,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.963402689526757e-06,
+      "loss": 1.068,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9632493952296845e-06,
+      "loss": 0.927,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.963095782531676e-06,
+      "loss": 1.0042,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.962941851449236e-06,
+      "loss": 0.9498,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.962787601998906e-06,
+      "loss": 0.8147,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.96263303419726e-06,
+      "loss": 0.9474,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.962478148060905e-06,
+      "loss": 0.9219,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.962322943606485e-06,
+      "loss": 1.0259,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.962167420850676e-06,
+      "loss": 1.0261,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.962011579810189e-06,
+      "loss": 0.943,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.96185542050177e-06,
+      "loss": 0.9979,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.961698942942196e-06,
+      "loss": 0.9773,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.961542147148282e-06,
+      "loss": 0.8481,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.961385033136876e-06,
+      "loss": 0.8482,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.961227600924859e-06,
+      "loss": 1.1898,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.961069850529149e-06,
+      "loss": 0.9799,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.960911781966694e-06,
+      "loss": 1.0628,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.960753395254479e-06,
+      "loss": 1.0569,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.960594690409524e-06,
+      "loss": 0.8862,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.96043566744888e-06,
+      "loss": 1.1338,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9602763263896345e-06,
+      "loss": 0.908,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.960116667248909e-06,
+      "loss": 1.006,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.959956690043859e-06,
+      "loss": 1.2493,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.959796394791674e-06,
+      "loss": 0.9802,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.959635781509578e-06,
+      "loss": 0.9829,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.959474850214828e-06,
+      "loss": 0.8838,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.959313600924717e-06,
+      "loss": 0.9058,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.95915203365657e-06,
+      "loss": 1.1249,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9589901484277485e-06,
+      "loss": 0.9173,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.958827945255647e-06,
+      "loss": 0.8104,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.958665424157693e-06,
+      "loss": 0.8825,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.958502585151351e-06,
+      "loss": 0.985,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.958339428254118e-06,
+      "loss": 1.1024,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.958175953483523e-06,
+      "loss": 0.9831,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.958012160857134e-06,
+      "loss": 0.9676,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.957848050392549e-06,
+      "loss": 1.0474,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.957683622107402e-06,
+      "loss": 0.9944,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.957518876019361e-06,
+      "loss": 1.0794,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.957353812146129e-06,
+      "loss": 0.7696,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.957188430505438e-06,
+      "loss": 1.2139,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.957022731115064e-06,
+      "loss": 0.8747,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.956856713992807e-06,
+      "loss": 1.101,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.956690379156508e-06,
+      "loss": 1.1761,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9565237266240375e-06,
+      "loss": 0.8549,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.956356756413305e-06,
+      "loss": 0.9771,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9561894685422495e-06,
+      "loss": 1.1107,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.956021863028845e-06,
+      "loss": 1.0797,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.955853939891105e-06,
+      "loss": 0.8847,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.955685699147068e-06,
+      "loss": 0.9926,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.955517140814814e-06,
+      "loss": 1.0678,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.955348264912455e-06,
+      "loss": 1.0061,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.955179071458135e-06,
+      "loss": 0.9567,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.955009560470034e-06,
+      "loss": 0.8493,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.954839731966368e-06,
+      "loss": 0.9829,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.954669585965383e-06,
+      "loss": 0.8007,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.954499122485362e-06,
+      "loss": 1.1774,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.954328341544621e-06,
+      "loss": 0.8259,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.954157243161511e-06,
+      "loss": 1.0526,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.953985827354415e-06,
+      "loss": 0.8831,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.953814094141754e-06,
+      "loss": 0.9272,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9536420435419785e-06,
+      "loss": 1.1278,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.953469675573576e-06,
+      "loss": 0.8965,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.953296990255068e-06,
+      "loss": 0.9866,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.953123987605008e-06,
+      "loss": 0.99,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.952950667641988e-06,
+      "loss": 0.9218,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.952777030384629e-06,
+      "loss": 1.0053,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.952603075851588e-06,
+      "loss": 1.0152,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.952428804061558e-06,
+      "loss": 0.9785,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.952254215033263e-06,
+      "loss": 0.8592,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.952079308785465e-06,
+      "loss": 1.031,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.951904085336953e-06,
+      "loss": 1.194,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.95172854470656e-06,
+      "loss": 1.0446,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.951552686913145e-06,
+      "loss": 1.0007,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.951376511975604e-06,
+      "loss": 0.8843,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.951200019912868e-06,
+      "loss": 0.8538,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9510232107438995e-06,
+      "loss": 1.0139,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9508460844876984e-06,
+      "loss": 0.9429,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.950668641163296e-06,
+      "loss": 1.2218,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.950490880789758e-06,
+      "loss": 0.9773,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.950312803386186e-06,
+      "loss": 1.0143,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.950134408971714e-06,
+      "loss": 1.088,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.949955697565509e-06,
+      "loss": 1.0467,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.949776669186776e-06,
+      "loss": 0.8823,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.949597323854751e-06,
+      "loss": 1.0345,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.949417661588703e-06,
+      "loss": 1.2222,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.949237682407937e-06,
+      "loss": 0.7486,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9490573863317935e-06,
+      "loss": 0.957,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.948876773379645e-06,
+      "loss": 0.9091,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9486958435708966e-06,
+      "loss": 0.8557,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9485145969249895e-06,
+      "loss": 1.1901,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.948333033461401e-06,
+      "loss": 0.9139,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.948151153199637e-06,
+      "loss": 0.9506,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.947968956159243e-06,
+      "loss": 1.0845,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.947786442359795e-06,
+      "loss": 1.0896,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.947603611820904e-06,
+      "loss": 1.0381,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9474204645622144e-06,
+      "loss": 0.9002,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.947237000603407e-06,
+      "loss": 1.33,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9470532199641946e-06,
+      "loss": 0.9552,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.946869122664323e-06,
+      "loss": 0.886,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.946684708723575e-06,
+      "loss": 0.8843,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.946499978161765e-06,
+      "loss": 0.9104,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.946314930998743e-06,
+      "loss": 1.1829,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.946129567254392e-06,
+      "loss": 0.9859,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945943886948629e-06,
+      "loss": 1.0122,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945757890101405e-06,
+      "loss": 0.9271,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945571576732706e-06,
+      "loss": 0.9198,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945384946862552e-06,
+      "loss": 0.9032,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945198000510996e-06,
+      "loss": 1.1748,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945010737698124e-06,
+      "loss": 0.9531,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.944823158444059e-06,
+      "loss": 0.9518,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.944635262768956e-06,
+      "loss": 0.8515,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.944447050693003e-06,
+      "loss": 0.9075,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.944258522236426e-06,
+      "loss": 1.2077,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.944069677419479e-06,
+      "loss": 0.7758,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.943880516262457e-06,
+      "loss": 0.901,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9436910387856835e-06,
+      "loss": 0.8593,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9435012450095165e-06,
+      "loss": 1.041,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9433111349543515e-06,
+      "loss": 0.8667,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.943120708640615e-06,
+      "loss": 0.9762,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.942929966088769e-06,
+      "loss": 0.8367,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.942738907319308e-06,
+      "loss": 1.2601,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.942547532352761e-06,
+      "loss": 1.1113,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.942355841209692e-06,
+      "loss": 0.9258,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.942163833910697e-06,
+      "loss": 1.1866,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.941971510476409e-06,
+      "loss": 1.3457,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.941778870927491e-06,
+      "loss": 0.8497,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.941585915284644e-06,
+      "loss": 0.9485,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.941392643568601e-06,
+      "loss": 0.8543,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.941199055800127e-06,
+      "loss": 1.0481,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.941005152000025e-06,
+      "loss": 0.8876,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.940810932189129e-06,
+      "loss": 1.1715,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.940616396388309e-06,
+      "loss": 0.9594,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.940421544618466e-06,
+      "loss": 1.0175,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.940226376900539e-06,
+      "loss": 1.1028,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9400308932554965e-06,
+      "loss": 1.4146,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.939835093704344e-06,
+      "loss": 0.7309,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.939638978268121e-06,
+      "loss": 1.001,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9394425469679005e-06,
+      "loss": 0.8115,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.939245799824787e-06,
+      "loss": 0.8122,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.939048736859923e-06,
+      "loss": 0.8703,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9388513580944806e-06,
+      "loss": 0.8944,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.93865366354967e-06,
+      "loss": 1.3751,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.938455653246733e-06,
+      "loss": 0.8435,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.938257327206946e-06,
+      "loss": 1.1464,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.938058685451618e-06,
+      "loss": 1.0687,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9378597280020955e-06,
+      "loss": 0.9041,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9376604548797545e-06,
+      "loss": 1.1421,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.937460866106007e-06,
+      "loss": 1.0545,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.937260961702299e-06,
+      "loss": 0.9779,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.937060741690109e-06,
+      "loss": 0.9814,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.936860206090953e-06,
+      "loss": 0.9644,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.936659354926378e-06,
+      "loss": 0.9857,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.936458188217964e-06,
+      "loss": 1.0309,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.936256705987327e-06,
+      "loss": 1.0658,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9360549082561175e-06,
+      "loss": 0.8577,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.935852795046017e-06,
+      "loss": 1.1155,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.935650366378743e-06,
+      "loss": 0.9284,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.935447622276047e-06,
+      "loss": 0.8834,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.935244562759714e-06,
+      "loss": 1.0205,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.935041187851561e-06,
+      "loss": 0.9797,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.934837497573442e-06,
+      "loss": 0.9251,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.934633491947243e-06,
+      "loss": 0.9555,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.934429170994886e-06,
+      "loss": 1.0192,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.934224534738323e-06,
+      "loss": 0.8867,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.934019583199544e-06,
+      "loss": 0.9041,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.933814316400569e-06,
+      "loss": 0.8963,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.933608734363456e-06,
+      "loss": 1.1032,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.933402837110293e-06,
+      "loss": 0.8737,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9331966246632054e-06,
+      "loss": 1.155,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9329900970443495e-06,
+      "loss": 1.0075,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9327832542759165e-06,
+      "loss": 1.0866,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.932576096380132e-06,
+      "loss": 0.7347,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.932368623379256e-06,
+      "loss": 1.1814,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.93216083529558e-06,
+      "loss": 0.887,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.931952732151432e-06,
+      "loss": 0.6585,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.931744313969172e-06,
+      "loss": 0.9091,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9315355807711945e-06,
+      "loss": 0.8395,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.931326532579928e-06,
+      "loss": 1.2051,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.931117169417834e-06,
+      "loss": 1.2323,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.930907491307411e-06,
+      "loss": 1.1969,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.930697498271187e-06,
+      "loss": 1.0681,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.930487190331724e-06,
+      "loss": 1.2469,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.930276567511623e-06,
+      "loss": 0.8429,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.930065629833514e-06,
+      "loss": 1.0107,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.929854377320062e-06,
+      "loss": 0.6646,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.929642809993966e-06,
+      "loss": 1.2231,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.929430927877959e-06,
+      "loss": 0.9669,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.929218730994809e-06,
+      "loss": 0.9625,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.929006219367313e-06,
+      "loss": 1.0141,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.92879339301831e-06,
+      "loss": 0.9143,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.928580251970665e-06,
+      "loss": 1.2122,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.92836679624728e-06,
+      "loss": 0.9764,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.928153025871093e-06,
+      "loss": 0.9091,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.927938940865073e-06,
+      "loss": 1.0914,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.927724541252222e-06,
+      "loss": 0.9354,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.927509827055577e-06,
+      "loss": 1.0023,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.927294798298211e-06,
+      "loss": 1.1626,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.927079455003227e-06,
+      "loss": 1.1872,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.926863797193765e-06,
+      "loss": 1.106,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.926647824892996e-06,
+      "loss": 1.0565,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.926431538124129e-06,
+      "loss": 0.994,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.926214936910401e-06,
+      "loss": 1.2386,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.925998021275087e-06,
+      "loss": 0.9908,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.925780791241494e-06,
+      "loss": 1.1369,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.925563246832964e-06,
+      "loss": 0.8871,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.925345388072872e-06,
+      "loss": 0.9397,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.925127214984626e-06,
+      "loss": 1.0413,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.92490872759167e-06,
+      "loss": 1.0225,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.924689925917481e-06,
+      "loss": 0.8197,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.924470809985566e-06,
+      "loss": 0.9611,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.924251379819473e-06,
+      "loss": 0.912,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.924031635442776e-06,
+      "loss": 0.8961,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.92381157687909e-06,
+      "loss": 1.1176,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9235912041520596e-06,
+      "loss": 1.0073,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.923370517285362e-06,
+      "loss": 1.1146,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.92314951630271e-06,
+      "loss": 0.9527,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.922928201227852e-06,
+      "loss": 0.9573,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.922706572084567e-06,
+      "loss": 0.8405,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.922484628896669e-06,
+      "loss": 1.1182,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.922262371688007e-06,
+      "loss": 1.1089,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.922039800482461e-06,
+      "loss": 0.8844,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.921816915303948e-06,
+      "loss": 0.937,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.921593716176414e-06,
+      "loss": 1.1841,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.921370203123845e-06,
+      "loss": 0.9058,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.921146376170257e-06,
+      "loss": 0.8316,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.920922235339697e-06,
+      "loss": 0.8517,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.920697780656253e-06,
+      "loss": 1.0168,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.920473012144041e-06,
+      "loss": 1.2233,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.920247929827212e-06,
+      "loss": 1.1049,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.920022533729951e-06,
+      "loss": 0.9317,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.919796823876477e-06,
+      "loss": 0.8863,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.919570800291042e-06,
+      "loss": 0.7835,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.919344462997934e-06,
+      "loss": 0.8907,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.919117812021472e-06,
+      "loss": 0.9031,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.918890847386009e-06,
+      "loss": 0.705,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.918663569115933e-06,
+      "loss": 0.9496,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.918435977235663e-06,
+      "loss": 0.908,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.918208071769656e-06,
+      "loss": 0.7387,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.917979852742402e-06,
+      "loss": 1.1984,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.91775132017842e-06,
+      "loss": 0.8959,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.917522474102266e-06,
+      "loss": 0.9789,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.917293314538531e-06,
+      "loss": 1.2681,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9170638415118375e-06,
+      "loss": 0.9394,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.916834055046843e-06,
+      "loss": 0.9508,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.916603955168236e-06,
+      "loss": 1.0739,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.916373541900744e-06,
+      "loss": 0.9681,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.916142815269122e-06,
+      "loss": 0.8433,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.915911775298162e-06,
+      "loss": 0.9494,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9156804220126906e-06,
+      "loss": 0.9702,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9154487554375654e-06,
+      "loss": 1.0904,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.91521677559768e-06,
+      "loss": 1.1456,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.914984482517959e-06,
+      "loss": 0.9152,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.914751876223364e-06,
+      "loss": 0.915,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9145189567388885e-06,
+      "loss": 1.0844,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.914285724089559e-06,
+      "loss": 0.9461,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9140521783004365e-06,
+      "loss": 0.8494,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.913818319396615e-06,
+      "loss": 0.978,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.913584147403222e-06,
+      "loss": 1.0596,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.913349662345422e-06,
+      "loss": 0.9942,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9131148642484076e-06,
+      "loss": 0.9661,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.912879753137409e-06,
+      "loss": 1.0946,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.91264432903769e-06,
+      "loss": 1.2015,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.912408591974544e-06,
+      "loss": 1.0376,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9121725419733045e-06,
+      "loss": 0.8373,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9119361790593306e-06,
+      "loss": 0.8354,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.911699503258024e-06,
+      "loss": 0.6637,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.911462514594813e-06,
+      "loss": 0.9489,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.911225213095163e-06,
+      "loss": 0.8387,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.910987598784571e-06,
+      "loss": 1.0826,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.910749671688569e-06,
+      "loss": 0.9352,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.910511431832723e-06,
+      "loss": 1.0935,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.910272879242631e-06,
+      "loss": 0.9465,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9100340139439265e-06,
+      "loss": 0.9386,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.909794835962274e-06,
+      "loss": 1.0466,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.909555345323375e-06,
+      "loss": 1.0514,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.909315542052961e-06,
+      "loss": 0.9349,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9090754261768e-06,
+      "loss": 0.8714,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.908834997720692e-06,
+      "loss": 0.8605,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.90859425671047e-06,
+      "loss": 0.9651,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.908353203172004e-06,
+      "loss": 1.0957,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9081118371311926e-06,
+      "loss": 0.954,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.907870158613974e-06,
+      "loss": 0.8664,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.907628167646313e-06,
+      "loss": 0.9462,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.907385864254212e-06,
+      "loss": 1.0669,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.907143248463708e-06,
+      "loss": 1.05,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.90690032030087e-06,
+      "loss": 0.9395,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.906657079791799e-06,
+      "loss": 1.0652,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9064135269626335e-06,
+      "loss": 0.9278,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.906169661839541e-06,
+      "loss": 0.9545,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.905925484448727e-06,
+      "loss": 1.0064,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.905680994816425e-06,
+      "loss": 0.9739,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9054361929689085e-06,
+      "loss": 0.7654,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.905191078932481e-06,
+      "loss": 0.9001,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.904945652733478e-06,
+      "loss": 0.8613,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.904699914398272e-06,
+      "loss": 1.0806,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.904453863953269e-06,
+      "loss": 1.0699,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.904207501424905e-06,
+      "loss": 0.9156,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.903960826839652e-06,
+      "loss": 0.8991,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.903713840224014e-06,
+      "loss": 1.1375,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.903466541604533e-06,
+      "loss": 0.7076,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.903218931007779e-06,
+      "loss": 0.796,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.902971008460357e-06,
+      "loss": 0.9846,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.902722773988908e-06,
+      "loss": 1.2552,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.902474227620104e-06,
+      "loss": 1.0545,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9022253693806515e-06,
+      "loss": 0.7086,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.90197619929729e-06,
+      "loss": 1.1663,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9017267173967934e-06,
+      "loss": 0.8925,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.901476923705968e-06,
+      "loss": 0.9389,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9012268182516535e-06,
+      "loss": 1.0029,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.900976401060724e-06,
+      "loss": 0.6915,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9007256721600884e-06,
+      "loss": 0.9245,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.900474631576686e-06,
+      "loss": 1.0054,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.900223279337491e-06,
+      "loss": 1.1444,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.899971615469512e-06,
+      "loss": 1.1102,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8997196399997905e-06,
+      "loss": 0.825,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.899467352955399e-06,
+      "loss": 1.071,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.899214754363449e-06,
+      "loss": 0.8469,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.89896184425108e-06,
+      "loss": 1.0754,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.898708622645467e-06,
+      "loss": 0.9485,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8984550895738195e-06,
+      "loss": 1.1359,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.89820124506338e-06,
+      "loss": 1.0604,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.897947089141424e-06,
+      "loss": 0.9478,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.897692621835258e-06,
+      "loss": 0.9235,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.897437843172229e-06,
+      "loss": 1.0296,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.897182753179709e-06,
+      "loss": 1.2143,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.896927351885109e-06,
+      "loss": 0.839,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.896671639315872e-06,
+      "loss": 1.2925,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8964156154994746e-06,
+      "loss": 0.971,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.896159280463425e-06,
+      "loss": 0.9567,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.895902634235268e-06,
+      "loss": 1.0743,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.89564567684258e-06,
+      "loss": 1.1548,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.895388408312969e-06,
+      "loss": 0.8426,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.895130828674082e-06,
+      "loss": 1.0337,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8948729379535936e-06,
+      "loss": 0.8381,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.894614736179214e-06,
+      "loss": 0.9898,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.894356223378688e-06,
+      "loss": 0.9441,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.894097399579792e-06,
+      "loss": 1.2999,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.893838264810338e-06,
+      "loss": 0.9017,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.893578819098167e-06,
+      "loss": 1.0466,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.893319062471159e-06,
+      "loss": 1.0647,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.893058994957226e-06,
+      "loss": 0.846,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8927986165843085e-06,
+      "loss": 0.9282,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.892537927380386e-06,
+      "loss": 0.9183,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.892276927373471e-06,
+      "loss": 1.1419,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.892015616591606e-06,
+      "loss": 0.8593,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.891753995062869e-06,
+      "loss": 0.8307,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.891492062815372e-06,
+      "loss": 0.9788,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.891229819877258e-06,
+      "loss": 1.0037,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.890967266276708e-06,
+      "loss": 1.0128,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8907044020419305e-06,
+      "loss": 0.8747,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.890441227201172e-06,
+      "loss": 1.1252,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.890177741782708e-06,
+      "loss": 0.7478,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8899139458148535e-06,
+      "loss": 0.9022,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.889649839325952e-06,
+      "loss": 1.1451,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8893854223443805e-06,
+      "loss": 1.0125,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.889120694898552e-06,
+      "loss": 0.9068,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.888855657016912e-06,
+      "loss": 0.8344,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.888590308727936e-06,
+      "loss": 0.8988,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.888324650060138e-06,
+      "loss": 1.1404,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.888058681042063e-06,
+      "loss": 0.7589,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.887792401702288e-06,
+      "loss": 1.0497,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.887525812069426e-06,
+      "loss": 0.9908,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.887258912172122e-06,
+      "loss": 1.0755,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.886991702039053e-06,
+      "loss": 0.9894,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.886724181698933e-06,
+      "loss": 0.9333,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.886456351180504e-06,
+      "loss": 0.7531,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.886188210512546e-06,
+      "loss": 0.8196,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.885919759723872e-06,
+      "loss": 0.9511,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.885650998843325e-06,
+      "loss": 0.8569,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.885381927899784e-06,
+      "loss": 1.0847,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.885112546922162e-06,
+      "loss": 0.9305,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.884842855939402e-06,
+      "loss": 1.1224,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8845728549804825e-06,
+      "loss": 0.9977,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.884302544074416e-06,
+      "loss": 1.1268,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.884031923250247e-06,
+      "loss": 1.1065,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.883760992537053e-06,
+      "loss": 0.9434,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8834897519639475e-06,
+      "loss": 0.8939,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.883218201560072e-06,
+      "loss": 0.8315,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.882946341354607e-06,
+      "loss": 0.9025,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.8826741713767645e-06,
+      "loss": 1.0897,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.882401691655787e-06,
+      "loss": 1.0116,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8821289022209545e-06,
+      "loss": 0.8393,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8818558031015764e-06,
+      "loss": 0.9722,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.881582394326998e-06,
+      "loss": 1.017,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.881308675926597e-06,
+      "loss": 1.1734,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.881034647929784e-06,
+      "loss": 0.8119,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.880760310366003e-06,
+      "loss": 0.8195,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.880485663264733e-06,
+      "loss": 1.1075,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.880210706655485e-06,
+      "loss": 1.0282,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.879935440567801e-06,
+      "loss": 1.0808,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.879659865031259e-06,
+      "loss": 0.7812,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.879383980075471e-06,
+      "loss": 1.2825,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.87910778573008e-06,
+      "loss": 0.9601,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8788312820247614e-06,
+      "loss": 0.7489,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.878554468989228e-06,
+      "loss": 1.0556,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.878277346653222e-06,
+      "loss": 0.8035,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.877999915046522e-06,
+      "loss": 0.9809,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.877722174198934e-06,
+      "loss": 0.8646,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.877444124140305e-06,
+      "loss": 0.8263,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8771657649005105e-06,
+      "loss": 1.0161,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.87688709650946e-06,
+      "loss": 1.115,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.876608118997095e-06,
+      "loss": 0.937,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.876328832393394e-06,
+      "loss": 0.9749,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.876049236728364e-06,
+      "loss": 0.8744,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.87576933203205e-06,
+      "loss": 0.9913,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.875489118334526e-06,
+      "loss": 1.0588,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.875208595665902e-06,
+      "loss": 1.0019,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.874927764056319e-06,
+      "loss": 0.9127,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.874646623535952e-06,
+      "loss": 1.0317,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.874365174135011e-06,
+      "loss": 1.0287,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.874083415883738e-06,
+      "loss": 0.903,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.873801348812407e-06,
+      "loss": 0.8883,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.873518972951326e-06,
+      "loss": 0.9026,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.873236288330836e-06,
+      "loss": 0.9646,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.872953294981312e-06,
+      "loss": 0.9352,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.872669992933161e-06,
+      "loss": 1.099,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.872386382216825e-06,
+      "loss": 0.9414,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.872102462862777e-06,
+      "loss": 0.9201,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.871818234901524e-06,
+      "loss": 0.9896,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.871533698363606e-06,
+      "loss": 0.9712,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.871248853279599e-06,
+      "loss": 0.9275,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8709636996801055e-06,
+      "loss": 1.1458,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.870678237595768e-06,
+      "loss": 0.9368,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.87039246705726e-06,
+      "loss": 0.9559,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.870106388095285e-06,
+      "loss": 1.1755,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.869820000740584e-06,
+      "loss": 1.0806,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.869533305023928e-06,
+      "loss": 0.9337,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.869246300976124e-06,
+      "loss": 1.0708,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.868958988628008e-06,
+      "loss": 1.1363,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.868671368010455e-06,
+      "loss": 1.0252,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.868383439154368e-06,
+      "loss": 0.828,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.868095202090685e-06,
+      "loss": 0.9052,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8678066568503766e-06,
+      "loss": 1.1019,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.867517803464449e-06,
+      "loss": 1.1023,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8672286419639366e-06,
+      "loss": 0.8174,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.866939172379911e-06,
+      "loss": 1.0403,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.866649394743477e-06,
+      "loss": 1.0235,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8663593090857705e-06,
+      "loss": 0.9484,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.866068915437959e-06,
+      "loss": 1.1613,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.865778213831249e-06,
+      "loss": 0.9908,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.865487204296873e-06,
+      "loss": 0.7886,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.865195886866101e-06,
+      "loss": 1.1963,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8649042615702375e-06,
+      "loss": 1.0187,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.864612328440614e-06,
+      "loss": 1.1205,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8643200875086005e-06,
+      "loss": 0.848,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.864027538805598e-06,
+      "loss": 1.2144,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.863734682363041e-06,
+      "loss": 0.8916,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8634415182123966e-06,
+      "loss": 0.8845,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.863148046385165e-06,
+      "loss": 1.1424,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.862854266912881e-06,
+      "loss": 0.9103,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.86256017982711e-06,
+      "loss": 0.9764,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.862265785159451e-06,
+      "loss": 1.11,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.861971082941539e-06,
+      "loss": 0.9905,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.861676073205037e-06,
+      "loss": 1.325,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.861380755981647e-06,
+      "loss": 0.9543,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.861085131303097e-06,
+      "loss": 1.1177,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.860789199201155e-06,
+      "loss": 0.9323,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.860492959707618e-06,
+      "loss": 1.1,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.860196412854315e-06,
+      "loss": 0.9953,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.859899558673113e-06,
+      "loss": 1.0804,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.859602397195907e-06,
+      "loss": 1.0198,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.859304928454627e-06,
+      "loss": 0.9311,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.859007152481237e-06,
+      "loss": 1.0267,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.858709069307732e-06,
+      "loss": 0.9001,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.858410678966142e-06,
+      "loss": 0.8055,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.858111981488529e-06,
+      "loss": 0.7812,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8578129769069865e-06,
+      "loss": 1.0499,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.857513665253644e-06,
+      "loss": 1.0347,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.857214046560663e-06,
+      "loss": 1.1178,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.856914120860235e-06,
+      "loss": 1.0785,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.856613888184589e-06,
+      "loss": 1.0662,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.856313348565986e-06,
+      "loss": 0.7728,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.856012502036715e-06,
+      "loss": 1.1729,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.855711348629106e-06,
+      "loss": 0.9995,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.855409888375516e-06,
+      "loss": 1.0877,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.855108121308338e-06,
+      "loss": 0.7771,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8548060474599955e-06,
+      "loss": 1.0469,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.854503666862947e-06,
+      "loss": 0.9618,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.854200979549683e-06,
+      "loss": 1.0187,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.853897985552728e-06,
+      "loss": 0.8451,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.853594684904637e-06,
+      "loss": 1.1464,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.853291077638001e-06,
+      "loss": 1.0543,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.852987163785443e-06,
+      "loss": 0.9622,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.852682943379618e-06,
+      "loss": 0.9454,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.852378416453213e-06,
+      "loss": 1.0055,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.8520735830389515e-06,
+      "loss": 0.9348,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.8517684431695874e-06,
+      "loss": 1.1165,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.851462996877907e-06,
+      "loss": 0.9761,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.851157244196732e-06,
+      "loss": 1.0265,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.850851185158913e-06,
+      "loss": 0.8383,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.8505448197973386e-06,
+      "loss": 0.7758,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.850238148144927e-06,
+      "loss": 1.0241,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0405,
+      "eval_samples_per_second": 352.05,
+      "eval_steps_per_second": 44.015,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.84993117023463e-06,
+      "loss": 1.1191,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.84962388609943e-06,
+      "loss": 0.9286,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.849316295772349e-06,
+      "loss": 0.8805,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.849008399286434e-06,
+      "loss": 0.8548,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.848700196674771e-06,
+      "loss": 1.1708,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.848391687970474e-06,
+      "loss": 0.9909,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.848082873206694e-06,
+      "loss": 0.9793,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.847773752416613e-06,
+      "loss": 0.9137,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.847464325633445e-06,
+      "loss": 0.9185,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.847154592890438e-06,
+      "loss": 1.0124,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.846844554220874e-06,
+      "loss": 1.1031,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.846534209658064e-06,
+      "loss": 0.9711,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.846223559235359e-06,
+      "loss": 1.092,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.8459126029861335e-06,
+      "loss": 1.0362,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.845601340943803e-06,
+      "loss": 0.9165,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.845289773141811e-06,
+      "loss": 0.7596,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.844977899613636e-06,
+      "loss": 1.0927,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.844665720392788e-06,
+      "loss": 0.8265,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.844353235512812e-06,
+      "loss": 0.877,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.844040445007284e-06,
+      "loss": 0.7953,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.8437273489098135e-06,
+      "loss": 1.0665,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.843413947254041e-06,
+      "loss": 0.9571,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.843100240073644e-06,
+      "loss": 0.9639,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.842786227402328e-06,
+      "loss": 0.8448,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.842471909273836e-06,
+      "loss": 1.0678,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.8421572857219386e-06,
+      "loss": 0.811,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.841842356780446e-06,
+      "loss": 0.8327,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.841527122483193e-06,
+      "loss": 0.8476,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.841211582864053e-06,
+      "loss": 1.0481,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.840895737956933e-06,
+      "loss": 1.0231,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.840579587795768e-06,
+      "loss": 1.0444,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.840263132414529e-06,
+      "loss": 1.0048,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.839946371847219e-06,
+      "loss": 1.1287,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.839629306127873e-06,
+      "loss": 1.0028,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.839311935290563e-06,
+      "loss": 0.7991,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.838994259369386e-06,
+      "loss": 0.9016,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.838676278398479e-06,
+      "loss": 0.9892,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.838357992412009e-06,
+      "loss": 0.8376,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.838039401444175e-06,
+      "loss": 1.1115,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8377205055292094e-06,
+      "loss": 0.8653,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.837401304701379e-06,
+      "loss": 1.0482,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.83708179899498e-06,
+      "loss": 1.1358,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.836761988444344e-06,
+      "loss": 0.8272,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8364418730838346e-06,
+      "loss": 0.8071,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.836121452947849e-06,
+      "loss": 0.9094,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8358007280708155e-06,
+      "loss": 1.0294,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.835479698487196e-06,
+      "loss": 0.9591,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8351583642314854e-06,
+      "loss": 0.9853,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8348367253382115e-06,
+      "loss": 1.1801,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.834514781841933e-06,
+      "loss": 0.7398,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.834192533777244e-06,
+      "loss": 0.9428,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8338699811787684e-06,
+      "loss": 0.8812,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.833547124081166e-06,
+      "loss": 0.8458,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.833223962519128e-06,
+      "loss": 0.947,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.832900496527376e-06,
+      "loss": 0.8629,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.83257672614067e-06,
+      "loss": 0.8939,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8322526513937944e-06,
+      "loss": 1.0234,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.831928272321575e-06,
+      "loss": 1.0615,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.831603588958865e-06,
+      "loss": 0.9492,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.831278601340551e-06,
+      "loss": 0.8182,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8309533095015524e-06,
+      "loss": 0.8909,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.830627713476824e-06,
+      "loss": 0.8554,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.83030181330135e-06,
+      "loss": 0.9429,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.829975609010147e-06,
+      "loss": 0.9133,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.829649100638267e-06,
+      "loss": 0.856,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.829322288220794e-06,
+      "loss": 0.9694,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.828995171792842e-06,
+      "loss": 1.0192,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.82866775138956e-06,
+      "loss": 1.3357,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.828340027046132e-06,
+      "loss": 0.8133,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.828011998797768e-06,
+      "loss": 0.9807,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8276836666797185e-06,
+      "loss": 1.3123,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.827355030727259e-06,
+      "loss": 1.0884,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.827026090975705e-06,
+      "loss": 1.0509,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.826696847460397e-06,
+      "loss": 0.7968,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.826367300216715e-06,
+      "loss": 0.9767,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.826037449280069e-06,
+      "loss": 0.9132,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.825707294685901e-06,
+      "loss": 0.9952,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.825376836469685e-06,
+      "loss": 1.0592,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.82504607466693e-06,
+      "loss": 1.088,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.824715009313176e-06,
+      "loss": 1.0191,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8243836404439965e-06,
+      "loss": 0.8266,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.824051968094995e-06,
+      "loss": 1.0313,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.823719992301812e-06,
+      "loss": 0.9433,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8233877131001166e-06,
+      "loss": 1.0096,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8230551305256134e-06,
+      "loss": 0.893,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.822722244614038e-06,
+      "loss": 1.0082,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8223890554011575e-06,
+      "loss": 1.1905,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8220555629227756e-06,
+      "loss": 0.8771,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.821721767214725e-06,
+      "loss": 0.9725,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8213876683128715e-06,
+      "loss": 0.8936,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.821053266253115e-06,
+      "loss": 0.9213,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.820718561071386e-06,
+      "loss": 0.8797,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.820383552803649e-06,
+      "loss": 0.9489,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.820048241485899e-06,
+      "loss": 0.9349,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.819712627154168e-06,
+      "loss": 0.9401,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.819376709844516e-06,
+      "loss": 1.0598,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.819040489593039e-06,
+      "loss": 1.0598,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.81870396643586e-06,
+      "loss": 0.8915,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.818367140409143e-06,
+      "loss": 1.0263,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.818030011549078e-06,
+      "loss": 1.0763,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.817692579891888e-06,
+      "loss": 1.0323,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.817354845473833e-06,
+      "loss": 0.9518,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8170168083311996e-06,
+      "loss": 1.0452,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.816678468500311e-06,
+      "loss": 0.8417,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8163398260175225e-06,
+      "loss": 0.7488,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.816000880919221e-06,
+      "loss": 0.7844,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8156616332418255e-06,
+      "loss": 0.9908,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.815322083021788e-06,
+      "loss": 0.9178,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.814982230295594e-06,
+      "loss": 0.9423,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.814642075099761e-06,
+      "loss": 0.8646,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.814301617470836e-06,
+      "loss": 1.0985,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.813960857445404e-06,
+      "loss": 1.0272,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.813619795060079e-06,
+      "loss": 1.0584,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8132784303515074e-06,
+      "loss": 0.9315,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.81293676335637e-06,
+      "loss": 1.0742,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.812594794111377e-06,
+      "loss": 0.8728,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.812252522653274e-06,
+      "loss": 1.0212,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8119099490188395e-06,
+      "loss": 1.4132,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8115670732448805e-06,
+      "loss": 0.8776,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.81122389536824e-06,
+      "loss": 0.9742,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.810880415425794e-06,
+      "loss": 1.2125,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.810536633454447e-06,
+      "loss": 0.867,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8101925494911375e-06,
+      "loss": 1.2472,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.809848163572842e-06,
+      "loss": 0.8118,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.80950347573656e-06,
+      "loss": 1.2535,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.809158486019331e-06,
+      "loss": 0.9405,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8088131944582215e-06,
+      "loss": 0.8118,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.808467601090337e-06,
+      "loss": 1.2759,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.808121705952808e-06,
+      "loss": 0.867,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.807775509082801e-06,
+      "loss": 0.826,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.807429010517517e-06,
+      "loss": 1.0415,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.807082210294187e-06,
+      "loss": 1.0759,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.806735108450073e-06,
+      "loss": 1.1105,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.806387705022472e-06,
+      "loss": 1.0285,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.806040000048712e-06,
+      "loss": 1.0037,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.805691993566155e-06,
+      "loss": 0.708,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.805343685612192e-06,
+      "loss": 0.9751,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.804995076224252e-06,
+      "loss": 0.9608,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.8046461654397905e-06,
+      "loss": 0.8493,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.804296953296299e-06,
+      "loss": 1.0242,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.803947439831299e-06,
+      "loss": 0.947,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.803597625082348e-06,
+      "loss": 1.0318,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.80324750908703e-06,
+      "loss": 0.9022,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.80289709188297e-06,
+      "loss": 0.8594,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.802546373507815e-06,
+      "loss": 1.0755,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.802195353999254e-06,
+      "loss": 0.9491,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.801844033395002e-06,
+      "loss": 1.0117,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.801492411732807e-06,
+      "loss": 0.919,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.8011404890504536e-06,
+      "loss": 0.9064,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.800788265385754e-06,
+      "loss": 1.1277,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.800435740776555e-06,
+      "loss": 0.9631,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.800082915260736e-06,
+      "loss": 1.1648,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.799729788876207e-06,
+      "loss": 0.8331,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.799376361660913e-06,
+      "loss": 0.8657,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.799022633652828e-06,
+      "loss": 0.8781,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.798668604889961e-06,
+      "loss": 0.8884,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.798314275410351e-06,
+      "loss": 0.8026,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.797959645252074e-06,
+      "loss": 0.8634,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.7976047144532305e-06,
+      "loss": 0.8528,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.7972494830519614e-06,
+      "loss": 1.106,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.796893951086435e-06,
+      "loss": 1.2337,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.796538118594852e-06,
+      "loss": 1.0184,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.796181985615449e-06,
+      "loss": 0.9233,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.79582555218649e-06,
+      "loss": 1.009,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.795468818346276e-06,
+      "loss": 1.2614,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.795111784133137e-06,
+      "loss": 0.9966,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.7947544495854355e-06,
+      "loss": 1.0682,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.794396814741568e-06,
+      "loss": 0.908,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.794038879639962e-06,
+      "loss": 1.3163,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.793680644319078e-06,
+      "loss": 0.8533,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.793322108817408e-06,
+      "loss": 1.0113,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.792963273173477e-06,
+      "loss": 0.9631,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.792604137425841e-06,
+      "loss": 0.7872,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.7922447016130905e-06,
+      "loss": 0.8505,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.791884965773846e-06,
+      "loss": 0.9717,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.79152492994676e-06,
+      "loss": 0.9567,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.791164594170521e-06,
+      "loss": 1.1013,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7908039584838446e-06,
+      "loss": 1.0333,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.790443022925483e-06,
+      "loss": 0.8177,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.790081787534217e-06,
+      "loss": 0.9923,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7897202523488615e-06,
+      "loss": 1.0255,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.789358417408264e-06,
+      "loss": 1.0225,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.788996282751304e-06,
+      "loss": 0.9937,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.788633848416893e-06,
+      "loss": 0.8712,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.788271114443973e-06,
+      "loss": 0.986,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.787908080871521e-06,
+      "loss": 1.0847,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.787544747738546e-06,
+      "loss": 0.914,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.787181115084085e-06,
+      "loss": 0.9976,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7868171829472125e-06,
+      "loss": 0.9926,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.786452951367033e-06,
+      "loss": 1.0915,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.786088420382682e-06,
+      "loss": 0.8889,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.78572359003333e-06,
+      "loss": 0.8812,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7853584603581766e-06,
+      "loss": 1.2079,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.784993031396455e-06,
+      "loss": 0.8723,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7846273031874316e-06,
+      "loss": 0.9157,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.784261275770403e-06,
+      "loss": 0.9028,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.783894949184699e-06,
+      "loss": 0.7473,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.783528323469681e-06,
+      "loss": 0.948,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.783161398664743e-06,
+      "loss": 1.4668,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.782794174809311e-06,
+      "loss": 0.9362,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.782426651942845e-06,
+      "loss": 0.981,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.782058830104833e-06,
+      "loss": 0.9583,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.781690709334798e-06,
+      "loss": 1.0419,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.781322289672296e-06,
+      "loss": 1.0876,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.780953571156911e-06,
+      "loss": 1.0675,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.780584553828264e-06,
+      "loss": 1.1144,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.780215237726004e-06,
+      "loss": 0.9277,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.779845622889816e-06,
+      "loss": 1.1704,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.779475709359414e-06,
+      "loss": 1.0067,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.779105497174546e-06,
+      "loss": 0.8787,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.778734986374991e-06,
+      "loss": 0.9275,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.778364177000559e-06,
+      "loss": 0.936,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.777993069091094e-06,
+      "loss": 1.1035,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.777621662686472e-06,
+      "loss": 0.9816,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.777249957826602e-06,
+      "loss": 0.9697,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.776877954551421e-06,
+      "loss": 0.9033,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.776505652900902e-06,
+      "loss": 0.8793,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.776133052915049e-06,
+      "loss": 1.1023,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.775760154633897e-06,
+      "loss": 0.8613,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.775386958097515e-06,
+      "loss": 0.9521,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.775013463346003e-06,
+      "loss": 1.1608,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.774639670419492e-06,
+      "loss": 0.8457,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.7742655793581455e-06,
+      "loss": 1.0596,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.773891190202161e-06,
+      "loss": 1.0445,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.773516502991767e-06,
+      "loss": 0.8873,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.773141517767222e-06,
+      "loss": 0.9935,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.772766234568818e-06,
+      "loss": 1.1657,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.772390653436881e-06,
+      "loss": 1.0227,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.772014774411765e-06,
+      "loss": 0.875,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.77163859753386e-06,
+      "loss": 0.8304,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.771262122843586e-06,
+      "loss": 0.8788,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.770885350381395e-06,
+      "loss": 0.83,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.77050828018777e-06,
+      "loss": 0.7981,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.770130912303227e-06,
+      "loss": 1.0351,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.769753246768316e-06,
+      "loss": 0.959,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7693752836236166e-06,
+      "loss": 0.7759,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.76899702290974e-06,
+      "loss": 1.1716,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.768618464667332e-06,
+      "loss": 1.1607,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.768239608937066e-06,
+      "loss": 1.1087,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.767860455759653e-06,
+      "loss": 0.8451,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.767481005175831e-06,
+      "loss": 1.0777,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.767101257226372e-06,
+      "loss": 0.8386,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.766721211952081e-06,
+      "loss": 0.8005,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.766340869393793e-06,
+      "loss": 1.1504,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.765960229592376e-06,
+      "loss": 0.9652,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.76557929258873e-06,
+      "loss": 0.893,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.765198058423786e-06,
+      "loss": 0.9775,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7648165271385074e-06,
+      "loss": 0.944,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.764434698773891e-06,
+      "loss": 0.9917,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7640525733709625e-06,
+      "loss": 0.9783,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7636701509707835e-06,
+      "loss": 1.0032,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.763287431614444e-06,
+      "loss": 0.9289,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.762904415343066e-06,
+      "loss": 0.6659,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.762521102197807e-06,
+      "loss": 1.0519,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.762137492219852e-06,
+      "loss": 1.0459,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.76175358545042e-06,
+      "loss": 0.9545,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.761369381930763e-06,
+      "loss": 0.8369,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.760984881702163e-06,
+      "loss": 1.1865,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.760600084805934e-06,
+      "loss": 0.9856,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.760214991283424e-06,
+      "loss": 0.7864,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.759829601176009e-06,
+      "loss": 0.9418,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.759443914525101e-06,
+      "loss": 0.8257,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.759057931372141e-06,
+      "loss": 0.8736,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.758671651758603e-06,
+      "loss": 1.1846,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.758285075725993e-06,
+      "loss": 0.9357,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.757898203315848e-06,
+      "loss": 1.132,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7575110345697384e-06,
+      "loss": 0.7936,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.757123569529265e-06,
+      "loss": 0.9026,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.756735808236061e-06,
+      "loss": 0.857,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.756347750731791e-06,
+      "loss": 0.7675,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.755959397058152e-06,
+      "loss": 1.0567,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7555707472568715e-06,
+      "loss": 0.8697,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.755181801369711e-06,
+      "loss": 0.9349,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.7547925594384625e-06,
+      "loss": 1.029,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.754403021504951e-06,
+      "loss": 1.0046,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.754013187611032e-06,
+      "loss": 0.8283,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7536230577985915e-06,
+      "loss": 1.0234,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.753232632109549e-06,
+      "loss": 0.7838,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.752841910585857e-06,
+      "loss": 1.1348,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7524508932695e-06,
+      "loss": 1.0224,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.752059580202488e-06,
+      "loss": 0.8442,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7516679714268736e-06,
+      "loss": 0.8602,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.75127606698473e-06,
+      "loss": 0.9216,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.75088386691817e-06,
+      "loss": 1.0046,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.750491371269334e-06,
+      "loss": 0.9962,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.750098580080397e-06,
+      "loss": 0.7934,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.749705493393564e-06,
+      "loss": 1.1718,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.749312111251072e-06,
+      "loss": 1.0628,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.748918433695189e-06,
+      "loss": 0.997,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7485244607682185e-06,
+      "loss": 1.0559,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.748130192512489e-06,
+      "loss": 0.7938,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.747735628970367e-06,
+      "loss": 1.2739,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.747340770184247e-06,
+      "loss": 1.2369,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7469456161965585e-06,
+      "loss": 0.9868,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7465501670497585e-06,
+      "loss": 0.8985,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.74615442278634e-06,
+      "loss": 0.6933,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.745758383448824e-06,
+      "loss": 1.133,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.745362049079766e-06,
+      "loss": 1.0458,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.744965419721752e-06,
+      "loss": 0.9643,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.744568495417398e-06,
+      "loss": 0.986,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.7441712762093565e-06,
+      "loss": 0.7875,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.743773762140307e-06,
+      "loss": 1.0358,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.743375953252962e-06,
+      "loss": 1.1892,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.742977849590067e-06,
+      "loss": 0.7553,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.742579451194398e-06,
+      "loss": 1.0576,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.742180758108762e-06,
+      "loss": 1.2428,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.741781770376e-06,
+      "loss": 0.8215,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.741382488038982e-06,
+      "loss": 0.8812,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.740982911140611e-06,
+      "loss": 1.0587,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.740583039723822e-06,
+      "loss": 1.1087,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.740182873831581e-06,
+      "loss": 0.9946,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.739782413506886e-06,
+      "loss": 0.9835,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.739381658792767e-06,
+      "loss": 0.9301,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.738980609732283e-06,
+      "loss": 0.8302,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.738579266368528e-06,
+      "loss": 0.886,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.738177628744626e-06,
+      "loss": 0.922,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.737775696903734e-06,
+      "loss": 1.0018,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.737373470889037e-06,
+      "loss": 1.0193,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.736970950743758e-06,
+      "loss": 0.8838,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.736568136511145e-06,
+      "loss": 1.16,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.736165028234481e-06,
+      "loss": 1.0633,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.73576162595708e-06,
+      "loss": 0.7789,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.735357929722287e-06,
+      "loss": 1.013,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.73495393957348e-06,
+      "loss": 0.7331,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.734549655554067e-06,
+      "loss": 0.9179,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.734145077707489e-06,
+      "loss": 0.905,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7337402060772174e-06,
+      "loss": 0.9427,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.733335040706756e-06,
+      "loss": 0.9035,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.73292958163964e-06,
+      "loss": 1.0931,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.732523828919435e-06,
+      "loss": 1.2399,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.732117782589739e-06,
+      "loss": 0.9637,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.731711442694183e-06,
+      "loss": 1.1295,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.731304809276426e-06,
+      "loss": 0.9329,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.730897882380163e-06,
+      "loss": 0.8749,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7304906620491176e-06,
+      "loss": 0.8793,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.730083148327045e-06,
+      "loss": 0.9065,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.729675341257734e-06,
+      "loss": 0.9268,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.729267240885002e-06,
+      "loss": 1.0619,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.728858847252699e-06,
+      "loss": 1.0314,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.728450160404708e-06,
+      "loss": 0.8773,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7280411803849425e-06,
+      "loss": 1.0221,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.727631907237346e-06,
+      "loss": 0.9704,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.727222341005896e-06,
+      "loss": 0.953,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7268124817346e-06,
+      "loss": 0.8026,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.726402329467497e-06,
+      "loss": 1.0338,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.72599188424866e-06,
+      "loss": 1.1201,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.725581146122188e-06,
+      "loss": 0.829,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.725170115132216e-06,
+      "loss": 1.1731,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.72475879132291e-06,
+      "loss": 1.0117,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.724347174738467e-06,
+      "loss": 1.0752,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.723935265423114e-06,
+      "loss": 1.0349,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.723523063421111e-06,
+      "loss": 1.0779,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.723110568776749e-06,
+      "loss": 1.0484,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7226977815343505e-06,
+      "loss": 0.9359,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.722284701738271e-06,
+      "loss": 1.0503,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.721871329432894e-06,
+      "loss": 1.0998,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7214576646626355e-06,
+      "loss": 0.8949,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7210437074719475e-06,
+      "loss": 0.9159,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.720629457905305e-06,
+      "loss": 0.9796,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.720214916007223e-06,
+      "loss": 1.0845,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.719800081822242e-06,
+      "loss": 0.7977,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.719384955394936e-06,
+      "loss": 0.84,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.718969536769911e-06,
+      "loss": 0.8895,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.718553825991802e-06,
+      "loss": 1.0354,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7181378231052794e-06,
+      "loss": 1.1423,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.717721528155042e-06,
+      "loss": 0.973,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.717304941185821e-06,
+      "loss": 0.9532,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.716888062242376e-06,
+      "loss": 1.0556,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.7164708913695035e-06,
+      "loss": 0.7945,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.716053428612027e-06,
+      "loss": 1.151,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.715635674014804e-06,
+      "loss": 1.0896,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.715217627622722e-06,
+      "loss": 0.8814,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7147992894807e-06,
+      "loss": 0.913,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7143806596336875e-06,
+      "loss": 0.9334,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.713961738126666e-06,
+      "loss": 0.8147,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.71354252500465e-06,
+      "loss": 0.8078,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7131230203126835e-06,
+      "loss": 0.932,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.712703224095841e-06,
+      "loss": 1.1714,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.712283136399233e-06,
+      "loss": 0.9776,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.711862757267993e-06,
+      "loss": 0.8477,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.711442086747295e-06,
+      "loss": 1.2616,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.711021124882338e-06,
+      "loss": 0.9116,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.710599871718354e-06,
+      "loss": 0.9108,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7101783273006076e-06,
+      "loss": 0.8877,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.709756491674392e-06,
+      "loss": 1.0946,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.709334364885037e-06,
+      "loss": 1.0094,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.708911946977897e-06,
+      "loss": 0.8305,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.708489237998361e-06,
+      "loss": 0.7183,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.70806623799185e-06,
+      "loss": 1.0361,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.707642947003815e-06,
+      "loss": 0.915,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.707219365079738e-06,
+      "loss": 1.1871,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7067954922651335e-06,
+      "loss": 0.7822,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7063713286055475e-06,
+      "loss": 1.109,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.705946874146555e-06,
+      "loss": 0.9688,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.705522128933764e-06,
+      "loss": 0.8815,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.705097093012813e-06,
+      "loss": 1.0187,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.704671766429372e-06,
+      "loss": 0.9128,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.704246149229143e-06,
+      "loss": 1.0993,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.703820241457859e-06,
+      "loss": 0.8005,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.703394043161282e-06,
+      "loss": 0.9795,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.702967554385209e-06,
+      "loss": 0.922,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.702540775175464e-06,
+      "loss": 1.0006,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.702113705577906e-06,
+      "loss": 0.8622,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.701686345638423e-06,
+      "loss": 1.0162,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.701258695402935e-06,
+      "loss": 1.2302,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.7008307549173935e-06,
+      "loss": 0.8502,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.70040252422778e-06,
+      "loss": 0.9586,
+      "step": 1407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.699974003380107e-06,
+      "loss": 1.107,
+      "step": 1408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.699545192420421e-06,
+      "loss": 0.8121,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.699116091394796e-06,
+      "loss": 1.0965,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.69868670034934e-06,
+      "loss": 1.1325,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.6982570193301906e-06,
+      "loss": 0.9149,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.697827048383517e-06,
+      "loss": 0.9861,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.697396787555521e-06,
+      "loss": 0.9495,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.69696623689243e-06,
+      "loss": 0.8327,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.696535396440511e-06,
+      "loss": 0.8489,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.696104266246055e-06,
+      "loss": 1.0336,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.695672846355389e-06,
+      "loss": 0.9696,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.695241136814868e-06,
+      "loss": 0.6807,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6948091376708786e-06,
+      "loss": 1.1509,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.69437684896984e-06,
+      "loss": 0.8675,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.693944270758201e-06,
+      "loss": 0.8921,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.693511403082442e-06,
+      "loss": 0.8315,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.693078245989076e-06,
+      "loss": 1.0159,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.692644799524644e-06,
+      "loss": 0.8898,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6922110637357204e-06,
+      "loss": 1.0466,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.69177703866891e-06,
+      "loss": 0.7908,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.691342724370848e-06,
+      "loss": 1.2833,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6909081208882045e-06,
+      "loss": 0.9524,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.690473228267674e-06,
+      "loss": 0.896,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.690038046555987e-06,
+      "loss": 0.7239,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.689602575799904e-06,
+      "loss": 0.9884,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.689166816046217e-06,
+      "loss": 1.0005,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.688730767341747e-06,
+      "loss": 0.9108,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.688294429733347e-06,
+      "loss": 0.9662,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.687857803267904e-06,
+      "loss": 1.0749,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6874208879923305e-06,
+      "loss": 0.9167,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6869836839535754e-06,
+      "loss": 1.001,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.686546191198615e-06,
+      "loss": 1.0438,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6861084097744574e-06,
+      "loss": 1.1308,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.685670339728143e-06,
+      "loss": 1.0159,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.685231981106742e-06,
+      "loss": 0.9382,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.684793333957357e-06,
+      "loss": 1.0651,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.68435439832712e-06,
+      "loss": 0.9745,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6839151742631936e-06,
+      "loss": 0.937,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6834756618127734e-06,
+      "loss": 1.0221,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.683035861023085e-06,
+      "loss": 1.2318,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.682595771941385e-06,
+      "loss": 1.256,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6821553946149605e-06,
+      "loss": 0.9983,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.68171472909113e-06,
+      "loss": 1.0316,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.681273775417244e-06,
+      "loss": 0.9687,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.680832533640682e-06,
+      "loss": 1.1146,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.680391003808855e-06,
+      "loss": 0.9232,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.679949185969206e-06,
+      "loss": 0.9343,
+      "step": 1454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.679507080169208e-06,
+      "loss": 0.9131,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6790646864563665e-06,
+      "loss": 1.0548,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6786220048782155e-06,
+      "loss": 1.089,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.67817903548232e-06,
+      "loss": 0.7688,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6777357783162786e-06,
+      "loss": 0.8813,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.677292233427719e-06,
+      "loss": 1.0575,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6768484008643e-06,
+      "loss": 0.8079,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.676404280673712e-06,
+      "loss": 0.9165,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6759598729036745e-06,
+      "loss": 0.9849,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.67551517760194e-06,
+      "loss": 0.8789,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.675070194816289e-06,
+      "loss": 1.0595,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.674624924594538e-06,
+      "loss": 0.9151,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.674179366984529e-06,
+      "loss": 0.8268,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.673733522034139e-06,
+      "loss": 1.0386,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.673287389791271e-06,
+      "loss": 1.1315,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.672840970303866e-06,
+      "loss": 0.7857,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.672394263619888e-06,
+      "loss": 0.9271,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.6719472697873376e-06,
+      "loss": 0.9606,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.671499988854243e-06,
+      "loss": 1.0088,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.671052420868667e-06,
+      "loss": 1.0728,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.670604565878697e-06,
+      "loss": 0.9808,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.670156423932457e-06,
+      "loss": 1.0131,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.669707995078101e-06,
+      "loss": 0.813,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.66925927936381e-06,
+      "loss": 1.1129,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.668810276837801e-06,
+      "loss": 1.2084,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.668360987548318e-06,
+      "loss": 1.0459,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.667911411543637e-06,
+      "loss": 0.7099,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.667461548872066e-06,
+      "loss": 1.1324,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.667011399581941e-06,
+      "loss": 0.9818,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.666560963721632e-06,
+      "loss": 0.9237,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.666110241339538e-06,
+      "loss": 0.8925,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.665659232484088e-06,
+      "loss": 0.9249,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.665207937203744e-06,
+      "loss": 0.9443,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.6647563555469975e-06,
+      "loss": 0.971,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.664304487562371e-06,
+      "loss": 0.9653,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.663852333298417e-06,
+      "loss": 0.7548,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.66339989280372e-06,
+      "loss": 0.8874,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.662947166126894e-06,
+      "loss": 1.0038,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.662494153316586e-06,
+      "loss": 1.1689,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.66204085442147e-06,
+      "loss": 1.0142,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.6615872694902555e-06,
+      "loss": 1.1814,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.661133398571679e-06,
+      "loss": 0.8549,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.660679241714507e-06,
+      "loss": 1.0472,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.660224798967541e-06,
+      "loss": 0.8002,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.659770070379611e-06,
+      "loss": 1.1598,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.659315055999576e-06,
+      "loss": 0.9095,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.658859755876329e-06,
+      "loss": 1.0044,
+      "step": 1501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.65840417005879e-06,
+      "loss": 0.9663,
+      "step": 1502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.6579482985959115e-06,
+      "loss": 0.9543,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.65749214153668e-06,
+      "loss": 1.0791,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.657035698930107e-06,
+      "loss": 1.134,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.656578970825238e-06,
+      "loss": 1.0719,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.656121957271147e-06,
+      "loss": 0.8976,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.655664658316942e-06,
+      "loss": 1.1163,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.655207074011758e-06,
+      "loss": 0.8294,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.654749204404765e-06,
+      "loss": 1.0142,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.654291049545159e-06,
+      "loss": 0.9537,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.653832609482169e-06,
+      "loss": 0.7829,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.653373884265054e-06,
+      "loss": 0.936,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.652914873943105e-06,
+      "loss": 0.7713,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.652455578565642e-06,
+      "loss": 0.9055,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.651995998182016e-06,
+      "loss": 0.9663,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.65153613284161e-06,
+      "loss": 0.9383,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.651075982593835e-06,
+      "loss": 0.9123,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.650615547488135e-06,
+      "loss": 0.8806,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.650154827573983e-06,
+      "loss": 0.9273,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.649693822900885e-06,
+      "loss": 0.7591,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.649232533518374e-06,
+      "loss": 0.9271,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6487709594760175e-06,
+      "loss": 0.9739,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.648309100823409e-06,
+      "loss": 0.9957,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6478469576101775e-06,
+      "loss": 0.9038,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.647384529885978e-06,
+      "loss": 0.8403,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.646921817700501e-06,
+      "loss": 0.9371,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.646458821103463e-06,
+      "loss": 0.9699,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.645995540144614e-06,
+      "loss": 0.8837,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.645531974873733e-06,
+      "loss": 0.8479,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.64506812534063e-06,
+      "loss": 0.8054,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6446039915951456e-06,
+      "loss": 1.014,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.644139573687151e-06,
+      "loss": 1.0224,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.643674871666547e-06,
+      "loss": 0.9086,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.643209885583268e-06,
+      "loss": 0.8984,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.642744615487275e-06,
+      "loss": 0.9008,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.642279061428563e-06,
+      "loss": 1.0966,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6418132234571514e-06,
+      "loss": 0.9674,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6413471016231e-06,
+      "loss": 0.9553,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.64088069597649e-06,
+      "loss": 0.9807,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.640414006567438e-06,
+      "loss": 0.8946,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.639947033446088e-06,
+      "loss": 0.8217,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.639479776662619e-06,
+      "loss": 1.1152,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.639012236267237e-06,
+      "loss": 1.1307,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.638544412310177e-06,
+      "loss": 1.0581,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.638076304841709e-06,
+      "loss": 0.8373,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.637607913912129e-06,
+      "loss": 0.8884,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.637139239571769e-06,
+      "loss": 0.9577,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.636670281870984e-06,
+      "loss": 0.9501,
+      "step": 1549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.636201040860168e-06,
+      "loss": 1.0621,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.635731516589735e-06,
+      "loss": 0.8094,
+      "step": 1551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6352617091101405e-06,
+      "loss": 1.3197,
+      "step": 1552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.634791618471863e-06,
+      "loss": 0.9209,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.634321244725414e-06,
+      "loss": 0.8965,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.633850587921336e-06,
+      "loss": 0.899,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6333796481102e-06,
+      "loss": 1.0753,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.632908425342608e-06,
+      "loss": 0.9815,
+      "step": 1557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.632436919669194e-06,
+      "loss": 0.7991,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.6319651311406205e-06,
+      "loss": 0.9746,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.631493059807582e-06,
+      "loss": 0.8728,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.631020705720802e-06,
+      "loss": 1.0696,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.630548068931034e-06,
+      "loss": 0.9315,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.630075149489064e-06,
+      "loss": 0.966,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6296019474457086e-06,
+      "loss": 1.0017,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6291284628518095e-06,
+      "loss": 1.1794,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.628654695758247e-06,
+      "loss": 0.8329,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.628180646215922e-06,
+      "loss": 0.995,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.627706314275776e-06,
+      "loss": 0.8779,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.627231699988774e-06,
+      "loss": 1.0955,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.626756803405913e-06,
+      "loss": 0.9331,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.62628162457822e-06,
+      "loss": 0.8337,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.625806163556755e-06,
+      "loss": 1.0829,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.625330420392604e-06,
+      "loss": 0.8867,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.624854395136888e-06,
+      "loss": 0.8789,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6243780878407544e-06,
+      "loss": 1.1032,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.623901498555382e-06,
+      "loss": 1.0999,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.623424627331981e-06,
+      "loss": 0.989,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6229474742217905e-06,
+      "loss": 1.158,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.622470039276082e-06,
+      "loss": 0.8066,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.621992322546155e-06,
+      "loss": 0.9877,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.621514324083339e-06,
+      "loss": 0.8868,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.621036043938996e-06,
+      "loss": 0.8427,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.620557482164517e-06,
+      "loss": 0.9638,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.620078638811325e-06,
+      "loss": 0.7697,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6195995139308695e-06,
+      "loss": 0.8459,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.619120107574632e-06,
+      "loss": 1.1176,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.618640419794127e-06,
+      "loss": 0.8746,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.618160450640896e-06,
+      "loss": 0.8292,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.61768020016651e-06,
+      "loss": 1.0855,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.617199668422575e-06,
+      "loss": 1.1057,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.616718855460721e-06,
+      "loss": 1.0926,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6162377613326135e-06,
+      "loss": 0.8591,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.615756386089946e-06,
+      "loss": 0.8661,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.61527472978444e-06,
+      "loss": 1.0596,
+      "step": 1594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.614792792467851e-06,
+      "loss": 1.0261,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.614310574191964e-06,
+      "loss": 1.0694,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.613828075008591e-06,
+      "loss": 0.7104,
+      "step": 1597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6133452949695794e-06,
+      "loss": 1.0741,
+      "step": 1598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.612862234126802e-06,
+      "loss": 0.8046,
+      "step": 1599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.612378892532163e-06,
+      "loss": 1.1926,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.611895270237599e-06,
+      "loss": 0.9305,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6114113672950745e-06,
+      "loss": 0.9565,
+      "step": 1602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.610927183756585e-06,
+      "loss": 1.038,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.610442719674156e-06,
+      "loss": 0.9743,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.609957975099843e-06,
+      "loss": 0.7688,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.609472950085732e-06,
+      "loss": 0.9502,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.608987644683938e-06,
+      "loss": 1.1945,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.6085020589466095e-06,
+      "loss": 0.8844,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.608016192925919e-06,
+      "loss": 1.1184,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.607530046674076e-06,
+      "loss": 1.1156,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.6070436202433146e-06,
+      "loss": 0.8679,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.606556913685902e-06,
+      "loss": 0.8645,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.606069927054136e-06,
+      "loss": 0.9469,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.605582660400342e-06,
+      "loss": 0.8942,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.605095113776877e-06,
+      "loss": 0.9275,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.604607287236128e-06,
+      "loss": 0.9162,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.604119180830511e-06,
+      "loss": 0.8723,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.603630794612475e-06,
+      "loss": 0.9831,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.6031421286344956e-06,
+      "loss": 0.8881,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.60265318294908e-06,
+      "loss": 0.883,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.602163957608766e-06,
+      "loss": 1.02,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.60167445266612e-06,
+      "loss": 0.9012,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.60118466817374e-06,
+      "loss": 0.9556,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.6006946041842525e-06,
+      "loss": 0.8123,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.600204260750316e-06,
+      "loss": 0.9581,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.599713637924617e-06,
+      "loss": 1.1791,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.599222735759873e-06,
+      "loss": 1.0577,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.598731554308833e-06,
+      "loss": 0.7785,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.598240093624271e-06,
+      "loss": 1.005,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.597748353758997e-06,
+      "loss": 0.8346,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.5972563347658465e-06,
+      "loss": 1.2594,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.59676403669769e-06,
+      "loss": 0.7873,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.596271459607422e-06,
+      "loss": 0.8596,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.595778603547972e-06,
+      "loss": 0.9309,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.595285468572295e-06,
+      "loss": 0.8184,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.59479205473338e-06,
+      "loss": 0.886,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.594298362084245e-06,
+      "loss": 0.8726,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.5938043906779344e-06,
+      "loss": 0.9745,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.593310140567528e-06,
+      "loss": 0.9615,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.592815611806132e-06,
+      "loss": 1.0554,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.592320804446884e-06,
+      "loss": 0.7705,
+      "step": 1641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.59182571854295e-06,
+      "loss": 0.8278,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.591330354147527e-06,
+      "loss": 1.066,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.590834711313843e-06,
+      "loss": 0.9443,
+      "step": 1644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.590338790095154e-06,
+      "loss": 0.9332,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.589842590544747e-06,
+      "loss": 0.9522,
+      "step": 1646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.589346112715938e-06,
+      "loss": 0.9132,
+      "step": 1647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.5888493566620745e-06,
+      "loss": 1.2172,
+      "step": 1648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.588352322436532e-06,
+      "loss": 0.864,
+      "step": 1649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.5878550100927175e-06,
+      "loss": 0.8093,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.587357419684067e-06,
+      "loss": 0.9831,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.586859551264046e-06,
+      "loss": 0.8647,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.58636140488615e-06,
+      "loss": 0.9444,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.5858629806039075e-06,
+      "loss": 0.8881,
+      "step": 1654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.585364278470871e-06,
+      "loss": 0.8805,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.584865298540627e-06,
+      "loss": 0.9106,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.584366040866792e-06,
+      "loss": 0.8013,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.58386650550301e-06,
+      "loss": 1.0176,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.583366692502956e-06,
+      "loss": 0.991,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.582866601920336e-06,
+      "loss": 1.0193,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.582366233808883e-06,
+      "loss": 1.0892,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.581865588222363e-06,
+      "loss": 0.8265,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.581364665214569e-06,
+      "loss": 1.102,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.580863464839327e-06,
+      "loss": 0.9102,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.580361987150489e-06,
+      "loss": 0.7991,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.579860232201941e-06,
+      "loss": 0.6613,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.579358200047593e-06,
+      "loss": 1.0032,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.578855890741393e-06,
+      "loss": 1.1645,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.57835330433731e-06,
+      "loss": 1.1878,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.577850440889348e-06,
+      "loss": 0.967,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.577347300451541e-06,
+      "loss": 0.9729,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.5768438830779505e-06,
+      "loss": 1.1712,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.576340188822669e-06,
+      "loss": 0.9076,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.575836217739817e-06,
+      "loss": 0.9007,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.575331969883548e-06,
+      "loss": 0.8823,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.574827445308042e-06,
+      "loss": 1.0988,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.574322644067511e-06,
+      "loss": 0.765,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.573817566216195e-06,
+      "loss": 1.0129,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.573312211808367e-06,
+      "loss": 0.9812,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.572806580898323e-06,
+      "loss": 0.846,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.572300673540397e-06,
+      "loss": 0.9062,
+      "step": 1681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.571794489788947e-06,
+      "loss": 0.9187,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.571288029698362e-06,
+      "loss": 0.9705,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.5707812933230625e-06,
+      "loss": 1.2218,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.570274280717497e-06,
+      "loss": 0.9861,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.569766991936142e-06,
+      "loss": 1.0415,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.5692594270335074e-06,
+      "loss": 0.841,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.568751586064132e-06,
+      "loss": 0.9141,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.56824346908258e-06,
+      "loss": 1.0927,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.567735076143452e-06,
+      "loss": 1.2106,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.567226407301373e-06,
+      "loss": 1.1638,
+      "step": 1691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.566717462610999e-06,
+      "loss": 0.8556,
+      "step": 1692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.566208242127018e-06,
+      "loss": 0.7376,
+      "step": 1693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.565698745904143e-06,
+      "loss": 0.867,
+      "step": 1694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.565188973997122e-06,
+      "loss": 1.131,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.564678926460729e-06,
+      "loss": 0.9388,
+      "step": 1696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.564168603349768e-06,
+      "loss": 1.0934,
+      "step": 1697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.563658004719073e-06,
+      "loss": 1.0889,
+      "step": 1698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.563147130623509e-06,
+      "loss": 1.3718,
+      "step": 1699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.562635981117968e-06,
+      "loss": 0.8334,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.562124556257374e-06,
+      "loss": 1.0896,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.56161285609668e-06,
+      "loss": 0.8314,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.561100880690866e-06,
+      "loss": 0.9614,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.560588630094946e-06,
+      "loss": 1.0356,
+      "step": 1704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.560076104363961e-06,
+      "loss": 1.044,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.559563303552979e-06,
+      "loss": 1.0273,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.559050227717105e-06,
+      "loss": 1.0569,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.558536876911466e-06,
+      "loss": 0.9424,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5580232511912225e-06,
+      "loss": 0.9129,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.557509350611563e-06,
+      "loss": 0.8904,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.556995175227706e-06,
+      "loss": 0.9675,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.556480725094899e-06,
+      "loss": 0.942,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5559660002684215e-06,
+      "loss": 0.8855,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.55545100080358e-06,
+      "loss": 1.0582,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.55493572675571e-06,
+      "loss": 0.9462,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.554420178180179e-06,
+      "loss": 0.9992,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.553904355132382e-06,
+      "loss": 0.9539,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.553388257667744e-06,
+      "loss": 0.8386,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5528718858417206e-06,
+      "loss": 0.7376,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.552355239709795e-06,
+      "loss": 1.2237,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5518383193274804e-06,
+      "loss": 0.9934,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.551321124750321e-06,
+      "loss": 0.9057,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.550803656033889e-06,
+      "loss": 1.1378,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.550285913233786e-06,
+      "loss": 1.3755,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5497678964056435e-06,
+      "loss": 0.9297,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5492496056051225e-06,
+      "loss": 0.9288,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.548731040887913e-06,
+      "loss": 0.9087,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.548212202309735e-06,
+      "loss": 0.9401,
+      "step": 1728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.547693089926339e-06,
+      "loss": 0.8736,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.547173703793501e-06,
+      "loss": 0.7821,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5466540439670304e-06,
+      "loss": 0.9998,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.546134110502765e-06,
+      "loss": 0.9619,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.545613903456571e-06,
+      "loss": 0.9549,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5450934228843445e-06,
+      "loss": 0.9704,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.544572668842011e-06,
+      "loss": 1.0626,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5440516413855264e-06,
+      "loss": 0.9908,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.543530340570874e-06,
+      "loss": 1.0535,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.543008766454069e-06,
+      "loss": 0.7915,
+      "step": 1738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.542486919091152e-06,
+      "loss": 1.1884,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.541964798538199e-06,
+      "loss": 1.0338,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.541442404851308e-06,
+      "loss": 0.948,
+      "step": 1741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.5409197380866125e-06,
+      "loss": 0.9903,
+      "step": 1742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.540396798300272e-06,
+      "loss": 1.0308,
+      "step": 1743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.539873585548478e-06,
+      "loss": 0.9442,
+      "step": 1744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.539350099887447e-06,
+      "loss": 0.9928,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.53882634137343e-06,
+      "loss": 0.9865,
+      "step": 1746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.538302310062703e-06,
+      "loss": 0.8636,
+      "step": 1747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.537778006011575e-06,
+      "loss": 1.0871,
+      "step": 1748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.537253429276381e-06,
+      "loss": 0.9433,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.536728579913486e-06,
+      "loss": 1.1315,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.536203457979288e-06,
+      "loss": 0.9099,
+      "step": 1751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.535678063530207e-06,
+      "loss": 1.0932,
+      "step": 1752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.535152396622701e-06,
+      "loss": 0.9387,
+      "step": 1753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.534626457313251e-06,
+      "loss": 1.2429,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.534100245658367e-06,
+      "loss": 0.8748,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.533573761714594e-06,
+      "loss": 0.869,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.5330470055385016e-06,
+      "loss": 0.9427,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.532519977186688e-06,
+      "loss": 0.9897,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.531992676715784e-06,
+      "loss": 0.7827,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.5314651041824474e-06,
+      "loss": 0.8794,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.530937259643366e-06,
+      "loss": 0.9551,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.530409143155257e-06,
+      "loss": 0.9972,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.529880754774865e-06,
+      "loss": 1.0226,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.529352094558969e-06,
+      "loss": 1.1152,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.528823162564369e-06,
+      "loss": 1.0828,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.528293958847901e-06,
+      "loss": 0.8319,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.527764483466427e-06,
+      "loss": 1.0631,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.527234736476841e-06,
+      "loss": 1.0912,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.526704717936061e-06,
+      "loss": 1.0184,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.526174427901041e-06,
+      "loss": 1.1469,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.525643866428758e-06,
+      "loss": 1.0254,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.525113033576222e-06,
+      "loss": 0.7984,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.524581929400471e-06,
+      "loss": 1.1154,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.524050553958571e-06,
+      "loss": 1.1021,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.52351890730762e-06,
+      "loss": 0.7883,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.522986989504742e-06,
+      "loss": 0.9802,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.522454800607093e-06,
+      "loss": 1.0664,
+      "step": 1777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.521922340671854e-06,
+      "loss": 1.0583,
+      "step": 1778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.521389609756241e-06,
+      "loss": 0.9338,
+      "step": 1779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.520856607917494e-06,
+      "loss": 1.2775,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.520323335212884e-06,
+      "loss": 1.198,
+      "step": 1781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.519789791699712e-06,
+      "loss": 0.9278,
+      "step": 1782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.519255977435305e-06,
+      "loss": 1.1096,
+      "step": 1783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.518721892477025e-06,
+      "loss": 0.9343,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.5181875368822564e-06,
+      "loss": 0.8726,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.517652910708417e-06,
+      "loss": 0.7923,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.517118014012952e-06,
+      "loss": 1.0582,
+      "step": 1787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.516582846853336e-06,
+      "loss": 0.8504,
+      "step": 1788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.516047409287073e-06,
+      "loss": 0.8887,
+      "step": 1789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.515511701371695e-06,
+      "loss": 0.8711,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.514975723164764e-06,
+      "loss": 0.7009,
+      "step": 1791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.514439474723872e-06,
+      "loss": 0.9755,
+      "step": 1792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.513902956106636e-06,
+      "loss": 0.9389,
+      "step": 1793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.513366167370708e-06,
+      "loss": 0.9317,
+      "step": 1794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.512829108573765e-06,
+      "loss": 0.797,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.512291779773513e-06,
+      "loss": 1.0107,
+      "step": 1796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.511754181027689e-06,
+      "loss": 1.2645,
+      "step": 1797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.511216312394056e-06,
+      "loss": 0.8949,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.510678173930411e-06,
+      "loss": 0.9163,
+      "step": 1799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.510139765694575e-06,
+      "loss": 0.8183,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.5096010877444005e-06,
+      "loss": 0.7972,
+      "step": 1801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.509062140137768e-06,
+      "loss": 0.9791,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.508522922932587e-06,
+      "loss": 0.9208,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.507983436186797e-06,
+      "loss": 1.0881,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.507443679958367e-06,
+      "loss": 0.8703,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.5069036543052915e-06,
+      "loss": 0.8585,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.5063633592855976e-06,
+      "loss": 0.8495,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.505822794957339e-06,
+      "loss": 1.1456,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.505281961378601e-06,
+      "loss": 0.9648,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.504740858607494e-06,
+      "loss": 1.4984,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.504199486702161e-06,
+      "loss": 0.9821,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.5036578457207715e-06,
+      "loss": 1.0686,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.503115935721526e-06,
+      "loss": 0.8111,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.50257375676265e-06,
+      "loss": 0.8554,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.502031308902404e-06,
+      "loss": 0.9558,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.501488592199072e-06,
+      "loss": 0.9142,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.500945606710969e-06,
+      "loss": 0.8669,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.500402352496439e-06,
+      "loss": 0.8755,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.499858829613856e-06,
+      "loss": 0.8915,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.499315038121618e-06,
+      "loss": 1.1692,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.498770978078159e-06,
+      "loss": 0.8042,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4982266495419365e-06,
+      "loss": 1.0297,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4976820525714385e-06,
+      "loss": 0.8719,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.497137187225182e-06,
+      "loss": 0.9056,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.496592053561713e-06,
+      "loss": 1.0025,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.496046651639607e-06,
+      "loss": 1.1314,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.495500981517465e-06,
+      "loss": 0.9233,
+      "step": 1827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4949550432539206e-06,
+      "loss": 0.9924,
+      "step": 1828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.494408836907636e-06,
+      "loss": 0.8188,
+      "step": 1829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4938623625373e-06,
+      "loss": 0.9573,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.493315620201629e-06,
+      "loss": 1.1856,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.492768609959374e-06,
+      "loss": 1.0134,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.49222133186931e-06,
+      "loss": 0.9265,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.491673785990241e-06,
+      "loss": 0.8263,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.491125972381002e-06,
+      "loss": 1.0469,
+      "step": 1835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4905778911004555e-06,
+      "loss": 0.831,
+      "step": 1836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4900295422074915e-06,
+      "loss": 1.004,
+      "step": 1837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.489480925761031e-06,
+      "loss": 1.141,
+      "step": 1838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4889320418200225e-06,
+      "loss": 0.9205,
+      "step": 1839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.488382890443444e-06,
+      "loss": 0.7996,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.4878334716903015e-06,
+      "loss": 1.0184,
+      "step": 1841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.48728378561963e-06,
+      "loss": 1.2054,
+      "step": 1842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.486733832290494e-06,
+      "loss": 0.9128,
+      "step": 1843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.486183611761983e-06,
+      "loss": 0.8991,
+      "step": 1844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.485633124093223e-06,
+      "loss": 0.7582,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.485082369343359e-06,
+      "loss": 1.0399,
+      "step": 1846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.484531347571573e-06,
+      "loss": 0.7783,
+      "step": 1847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.483980058837071e-06,
+      "loss": 0.8195,
+      "step": 1848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.483428503199087e-06,
+      "loss": 0.7576,
+      "step": 1849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.48287668071689e-06,
+      "loss": 0.8249,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.482324591449769e-06,
+      "loss": 0.7686,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.48177223545705e-06,
+      "loss": 1.0496,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.481219612798079e-06,
+      "loss": 0.8215,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4806667235322395e-06,
+      "loss": 1.1949,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.480113567718937e-06,
+      "loss": 1.1126,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.479560145417608e-06,
+      "loss": 0.7972,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.47900645668772e-06,
+      "loss": 1.0195,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.478452501588763e-06,
+      "loss": 0.8229,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.477898280180262e-06,
+      "loss": 0.9148,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.477343792521769e-06,
+      "loss": 0.9664,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.47678903867286e-06,
+      "loss": 0.9086,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.476234018693148e-06,
+      "loss": 1.0583,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4756787326422656e-06,
+      "loss": 1.1682,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4751231805798806e-06,
+      "loss": 1.0053,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.474567362565687e-06,
+      "loss": 1.0072,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.474011278659406e-06,
+      "loss": 1.1222,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.47345492892079e-06,
+      "loss": 0.8537,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.472898313409618e-06,
+      "loss": 0.823,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4723414321857e-06,
+      "loss": 0.9539,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.471784285308871e-06,
+      "loss": 0.9165,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.471226872838997e-06,
+      "loss": 0.9209,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4706691948359705e-06,
+      "loss": 0.8952,
+      "step": 1872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.470111251359716e-06,
+      "loss": 0.8653,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.469553042470184e-06,
+      "loss": 0.9445,
+      "step": 1874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.468994568227354e-06,
+      "loss": 0.9305,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.468435828691234e-06,
+      "loss": 1.0816,
+      "step": 1876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.46787682392186e-06,
+      "loss": 1.0109,
+      "step": 1877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.467317553979298e-06,
+      "loss": 0.9169,
+      "step": 1878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.466758018923639e-06,
+      "loss": 1.0254,
+      "step": 1879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4661982188150096e-06,
+      "loss": 1.1398,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.465638153713555e-06,
+      "loss": 0.9408,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.465077823679459e-06,
+      "loss": 0.9106,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.464517228772925e-06,
+      "loss": 0.9653,
+      "step": 1883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4639563690541925e-06,
+      "loss": 0.8979,
+      "step": 1884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.463395244583523e-06,
+      "loss": 1.1831,
+      "step": 1885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.462833855421209e-06,
+      "loss": 1.0886,
+      "step": 1886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.462272201627575e-06,
+      "loss": 0.793,
+      "step": 1887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.461710283262966e-06,
+      "loss": 1.0232,
+      "step": 1888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.4611481003877654e-06,
+      "loss": 0.8853,
+      "step": 1889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.460585653062377e-06,
+      "loss": 0.9787,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.460022941347234e-06,
+      "loss": 1.1529,
+      "step": 1891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.459459965302802e-06,
+      "loss": 0.8324,
+      "step": 1892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.458896724989572e-06,
+      "loss": 0.7484,
+      "step": 1893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.458333220468066e-06,
+      "loss": 0.8327,
+      "step": 1894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.457769451798829e-06,
+      "loss": 0.921,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.457205419042441e-06,
+      "loss": 0.8538,
+      "step": 1896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.456641122259506e-06,
+      "loss": 1.1132,
+      "step": 1897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.456076561510657e-06,
+      "loss": 1.0289,
+      "step": 1898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.455511736856557e-06,
+      "loss": 0.9874,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.454946648357896e-06,
+      "loss": 0.9705,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.4543812960753924e-06,
+      "loss": 1.1799,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.453815680069795e-06,
+      "loss": 0.9194,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.453249800401877e-06,
+      "loss": 1.214,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.4526836571324436e-06,
+      "loss": 1.0368,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.452117250322325e-06,
+      "loss": 0.9438,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.451550580032384e-06,
+      "loss": 1.0484,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.4509836463235075e-06,
+      "loss": 1.011,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.450416449256614e-06,
+      "loss": 0.8974,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.449848988892647e-06,
+      "loss": 0.8694,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.449281265292581e-06,
+      "loss": 0.8961,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.448713278517418e-06,
+      "loss": 1.1372,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.4481450286281874e-06,
+      "loss": 0.8131,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.447576515685949e-06,
+      "loss": 0.9954,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.447007739751788e-06,
+      "loss": 1.0506,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.446438700886821e-06,
+      "loss": 0.8486,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.445869399152189e-06,
+      "loss": 1.0307,
+      "step": 1916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.445299834609066e-06,
+      "loss": 1.0452,
+      "step": 1917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.444730007318649e-06,
+      "loss": 0.907,
+      "step": 1918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.444159917342167e-06,
+      "loss": 1.0056,
+      "step": 1919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.443589564740877e-06,
+      "loss": 0.8903,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.443018949576063e-06,
+      "loss": 1.2022,
+      "step": 1921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.442448071909037e-06,
+      "loss": 1.0649,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.441876931801139e-06,
+      "loss": 1.0977,
+      "step": 1923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.44130552931374e-06,
+      "loss": 0.9511,
+      "step": 1924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.440733864508235e-06,
+      "loss": 0.7768,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.440161937446052e-06,
+      "loss": 1.1062,
+      "step": 1926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.439589748188641e-06,
+      "loss": 0.8558,
+      "step": 1927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.439017296797487e-06,
+      "loss": 0.8351,
+      "step": 1928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.4384445833340985e-06,
+      "loss": 1.0016,
+      "step": 1929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.437871607860013e-06,
+      "loss": 1.0766,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.437298370436797e-06,
+      "loss": 0.8905,
+      "step": 1931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.436724871126045e-06,
+      "loss": 0.8563,
+      "step": 1932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.43615110998938e-06,
+      "loss": 1.119,
+      "step": 1933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.435577087088453e-06,
+      "loss": 0.9224,
+      "step": 1934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.43500280248494e-06,
+      "loss": 0.9588,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.434428256240551e-06,
+      "loss": 0.8386,
+      "step": 1936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.433853448417019e-06,
+      "loss": 0.9214,
+      "step": 1937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.4332783790761065e-06,
+      "loss": 0.84,
+      "step": 1938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.432703048279607e-06,
+      "loss": 1.0976,
+      "step": 1939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.432127456089339e-06,
+      "loss": 0.8978,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.431551602567149e-06,
+      "loss": 0.848,
+      "step": 1941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.430975487774912e-06,
+      "loss": 0.8924,
+      "step": 1942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.430399111774534e-06,
+      "loss": 1.0204,
+      "step": 1943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.429822474627943e-06,
+      "loss": 0.8708,
+      "step": 1944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.429245576397102e-06,
+      "loss": 1.068,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.428668417143996e-06,
+      "loss": 0.9335,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.428090996930642e-06,
+      "loss": 0.7742,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4275133158190845e-06,
+      "loss": 1.1121,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4269353738713944e-06,
+      "loss": 0.8645,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.426357171149671e-06,
+      "loss": 0.9939,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4257787077160424e-06,
+      "loss": 0.9021,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.425199983632666e-06,
+      "loss": 0.9096,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.424620998961723e-06,
+      "loss": 0.9227,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4240417537654275e-06,
+      "loss": 0.8134,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.423462248106018e-06,
+      "loss": 0.913,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4228824820457635e-06,
+      "loss": 0.9067,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4223024556469585e-06,
+      "loss": 1.062,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.421722168971928e-06,
+      "loss": 0.9732,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.421141622083024e-06,
+      "loss": 0.7693,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.420560815042624e-06,
+      "loss": 1.1427,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.419979747913138e-06,
+      "loss": 1.091,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.419398420757e-06,
+      "loss": 0.9646,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.418816833636676e-06,
+      "loss": 0.8042,
+      "step": 1963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.418234986614653e-06,
+      "loss": 0.9498,
+      "step": 1964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.417652879753456e-06,
+      "loss": 0.9812,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.417070513115628e-06,
+      "loss": 0.8327,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.416487886763747e-06,
+      "loss": 1.1095,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.415905000760413e-06,
+      "loss": 1.1187,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.415321855168262e-06,
+      "loss": 1.1365,
+      "step": 1969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.414738450049948e-06,
+      "loss": 0.7954,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.41415478546816e-06,
+      "loss": 1.1697,
+      "step": 1971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.413570861485613e-06,
+      "loss": 1.0876,
+      "step": 1972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.412986678165049e-06,
+      "loss": 0.9512,
+      "step": 1973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.412402235569239e-06,
+      "loss": 1.0945,
+      "step": 1974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.41181753376098e-06,
+      "loss": 0.9918,
+      "step": 1975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4112325728031e-06,
+      "loss": 0.9063,
+      "step": 1976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.410647352758452e-06,
+      "loss": 0.8894,
+      "step": 1977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.410061873689918e-06,
+      "loss": 0.9111,
+      "step": 1978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.409476135660407e-06,
+      "loss": 0.9663,
+      "step": 1979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.408890138732858e-06,
+      "loss": 0.8124,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4083038829702354e-06,
+      "loss": 1.0365,
+      "step": 1981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.407717368435532e-06,
+      "loss": 0.7885,
+      "step": 1982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.407130595191769e-06,
+      "loss": 1.0983,
+      "step": 1983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.406543563301996e-06,
+      "loss": 0.9095,
+      "step": 1984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.405956272829288e-06,
+      "loss": 0.9744,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.40536872383675e-06,
+      "loss": 0.9156,
+      "step": 1986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.404780916387515e-06,
+      "loss": 0.8263,
+      "step": 1987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.404192850544742e-06,
+      "loss": 0.8564,
+      "step": 1988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4036045263716195e-06,
+      "loss": 0.7952,
+      "step": 1989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.403015943931361e-06,
+      "loss": 1.0267,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.402427103287211e-06,
+      "loss": 0.8345,
+      "step": 1991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.40183800450244e-06,
+      "loss": 1.1449,
+      "step": 1992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4012486476403485e-06,
+      "loss": 0.8897,
+      "step": 1993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.4006590327642595e-06,
+      "loss": 0.9408,
+      "step": 1994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.400069159937529e-06,
+      "loss": 1.1026,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3994790292235385e-06,
+      "loss": 0.8509,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.398888640685698e-06,
+      "loss": 0.9691,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.398297994387444e-06,
+      "loss": 1.0327,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.397707090392242e-06,
+      "loss": 1.0577,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.397115928763583e-06,
+      "loss": 0.8344,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0286,
+      "eval_samples_per_second": 352.327,
+      "eval_steps_per_second": 44.049,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.396524509564989e-06,
+      "loss": 0.9744,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.395932832860008e-06,
+      "loss": 1.0089,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.395340898712215e-06,
+      "loss": 0.8524,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.394748707185213e-06,
+      "loss": 0.7612,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.394156258342634e-06,
+      "loss": 0.9477,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.393563552248136e-06,
+      "loss": 0.9264,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.392970588965404e-06,
+      "loss": 1.0177,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3923773685581535e-06,
+      "loss": 1.2797,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.391783891090125e-06,
+      "loss": 1.0346,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.391190156625089e-06,
+      "loss": 0.8052,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3905961652268406e-06,
+      "loss": 0.8956,
+      "step": 2011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.390001916959205e-06,
+      "loss": 1.025,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.389407411886034e-06,
+      "loss": 1.0725,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3888126500712074e-06,
+      "loss": 0.9196,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3882176315786315e-06,
+      "loss": 0.9083,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.387622356472242e-06,
+      "loss": 0.9892,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.387026824816e-06,
+      "loss": 0.8558,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.386431036673897e-06,
+      "loss": 0.8412,
+      "step": 2018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.385834992109948e-06,
+      "loss": 0.8734,
+      "step": 2019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.385238691188201e-06,
+      "loss": 0.9499,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.384642133972725e-06,
+      "loss": 1.2012,
+      "step": 2021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.384045320527623e-06,
+      "loss": 0.7914,
+      "step": 2022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.383448250917022e-06,
+      "loss": 0.9789,
+      "step": 2023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3828509252050755e-06,
+      "loss": 0.9152,
+      "step": 2024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.382253343455967e-06,
+      "loss": 0.8777,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.381655505733908e-06,
+      "loss": 1.0393,
+      "step": 2026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.381057412103134e-06,
+      "loss": 0.9291,
+      "step": 2027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.380459062627911e-06,
+      "loss": 1.026,
+      "step": 2028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.379860457372531e-06,
+      "loss": 1.091,
+      "step": 2029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.379261596401316e-06,
+      "loss": 0.862,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.378662479778611e-06,
+      "loss": 1.2462,
+      "step": 2031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.378063107568793e-06,
+      "loss": 0.8137,
+      "step": 2032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3774634798362646e-06,
+      "loss": 1.0157,
+      "step": 2033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.376863596645453e-06,
+      "loss": 1.0534,
+      "step": 2034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.376263458060819e-06,
+      "loss": 0.9251,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.375663064146846e-06,
+      "loss": 1.0599,
+      "step": 2036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.375062414968045e-06,
+      "loss": 1.1351,
+      "step": 2037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.374461510588958e-06,
+      "loss": 0.9465,
+      "step": 2038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.37386035107415e-06,
+      "loss": 1.0234,
+      "step": 2039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.373258936488218e-06,
+      "loss": 0.9426,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.372657266895783e-06,
+      "loss": 0.9145,
+      "step": 2041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.372055342361493e-06,
+      "loss": 1.0325,
+      "step": 2042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.371453162950026e-06,
+      "loss": 0.9737,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.370850728726086e-06,
+      "loss": 0.851,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.370248039754404e-06,
+      "loss": 0.8591,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3696450960997385e-06,
+      "loss": 1.0886,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.369041897826877e-06,
+      "loss": 0.9065,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.368438445000634e-06,
+      "loss": 0.9592,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.367834737685848e-06,
+      "loss": 1.0103,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.367230775947388e-06,
+      "loss": 1.0671,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.366626559850152e-06,
+      "loss": 1.0871,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.366022089459059e-06,
+      "loss": 0.9994,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.365417364839062e-06,
+      "loss": 0.8809,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.364812386055138e-06,
+      "loss": 1.1061,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.364207153172291e-06,
+      "loss": 1.0178,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3636016662555554e-06,
+      "loss": 0.9353,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.362995925369989e-06,
+      "loss": 0.926,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3623899305806786e-06,
+      "loss": 1.0147,
+      "step": 2058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.36178368195274e-06,
+      "loss": 0.9483,
+      "step": 2059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.361177179551312e-06,
+      "loss": 1.1992,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.360570423441564e-06,
+      "loss": 0.8016,
+      "step": 2061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.359963413688694e-06,
+      "loss": 0.9437,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3593561503579225e-06,
+      "loss": 0.9806,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.358748633514502e-06,
+      "loss": 1.1723,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.358140863223708e-06,
+      "loss": 0.8871,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.357532839550848e-06,
+      "loss": 0.8116,
+      "step": 2066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.356924562561251e-06,
+      "loss": 0.7259,
+      "step": 2067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.356316032320279e-06,
+      "loss": 0.9829,
+      "step": 2068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.355707248893318e-06,
+      "loss": 0.9762,
+      "step": 2069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.355098212345779e-06,
+      "loss": 0.8791,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.354488922743107e-06,
+      "loss": 0.9837,
+      "step": 2071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.353879380150768e-06,
+      "loss": 0.8277,
+      "step": 2072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.353269584634257e-06,
+      "loss": 0.8697,
+      "step": 2073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.352659536259098e-06,
+      "loss": 1.1657,
+      "step": 2074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.352049235090839e-06,
+      "loss": 1.0837,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3514386811950574e-06,
+      "loss": 0.9615,
+      "step": 2076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.350827874637358e-06,
+      "loss": 1.2017,
+      "step": 2077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.350216815483371e-06,
+      "loss": 0.8883,
+      "step": 2078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3496055037987535e-06,
+      "loss": 0.9686,
+      "step": 2079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.348993939649194e-06,
+      "loss": 1.0088,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.348382123100403e-06,
+      "loss": 1.2405,
+      "step": 2081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.34777005421812e-06,
+      "loss": 0.8991,
+      "step": 2082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.347157733068113e-06,
+      "loss": 1.2714,
+      "step": 2083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.346545159716174e-06,
+      "loss": 0.9427,
+      "step": 2084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.345932334228125e-06,
+      "loss": 0.9592,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.345319256669814e-06,
+      "loss": 0.8065,
+      "step": 2086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.344705927107118e-06,
+      "loss": 1.0332,
+      "step": 2087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.344092345605936e-06,
+      "loss": 0.9296,
+      "step": 2088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3434785122321986e-06,
+      "loss": 0.8416,
+      "step": 2089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.342864427051864e-06,
+      "loss": 0.7472,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.342250090130912e-06,
+      "loss": 0.7449,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.341635501535355e-06,
+      "loss": 0.8141,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.341020661331231e-06,
+      "loss": 0.9602,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.340405569584603e-06,
+      "loss": 1.0518,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.339790226361565e-06,
+      "loss": 0.9843,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3391746317282345e-06,
+      "loss": 0.9381,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.338558785750756e-06,
+      "loss": 1.0291,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3379426884953025e-06,
+      "loss": 0.9797,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.337326340028074e-06,
+      "loss": 1.057,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.336709740415298e-06,
+      "loss": 0.8881,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.336092889723226e-06,
+      "loss": 1.0142,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.33547578801814e-06,
+      "loss": 0.6481,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.334858435366347e-06,
+      "loss": 0.9247,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.334240831834182e-06,
+      "loss": 1.1453,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.333622977488007e-06,
+      "loss": 1.0225,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.333004872394209e-06,
+      "loss": 1.1377,
+      "step": 2106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.332386516619203e-06,
+      "loss": 0.8496,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.331767910229433e-06,
+      "loss": 1.0057,
+      "step": 2108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.331149053291367e-06,
+      "loss": 0.9775,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.330529945871503e-06,
+      "loss": 1.0833,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.329910588036361e-06,
+      "loss": 1.1037,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.329290979852494e-06,
+      "loss": 0.9516,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.328671121386477e-06,
+      "loss": 0.8411,
+      "step": 2113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.328051012704913e-06,
+      "loss": 0.876,
+      "step": 2114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.327430653874435e-06,
+      "loss": 1.0085,
+      "step": 2115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.326810044961699e-06,
+      "loss": 0.9423,
+      "step": 2116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3261891860333895e-06,
+      "loss": 0.9133,
+      "step": 2117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.325568077156218e-06,
+      "loss": 0.7395,
+      "step": 2118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.324946718396922e-06,
+      "loss": 0.8559,
+      "step": 2119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.324325109822269e-06,
+      "loss": 0.8962,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.323703251499047e-06,
+      "loss": 0.9687,
+      "step": 2121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.323081143494077e-06,
+      "loss": 0.9476,
+      "step": 2122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.322458785874203e-06,
+      "loss": 0.9525,
+      "step": 2123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3218361787062994e-06,
+      "loss": 1.0954,
+      "step": 2124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.321213322057264e-06,
+      "loss": 0.8106,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.320590215994022e-06,
+      "loss": 0.7288,
+      "step": 2126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.319966860583528e-06,
+      "loss": 1.0748,
+      "step": 2127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.319343255892759e-06,
+      "loss": 1.0312,
+      "step": 2128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.318719401988724e-06,
+      "loss": 0.9833,
+      "step": 2129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.318095298938454e-06,
+      "loss": 0.9763,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.317470946809009e-06,
+      "loss": 0.9666,
+      "step": 2131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3168463456674776e-06,
+      "loss": 0.8606,
+      "step": 2132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.31622149558097e-06,
+      "loss": 1.0479,
+      "step": 2133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3155963966166294e-06,
+      "loss": 0.9231,
+      "step": 2134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.314971048841621e-06,
+      "loss": 0.8621,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.314345452323137e-06,
+      "loss": 0.8674,
+      "step": 2136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.313719607128401e-06,
+      "loss": 1.0201,
+      "step": 2137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.313093513324658e-06,
+      "loss": 1.007,
+      "step": 2138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.3124671709791825e-06,
+      "loss": 0.9265,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.311840580159274e-06,
+      "loss": 1.0341,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.31121374093226e-06,
+      "loss": 1.0586,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.310586653365495e-06,
+      "loss": 1.0631,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.309959317526359e-06,
+      "loss": 0.9898,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.309331733482257e-06,
+      "loss": 1.161,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.308703901300627e-06,
+      "loss": 0.817,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.3080758210489275e-06,
+      "loss": 0.8678,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.307447492794646e-06,
+      "loss": 1.0541,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.306818916605295e-06,
+      "loss": 0.9602,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.306190092548417e-06,
+      "loss": 0.9767,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.305561020691578e-06,
+      "loss": 0.8718,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.304931701102371e-06,
+      "loss": 0.9833,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.304302133848418e-06,
+      "loss": 0.8651,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.303672318997365e-06,
+      "loss": 0.9819,
+      "step": 2153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.3030422566168855e-06,
+      "loss": 0.9429,
+      "step": 2154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.3024119467746796e-06,
+      "loss": 1.2254,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.301781389538475e-06,
+      "loss": 0.852,
+      "step": 2156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.301150584976024e-06,
+      "loss": 1.1446,
+      "step": 2157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.300519533155107e-06,
+      "loss": 0.9005,
+      "step": 2158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.29988823414353e-06,
+      "loss": 0.8197,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.299256688009127e-06,
+      "loss": 0.9281,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.298624894819755e-06,
+      "loss": 0.952,
+      "step": 2161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.297992854643304e-06,
+      "loss": 1.0952,
+      "step": 2162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.297360567547685e-06,
+      "loss": 1.0235,
+      "step": 2163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.296728033600836e-06,
+      "loss": 0.9051,
+      "step": 2164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.296095252870724e-06,
+      "loss": 1.0226,
+      "step": 2165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.295462225425341e-06,
+      "loss": 1.1001,
+      "step": 2166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2948289513327054e-06,
+      "loss": 1.2066,
+      "step": 2167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2941954306608625e-06,
+      "loss": 1.0407,
+      "step": 2168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.293561663477885e-06,
+      "loss": 0.9159,
+      "step": 2169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.29292764985187e-06,
+      "loss": 1.195,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.292293389850943e-06,
+      "loss": 0.9187,
+      "step": 2171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2916588835432544e-06,
+      "loss": 1.0759,
+      "step": 2172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.291024130996983e-06,
+      "loss": 1.1,
+      "step": 2173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.290389132280332e-06,
+      "loss": 1.042,
+      "step": 2174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.289753887461531e-06,
+      "loss": 1.0089,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.28911839660884e-06,
+      "loss": 1.1133,
+      "step": 2176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2884826597905385e-06,
+      "loss": 1.0183,
+      "step": 2177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.28784667707494e-06,
+      "loss": 0.9107,
+      "step": 2178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.287210448530379e-06,
+      "loss": 1.2817,
+      "step": 2179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.286573974225218e-06,
+      "loss": 0.8715,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.285937254227847e-06,
+      "loss": 1.1864,
+      "step": 2181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2853002886066825e-06,
+      "loss": 0.8947,
+      "step": 2182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.284663077430163e-06,
+      "loss": 1.0336,
+      "step": 2183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2840256207667595e-06,
+      "loss": 0.9945,
+      "step": 2184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2833879186849664e-06,
+      "loss": 0.8628,
+      "step": 2185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.282749971253305e-06,
+      "loss": 0.9275,
+      "step": 2186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.282111778540321e-06,
+      "loss": 0.8971,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.28147334061459e-06,
+      "loss": 0.8883,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.28083465754471e-06,
+      "loss": 0.8309,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.280195729399311e-06,
+      "loss": 0.8338,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.279556556247042e-06,
+      "loss": 1.0008,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.278917138156584e-06,
+      "loss": 1.1495,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.278277475196642e-06,
+      "loss": 0.8336,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.277637567435947e-06,
+      "loss": 1.0254,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.276997414943259e-06,
+      "loss": 0.9133,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.276357017787361e-06,
+      "loss": 1.0833,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.275716376037065e-06,
+      "loss": 0.909,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.275075489761205e-06,
+      "loss": 0.865,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.274434359028647e-06,
+      "loss": 0.8087,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.273792983908279e-06,
+      "loss": 0.9862,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.273151364469017e-06,
+      "loss": 1.0059,
+      "step": 2201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.272509500779804e-06,
+      "loss": 0.9221,
+      "step": 2202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.271867392909606e-06,
+      "loss": 0.9621,
+      "step": 2203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.27122504092742e-06,
+      "loss": 0.9615,
+      "step": 2204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.270582444902265e-06,
+      "loss": 0.8818,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.269939604903188e-06,
+      "loss": 1.0407,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.269296520999263e-06,
+      "loss": 0.894,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2686531932595885e-06,
+      "loss": 0.9077,
+      "step": 2208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2680096217532896e-06,
+      "loss": 0.9945,
+      "step": 2209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.267365806549519e-06,
+      "loss": 0.8381,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.266721747717455e-06,
+      "loss": 0.914,
+      "step": 2211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.266077445326301e-06,
+      "loss": 1.0046,
+      "step": 2212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2654328994452865e-06,
+      "loss": 0.8895,
+      "step": 2213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.264788110143668e-06,
+      "loss": 1.1502,
+      "step": 2214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.26414307749073e-06,
+      "loss": 0.9952,
+      "step": 2215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2634978015557795e-06,
+      "loss": 0.9824,
+      "step": 2216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.262852282408152e-06,
+      "loss": 1.0216,
+      "step": 2217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2622065201172075e-06,
+      "loss": 0.9112,
+      "step": 2218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.261560514752334e-06,
+      "loss": 0.9438,
+      "step": 2219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.260914266382945e-06,
+      "loss": 0.9328,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.260267775078479e-06,
+      "loss": 0.8888,
+      "step": 2221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.259621040908402e-06,
+      "loss": 0.7303,
+      "step": 2222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.258974063942206e-06,
+      "loss": 0.877,
+      "step": 2223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.258326844249407e-06,
+      "loss": 0.9019,
+      "step": 2224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.25767938189955e-06,
+      "loss": 0.9131,
+      "step": 2225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.257031676962205e-06,
+      "loss": 0.8729,
+      "step": 2226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.256383729506967e-06,
+      "loss": 1.0595,
+      "step": 2227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.255735539603458e-06,
+      "loss": 0.9507,
+      "step": 2228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.255087107321327e-06,
+      "loss": 0.9108,
+      "step": 2229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.254438432730246e-06,
+      "loss": 0.7734,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.253789515899918e-06,
+      "loss": 0.9291,
+      "step": 2231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2531403569000665e-06,
+      "loss": 1.0132,
+      "step": 2232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.252490955800445e-06,
+      "loss": 0.9676,
+      "step": 2233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.25184131267083e-06,
+      "loss": 0.9758,
+      "step": 2234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.251191427581027e-06,
+      "loss": 1.0558,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.250541300600865e-06,
+      "loss": 0.749,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.249890931800202e-06,
+      "loss": 1.1418,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.249240321248919e-06,
+      "loss": 0.9666,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.248589469016923e-06,
+      "loss": 1.1441,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.24793837517415e-06,
+      "loss": 0.8825,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2472870397905585e-06,
+      "loss": 1.1467,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.246635462936135e-06,
+      "loss": 0.8301,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.245983644680891e-06,
+      "loss": 0.763,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.245331585094865e-06,
+      "loss": 1.0763,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.244679284248121e-06,
+      "loss": 0.9151,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.244026742210747e-06,
+      "loss": 0.8194,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2433739590528605e-06,
+      "loss": 0.9766,
+      "step": 2247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.242720934844602e-06,
+      "loss": 0.996,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2420676696561385e-06,
+      "loss": 0.8958,
+      "step": 2249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.241414163557664e-06,
+      "loss": 0.954,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.240760416619398e-06,
+      "loss": 0.8122,
+      "step": 2251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.240106428911585e-06,
+      "loss": 0.912,
+      "step": 2252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.239452200504496e-06,
+      "loss": 1.1593,
+      "step": 2253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.238797731468428e-06,
+      "loss": 1.0324,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.238143021873703e-06,
+      "loss": 1.1349,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.237488071790671e-06,
+      "loss": 0.9285,
+      "step": 2256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2368328812897045e-06,
+      "loss": 0.8325,
+      "step": 2257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.236177450441205e-06,
+      "loss": 1.004,
+      "step": 2258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.235521779315597e-06,
+      "loss": 0.9767,
+      "step": 2259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2348658679833345e-06,
+      "loss": 0.9082,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.234209716514894e-06,
+      "loss": 1.0185,
+      "step": 2261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2335533249807784e-06,
+      "loss": 1.0599,
+      "step": 2262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.232896693451517e-06,
+      "loss": 0.8633,
+      "step": 2263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.232239821997666e-06,
+      "loss": 0.8465,
+      "step": 2264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.231582710689805e-06,
+      "loss": 0.779,
+      "step": 2265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.230925359598541e-06,
+      "loss": 1.0325,
+      "step": 2266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.230267768794507e-06,
+      "loss": 1.0861,
+      "step": 2267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.22960993834836e-06,
+      "loss": 0.9957,
+      "step": 2268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.228951868330784e-06,
+      "loss": 0.9446,
+      "step": 2269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.228293558812489e-06,
+      "loss": 0.9172,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.22763500986421e-06,
+      "loss": 1.1187,
+      "step": 2271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.226976221556708e-06,
+      "loss": 0.8433,
+      "step": 2272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.22631719396077e-06,
+      "loss": 0.7467,
+      "step": 2273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2256579271472086e-06,
+      "loss": 0.921,
+      "step": 2274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2249984211868614e-06,
+      "loss": 0.7918,
+      "step": 2275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2243386761505925e-06,
+      "loss": 1.2375,
+      "step": 2276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.223678692109292e-06,
+      "loss": 0.9202,
+      "step": 2277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.223018469133875e-06,
+      "loss": 0.9436,
+      "step": 2278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.222358007295282e-06,
+      "loss": 0.7107,
+      "step": 2279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.221697306664478e-06,
+      "loss": 1.075,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.221036367312459e-06,
+      "loss": 0.8818,
+      "step": 2281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.220375189310241e-06,
+      "loss": 0.9534,
+      "step": 2282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.219713772728866e-06,
+      "loss": 1.0613,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.219052117639405e-06,
+      "loss": 0.8267,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.218390224112952e-06,
+      "loss": 0.9093,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.217728092220628e-06,
+      "loss": 0.9458,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.217065722033579e-06,
+      "loss": 0.9036,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.216403113622976e-06,
+      "loss": 1.1513,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.215740267060017e-06,
+      "loss": 0.8926,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2150771824159235e-06,
+      "loss": 1.1308,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.214413859761946e-06,
+      "loss": 1.1498,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.213750299169357e-06,
+      "loss": 0.8312,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.213086500709455e-06,
+      "loss": 1.0423,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.212422464453568e-06,
+      "loss": 1.2037,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.211758190473045e-06,
+      "loss": 1.1828,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.211093678839262e-06,
+      "loss": 0.9032,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.210428929623622e-06,
+      "loss": 0.9327,
+      "step": 2297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.209763942897549e-06,
+      "loss": 0.9639,
+      "step": 2298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2090987187324985e-06,
+      "loss": 0.9087,
+      "step": 2299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2084332571999496e-06,
+      "loss": 1.2201,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.207767558371404e-06,
+      "loss": 0.9069,
+      "step": 2301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.207101622318392e-06,
+      "loss": 1.0056,
+      "step": 2302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.206435449112467e-06,
+      "loss": 0.9922,
+      "step": 2303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2057690388252124e-06,
+      "loss": 0.7829,
+      "step": 2304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.205102391528231e-06,
+      "loss": 0.8016,
+      "step": 2305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2044355072931545e-06,
+      "loss": 1.1343,
+      "step": 2306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2037683861916416e-06,
+      "loss": 1.0169,
+      "step": 2307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2031010282953715e-06,
+      "loss": 0.9702,
+      "step": 2308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.202433433676053e-06,
+      "loss": 0.9735,
+      "step": 2309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.201765602405421e-06,
+      "loss": 1.0213,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.20109753455523e-06,
+      "loss": 1.0009,
+      "step": 2311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.200429230197267e-06,
+      "loss": 1.0114,
+      "step": 2312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.19976068940334e-06,
+      "loss": 1.1238,
+      "step": 2313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.199091912245285e-06,
+      "loss": 1.1496,
+      "step": 2314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.198422898794959e-06,
+      "loss": 0.9576,
+      "step": 2315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.19775364912425e-06,
+      "loss": 0.9875,
+      "step": 2316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.197084163305068e-06,
+      "loss": 1.0115,
+      "step": 2317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.196414441409351e-06,
+      "loss": 0.924,
+      "step": 2318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.195744483509058e-06,
+      "loss": 0.854,
+      "step": 2319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.195074289676177e-06,
+      "loss": 0.9339,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1944038599827195e-06,
+      "loss": 0.9085,
+      "step": 2321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.193733194500725e-06,
+      "loss": 0.8463,
+      "step": 2322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.193062293302255e-06,
+      "loss": 1.0058,
+      "step": 2323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.192391156459397e-06,
+      "loss": 1.144,
+      "step": 2324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.191719784044267e-06,
+      "loss": 0.8709,
+      "step": 2325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1910481761290025e-06,
+      "loss": 0.9252,
+      "step": 2326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1903763327857675e-06,
+      "loss": 0.8354,
+      "step": 2327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1897042540867514e-06,
+      "loss": 0.9169,
+      "step": 2328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.18903194010417e-06,
+      "loss": 0.846,
+      "step": 2329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.188359390910263e-06,
+      "loss": 0.8188,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.187686606577295e-06,
+      "loss": 1.1429,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1870135871775575e-06,
+      "loss": 0.865,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1863403327833665e-06,
+      "loss": 0.9829,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.185666843467062e-06,
+      "loss": 0.9533,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.184993119301011e-06,
+      "loss": 0.7915,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.184319160357606e-06,
+      "loss": 1.0291,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.183644966709263e-06,
+      "loss": 0.8566,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.182970538428424e-06,
+      "loss": 0.9398,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.182295875587557e-06,
+      "loss": 1.0052,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1816209782591524e-06,
+      "loss": 0.9298,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.180945846515731e-06,
+      "loss": 0.7951,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1802704804298335e-06,
+      "loss": 0.794,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.179594880074028e-06,
+      "loss": 1.0035,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.178919045520908e-06,
+      "loss": 1.1776,
+      "step": 2344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.178242976843093e-06,
+      "loss": 0.8599,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.177566674113226e-06,
+      "loss": 1.2292,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.176890137403973e-06,
+      "loss": 0.959,
+      "step": 2347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.176213366788031e-06,
+      "loss": 0.9405,
+      "step": 2348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.175536362338119e-06,
+      "loss": 1.0396,
+      "step": 2349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.17485912412698e-06,
+      "loss": 0.8388,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.174181652227383e-06,
+      "loss": 0.8505,
+      "step": 2351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.173503946712122e-06,
+      "loss": 1.1923,
+      "step": 2352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.172826007654018e-06,
+      "loss": 0.8992,
+      "step": 2353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.172147835125914e-06,
+      "loss": 0.7107,
+      "step": 2354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.17146942920068e-06,
+      "loss": 0.835,
+      "step": 2355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1707907899512104e-06,
+      "loss": 0.9372,
+      "step": 2356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1701119174504265e-06,
+      "loss": 0.8694,
+      "step": 2357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.169432811771271e-06,
+      "loss": 1.0199,
+      "step": 2358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.168753472986715e-06,
+      "loss": 0.8876,
+      "step": 2359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.168073901169753e-06,
+      "loss": 0.9634,
+      "step": 2360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.167394096393405e-06,
+      "loss": 0.885,
+      "step": 2361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.166714058730716e-06,
+      "loss": 1.007,
+      "step": 2362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1660337882547565e-06,
+      "loss": 0.9285,
+      "step": 2363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.165353285038619e-06,
+      "loss": 1.0332,
+      "step": 2364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.164672549155428e-06,
+      "loss": 1.0712,
+      "step": 2365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1639915806783256e-06,
+      "loss": 0.8825,
+      "step": 2366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1633103796804815e-06,
+      "loss": 0.83,
+      "step": 2367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1626289462350916e-06,
+      "loss": 0.9619,
+      "step": 2368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.161947280415376e-06,
+      "loss": 1.074,
+      "step": 2369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1612653822945795e-06,
+      "loss": 1.0992,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.160583251945972e-06,
+      "loss": 1.1627,
+      "step": 2371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.159900889442848e-06,
+      "loss": 1.0249,
+      "step": 2372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.159218294858527e-06,
+      "loss": 0.8855,
+      "step": 2373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.158535468266356e-06,
+      "loss": 0.7929,
+      "step": 2374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.157852409739701e-06,
+      "loss": 0.868,
+      "step": 2375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.15716911935196e-06,
+      "loss": 0.7969,
+      "step": 2376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.15648559717655e-06,
+      "loss": 1.0913,
+      "step": 2377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.155801843286917e-06,
+      "loss": 0.7859,
+      "step": 2378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.15511785775653e-06,
+      "loss": 0.8983,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1544336406588826e-06,
+      "loss": 1.0068,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.153749192067494e-06,
+      "loss": 0.9066,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.153064512055909e-06,
+      "loss": 1.0031,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.152379600697696e-06,
+      "loss": 0.9813,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.151694458066447e-06,
+      "loss": 0.9333,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.151009084235783e-06,
+      "loss": 1.0152,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.150323479279347e-06,
+      "loss": 0.9322,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.149637643270807e-06,
+      "loss": 1.0654,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.148951576283853e-06,
+      "loss": 0.817,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1482652783922075e-06,
+      "loss": 1.1025,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.14757874966961e-06,
+      "loss": 1.1547,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.14689199018983e-06,
+      "loss": 0.964,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.14620500002666e-06,
+      "loss": 0.862,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.145517779253914e-06,
+      "loss": 1.0674,
+      "step": 2393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.144830327945436e-06,
+      "loss": 0.9691,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.144142646175093e-06,
+      "loss": 1.0154,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.143454734016775e-06,
+      "loss": 0.938,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1427665915444e-06,
+      "loss": 1.0227,
+      "step": 2397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1420782188319065e-06,
+      "loss": 0.9142,
+      "step": 2398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.141389615953261e-06,
+      "loss": 0.9136,
+      "step": 2399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.140700782982454e-06,
+      "loss": 1.0439,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.140011719993502e-06,
+      "loss": 0.876,
+      "step": 2401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.139322427060442e-06,
+      "loss": 0.931,
+      "step": 2402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.138632904257341e-06,
+      "loss": 0.9063,
+      "step": 2403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.137943151658287e-06,
+      "loss": 1.0587,
+      "step": 2404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1372531693373925e-06,
+      "loss": 0.9774,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1365629573687984e-06,
+      "loss": 0.7927,
+      "step": 2406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.135872515826667e-06,
+      "loss": 1.0598,
+      "step": 2407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.135181844785186e-06,
+      "loss": 0.7907,
+      "step": 2408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.134490944318569e-06,
+      "loss": 1.0296,
+      "step": 2409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.133799814501052e-06,
+      "loss": 0.9564,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.133108455406897e-06,
+      "loss": 0.7194,
+      "step": 2411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.132416867110389e-06,
+      "loss": 1.1497,
+      "step": 2412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.131725049685843e-06,
+      "loss": 0.872,
+      "step": 2413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.131033003207592e-06,
+      "loss": 0.8752,
+      "step": 2414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.130340727749996e-06,
+      "loss": 0.8921,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.129648223387442e-06,
+      "loss": 0.917,
+      "step": 2416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.128955490194338e-06,
+      "loss": 1.1003,
+      "step": 2417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.128262528245118e-06,
+      "loss": 1.0014,
+      "step": 2418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.127569337614242e-06,
+      "loss": 0.9083,
+      "step": 2419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.126875918376192e-06,
+      "loss": 0.9644,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.126182270605476e-06,
+      "loss": 0.8377,
+      "step": 2421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.125488394376628e-06,
+      "loss": 0.8545,
+      "step": 2422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.124794289764202e-06,
+      "loss": 1.0247,
+      "step": 2423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.124099956842783e-06,
+      "loss": 1.0255,
+      "step": 2424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.123405395686974e-06,
+      "loss": 1.012,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.122710606371407e-06,
+      "loss": 0.9471,
+      "step": 2426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.122015588970737e-06,
+      "loss": 0.7059,
+      "step": 2427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1213203435596425e-06,
+      "loss": 0.8125,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.120624870212829e-06,
+      "loss": 0.8683,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119929169005026e-06,
+      "loss": 1.3407,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119233240010982e-06,
+      "loss": 0.845,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11853708330548e-06,
+      "loss": 0.9113,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.117840698963319e-06,
+      "loss": 0.8465,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.117144087059327e-06,
+      "loss": 1.1675,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116447247668352e-06,
+      "loss": 0.9437,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1157501808652724e-06,
+      "loss": 0.803,
+      "step": 2436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1150528867249864e-06,
+      "loss": 0.8941,
+      "step": 2437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114355365322418e-06,
+      "loss": 1.2692,
+      "step": 2438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.113657616732518e-06,
+      "loss": 1.0514,
+      "step": 2439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.112959641030257e-06,
+      "loss": 0.8576,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.112261438290633e-06,
+      "loss": 1.0214,
+      "step": 2441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111563008588669e-06,
+      "loss": 1.0628,
+      "step": 2442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11086435199941e-06,
+      "loss": 1.007,
+      "step": 2443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1101654685979285e-06,
+      "loss": 0.8617,
+      "step": 2444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.109466358459316e-06,
+      "loss": 1.0714,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108767021658695e-06,
+      "loss": 1.1484,
+      "step": 2446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108067458271207e-06,
+      "loss": 1.0369,
+      "step": 2447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1073676683720216e-06,
+      "loss": 0.9214,
+      "step": 2448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.106667652036331e-06,
+      "loss": 1.0162,
+      "step": 2449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105967409339351e-06,
+      "loss": 1.0225,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105266940356322e-06,
+      "loss": 0.9784,
+      "step": 2451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.104566245162512e-06,
+      "loss": 0.8025,
+      "step": 2452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.103865323833209e-06,
+      "loss": 1.1102,
+      "step": 2453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.103164176443727e-06,
+      "loss": 1.0883,
+      "step": 2454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102462803069404e-06,
+      "loss": 0.8716,
+      "step": 2455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1017612037856036e-06,
+      "loss": 1.1199,
+      "step": 2456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101059378667711e-06,
+      "loss": 0.932,
+      "step": 2457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.100357327791138e-06,
+      "loss": 1.1086,
+      "step": 2458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.099655051231321e-06,
+      "loss": 0.7941,
+      "step": 2459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0989525490637176e-06,
+      "loss": 0.9141,
+      "step": 2460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0982498213638125e-06,
+      "loss": 1.0351,
+      "step": 2461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097546868207114e-06,
+      "loss": 0.9666,
+      "step": 2462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096843689669155e-06,
+      "loss": 0.908,
+      "step": 2463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096140285825491e-06,
+      "loss": 0.8696,
+      "step": 2464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.095436656751702e-06,
+      "loss": 0.9934,
+      "step": 2465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094732802523394e-06,
+      "loss": 0.7219,
+      "step": 2466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094028723216196e-06,
+      "loss": 0.7932,
+      "step": 2467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.093324418905761e-06,
+      "loss": 1.033,
+      "step": 2468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.092619889667767e-06,
+      "loss": 0.9572,
+      "step": 2469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091915135577915e-06,
+      "loss": 0.8775,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091210156711931e-06,
+      "loss": 0.873,
+      "step": 2471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090504953145565e-06,
+      "loss": 0.9255,
+      "step": 2472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08979952495459e-06,
+      "loss": 0.8991,
+      "step": 2473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.089093872214807e-06,
+      "loss": 0.8108,
+      "step": 2474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088387995002035e-06,
+      "loss": 1.1737,
+      "step": 2475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.087681893392122e-06,
+      "loss": 1.0651,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.086975567460939e-06,
+      "loss": 0.9521,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.086269017284381e-06,
+      "loss": 0.9331,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0855622429383635e-06,
+      "loss": 0.9467,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.084855244498834e-06,
+      "loss": 0.9204,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.084148022041757e-06,
+      "loss": 0.9802,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.083440575643123e-06,
+      "loss": 0.9266,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.082732905378948e-06,
+      "loss": 1.1413,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.082025011325272e-06,
+      "loss": 1.0594,
+      "step": 2484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.081316893558156e-06,
+      "loss": 0.9664,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.080608552153688e-06,
+      "loss": 0.8588,
+      "step": 2486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.07989998718798e-06,
+      "loss": 0.8075,
+      "step": 2487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.079191198737167e-06,
+      "loss": 0.9046,
+      "step": 2488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.078482186877408e-06,
+      "loss": 0.9701,
+      "step": 2489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.077772951684886e-06,
+      "loss": 0.9951,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.077063493235809e-06,
+      "loss": 0.8266,
+      "step": 2491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.076353811606408e-06,
+      "loss": 0.9048,
+      "step": 2492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.075643906872938e-06,
+      "loss": 0.695,
+      "step": 2493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.074933779111679e-06,
+      "loss": 1.0137,
+      "step": 2494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.074223428398932e-06,
+      "loss": 0.9308,
+      "step": 2495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.073512854811026e-06,
+      "loss": 0.8736,
+      "step": 2496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.072802058424312e-06,
+      "loss": 0.9095,
+      "step": 2497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.072091039315165e-06,
+      "loss": 0.978,
+      "step": 2498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0713797975599825e-06,
+      "loss": 0.8631,
+      "step": 2499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.07066833323519e-06,
+      "loss": 1.0013,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0699566464172326e-06,
+      "loss": 0.9105,
+      "step": 2501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.069244737182581e-06,
+      "loss": 1.1299,
+      "step": 2502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.06853260560773e-06,
+      "loss": 1.017,
+      "step": 2503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.067820251769198e-06,
+      "loss": 1.0599,
+      "step": 2504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.067107675743528e-06,
+      "loss": 0.9005,
+      "step": 2505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.066394877607285e-06,
+      "loss": 0.9718,
+      "step": 2506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.065681857437061e-06,
+      "loss": 0.9717,
+      "step": 2507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.064968615309468e-06,
+      "loss": 1.0909,
+      "step": 2508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.064255151301144e-06,
+      "loss": 1.0547,
+      "step": 2509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.063541465488752e-06,
+      "loss": 0.8874,
+      "step": 2510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0628275579489775e-06,
+      "loss": 0.9591,
+      "step": 2511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.062113428758528e-06,
+      "loss": 0.9388,
+      "step": 2512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.061399077994138e-06,
+      "loss": 1.0513,
+      "step": 2513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.060684505732565e-06,
+      "loss": 0.9671,
+      "step": 2514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.059969712050588e-06,
+      "loss": 0.8681,
+      "step": 2515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.059254697025013e-06,
+      "loss": 0.8135,
+      "step": 2516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.058539460732668e-06,
+      "loss": 0.8027,
+      "step": 2517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.057824003250403e-06,
+      "loss": 1.1504,
+      "step": 2518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.057108324655098e-06,
+      "loss": 0.9563,
+      "step": 2519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.056392425023649e-06,
+      "loss": 0.8946,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.055676304432981e-06,
+      "loss": 1.0663,
+      "step": 2521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.054959962960041e-06,
+      "loss": 0.7307,
+      "step": 2522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.054243400681799e-06,
+      "loss": 1.1529,
+      "step": 2523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.053526617675251e-06,
+      "loss": 0.9176,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.052809614017413e-06,
+      "loss": 0.911,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.05209238978533e-06,
+      "loss": 0.9896,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.051374945056065e-06,
+      "loss": 1.1208,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.050657279906709e-06,
+      "loss": 0.8298,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.049939394414375e-06,
+      "loss": 0.8746,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0492212886562e-06,
+      "loss": 0.9392,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.048502962709342e-06,
+      "loss": 0.8442,
+      "step": 2531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.047784416650988e-06,
+      "loss": 0.9574,
+      "step": 2532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.047065650558344e-06,
+      "loss": 1.3257,
+      "step": 2533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.046346664508642e-06,
+      "loss": 0.9249,
+      "step": 2534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0456274585791375e-06,
+      "loss": 1.0227,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0449080328471084e-06,
+      "loss": 0.949,
+      "step": 2536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0441883873898565e-06,
+      "loss": 0.9935,
+      "step": 2537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.043468522284709e-06,
+      "loss": 0.9295,
+      "step": 2538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0427484376090144e-06,
+      "loss": 0.9812,
+      "step": 2539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.042028133440144e-06,
+      "loss": 0.8358,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.041307609855499e-06,
+      "loss": 1.119,
+      "step": 2541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.040586866932496e-06,
+      "loss": 1.3557,
+      "step": 2542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.03986590474858e-06,
+      "loss": 0.9347,
+      "step": 2543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0391447233812185e-06,
+      "loss": 0.9196,
+      "step": 2544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.038423322907901e-06,
+      "loss": 0.9909,
+      "step": 2545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.037701703406143e-06,
+      "loss": 1.0625,
+      "step": 2546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.036979864953483e-06,
+      "loss": 1.2342,
+      "step": 2547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0362578076274815e-06,
+      "loss": 0.9071,
+      "step": 2548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.035535531505725e-06,
+      "loss": 0.8573,
+      "step": 2549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.034813036665819e-06,
+      "loss": 1.0338,
+      "step": 2550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.034090323185398e-06,
+      "loss": 1.0673,
+      "step": 2551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.033367391142118e-06,
+      "loss": 0.8365,
+      "step": 2552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.032644240613655e-06,
+      "loss": 0.9659,
+      "step": 2553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.031920871677715e-06,
+      "loss": 1.081,
+      "step": 2554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0311972844120215e-06,
+      "loss": 1.0689,
+      "step": 2555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0304734788943254e-06,
+      "loss": 0.8112,
+      "step": 2556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.029749455202398e-06,
+      "loss": 0.9363,
+      "step": 2557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.029025213414037e-06,
+      "loss": 1.176,
+      "step": 2558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.028300753607062e-06,
+      "loss": 0.9279,
+      "step": 2559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0275760758593154e-06,
+      "loss": 0.8483,
+      "step": 2560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.026851180248665e-06,
+      "loss": 0.975,
+      "step": 2561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.026126066852998e-06,
+      "loss": 0.9939,
+      "step": 2562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.025400735750231e-06,
+      "loss": 0.8144,
+      "step": 2563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.024675187018299e-06,
+      "loss": 0.9126,
+      "step": 2564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.023949420735163e-06,
+      "loss": 1.0224,
+      "step": 2565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.023223436978806e-06,
+      "loss": 1.2689,
+      "step": 2566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.022497235827235e-06,
+      "loss": 0.9545,
+      "step": 2567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.021770817358481e-06,
+      "loss": 0.6864,
+      "step": 2568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.021044181650596e-06,
+      "loss": 0.89,
+      "step": 2569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.020317328781658e-06,
+      "loss": 1.082,
+      "step": 2570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.019590258829767e-06,
+      "loss": 0.8674,
+      "step": 2571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.018862971873049e-06,
+      "loss": 1.0246,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.018135467989646e-06,
+      "loss": 1.028,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.017407747257732e-06,
+      "loss": 1.0678,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.0166798097555e-06,
+      "loss": 1.1637,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.015951655561165e-06,
+      "loss": 1.0397,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.015223284752969e-06,
+      "loss": 1.1779,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.014494697409176e-06,
+      "loss": 1.186,
+      "step": 2578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.01376589360807e-06,
+      "loss": 1.1003,
+      "step": 2579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.013036873427963e-06,
+      "loss": 0.8704,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.012307636947187e-06,
+      "loss": 1.0474,
+      "step": 2581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.0115781842441006e-06,
+      "loss": 1.297,
+      "step": 2582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.010848515397081e-06,
+      "loss": 0.8542,
+      "step": 2583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.010118630484531e-06,
+      "loss": 0.9307,
+      "step": 2584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.00938852958488e-06,
+      "loss": 0.8234,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.008658212776572e-06,
+      "loss": 0.8584,
+      "step": 2586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.007927680138084e-06,
+      "loss": 0.857,
+      "step": 2587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.007196931747911e-06,
+      "loss": 0.8757,
+      "step": 2588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.006465967684569e-06,
+      "loss": 1.0301,
+      "step": 2589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.0057347880266025e-06,
+      "loss": 0.9718,
+      "step": 2590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.005003392852576e-06,
+      "loss": 0.7893,
+      "step": 2591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.004271782241078e-06,
+      "loss": 1.119,
+      "step": 2592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.00353995627072e-06,
+      "loss": 0.863,
+      "step": 2593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.002807915020137e-06,
+      "loss": 1.0219,
+      "step": 2594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.002075658567987e-06,
+      "loss": 0.9247,
+      "step": 2595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.00134318699295e-06,
+      "loss": 1.0103,
+      "step": 2596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.00061050037373e-06,
+      "loss": 0.8803,
+      "step": 2597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.999877598789055e-06,
+      "loss": 0.7494,
+      "step": 2598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.999144482317674e-06,
+      "loss": 0.9234,
+      "step": 2599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.998411151038362e-06,
+      "loss": 0.9062,
+      "step": 2600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9976776050299125e-06,
+      "loss": 0.9348,
+      "step": 2601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.996943844371149e-06,
+      "loss": 0.9997,
+      "step": 2602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.996209869140911e-06,
+      "loss": 1.0604,
+      "step": 2603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.995475679418065e-06,
+      "loss": 1.0737,
+      "step": 2604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.994741275281499e-06,
+      "loss": 0.955,
+      "step": 2605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.994006656810127e-06,
+      "loss": 1.0295,
+      "step": 2606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.993271824082881e-06,
+      "loss": 1.1063,
+      "step": 2607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.99253677717872e-06,
+      "loss": 0.8617,
+      "step": 2608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.991801516176624e-06,
+      "loss": 1.202,
+      "step": 2609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.991066041155599e-06,
+      "loss": 0.971,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.990330352194667e-06,
+      "loss": 1.0037,
+      "step": 2611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.989594449372883e-06,
+      "loss": 0.9203,
+      "step": 2612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.988858332769316e-06,
+      "loss": 0.9964,
+      "step": 2613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.988122002463064e-06,
+      "loss": 0.9321,
+      "step": 2614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.987385458533244e-06,
+      "loss": 0.9853,
+      "step": 2615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.986648701058999e-06,
+      "loss": 0.9984,
+      "step": 2616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.985911730119492e-06,
+      "loss": 0.9028,
+      "step": 2617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.985174545793912e-06,
+      "loss": 0.8199,
+      "step": 2618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.984437148161468e-06,
+      "loss": 0.9792,
+      "step": 2619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.983699537301394e-06,
+      "loss": 0.8038,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.982961713292947e-06,
+      "loss": 1.0658,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.982223676215406e-06,
+      "loss": 1.0377,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9814854261480715e-06,
+      "loss": 0.8457,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9807469631702695e-06,
+      "loss": 1.1374,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.980008287361347e-06,
+      "loss": 1.0478,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.979269398800676e-06,
+      "loss": 1.1262,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.978530297567649e-06,
+      "loss": 1.0051,
+      "step": 2627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.977790983741683e-06,
+      "loss": 0.9621,
+      "step": 2628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9770514574022176e-06,
+      "loss": 0.933,
+      "step": 2629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.976311718628713e-06,
+      "loss": 0.8661,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.975571767500656e-06,
+      "loss": 1.1698,
+      "step": 2631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.974831604097554e-06,
+      "loss": 1.035,
+      "step": 2632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.974091228498936e-06,
+      "loss": 0.9496,
+      "step": 2633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.973350640784357e-06,
+      "loss": 0.898,
+      "step": 2634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.972609841033393e-06,
+      "loss": 0.9609,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.97186882932564e-06,
+      "loss": 1.0409,
+      "step": 2636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.971127605740723e-06,
+      "loss": 0.8878,
+      "step": 2637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.970386170358287e-06,
+      "loss": 0.9911,
+      "step": 2638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9696445232579966e-06,
+      "loss": 0.8893,
+      "step": 2639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.968902664519542e-06,
+      "loss": 1.0484,
+      "step": 2640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.968160594222638e-06,
+      "loss": 0.9679,
+      "step": 2641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.967418312447018e-06,
+      "loss": 0.7957,
+      "step": 2642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.96667581927244e-06,
+      "loss": 0.8141,
+      "step": 2643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9659331147786865e-06,
+      "loss": 1.026,
+      "step": 2644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.965190199045559e-06,
+      "loss": 0.8327,
+      "step": 2645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.964447072152887e-06,
+      "loss": 1.0568,
+      "step": 2646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.963703734180516e-06,
+      "loss": 0.8787,
+      "step": 2647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.962960185208318e-06,
+      "loss": 1.0242,
+      "step": 2648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.962216425316189e-06,
+      "loss": 1.0282,
+      "step": 2649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.961472454584045e-06,
+      "loss": 1.1661,
+      "step": 2650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.960728273091826e-06,
+      "loss": 0.8773,
+      "step": 2651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.959983880919494e-06,
+      "loss": 1.0649,
+      "step": 2652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.959239278147033e-06,
+      "loss": 0.9757,
+      "step": 2653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.958494464854452e-06,
+      "loss": 0.7974,
+      "step": 2654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9577494411217805e-06,
+      "loss": 1.0374,
+      "step": 2655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.957004207029072e-06,
+      "loss": 0.9779,
+      "step": 2656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9562587626564e-06,
+      "loss": 0.8656,
+      "step": 2657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.955513108083864e-06,
+      "loss": 0.9525,
+      "step": 2658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9547672433915844e-06,
+      "loss": 1.0833,
+      "step": 2659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.954021168659703e-06,
+      "loss": 0.8195,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.953274883968387e-06,
+      "loss": 1.0041,
+      "step": 2661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.952528389397824e-06,
+      "loss": 0.9358,
+      "step": 2662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.951781685028226e-06,
+      "loss": 1.1057,
+      "step": 2663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.951034770939823e-06,
+      "loss": 1.0556,
+      "step": 2664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.950287647212875e-06,
+      "loss": 0.9416,
+      "step": 2665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9495403139276584e-06,
+      "loss": 1.0965,
+      "step": 2666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.948792771164473e-06,
+      "loss": 0.9436,
+      "step": 2667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.948045019003644e-06,
+      "loss": 0.841,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.947297057525517e-06,
+      "loss": 0.7557,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.946548886810459e-06,
+      "loss": 1.1596,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.945800506938863e-06,
+      "loss": 1.1219,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9450519179911415e-06,
+      "loss": 0.8848,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.94430312004773e-06,
+      "loss": 0.7982,
+      "step": 2673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.943554113189087e-06,
+      "loss": 0.8329,
+      "step": 2674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.942804897495693e-06,
+      "loss": 1.0377,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9420554730480516e-06,
+      "loss": 1.0421,
+      "step": 2676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.941305839926688e-06,
+      "loss": 0.8797,
+      "step": 2677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.940555998212151e-06,
+      "loss": 0.8476,
+      "step": 2678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.939805947985011e-06,
+      "loss": 0.861,
+      "step": 2679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.939055689325858e-06,
+      "loss": 0.8023,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.938305222315312e-06,
+      "loss": 0.9713,
+      "step": 2681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.937554547034006e-06,
+      "loss": 0.8407,
+      "step": 2682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.936803663562604e-06,
+      "loss": 1.0309,
+      "step": 2683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.936052571981785e-06,
+      "loss": 0.9783,
+      "step": 2684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.935301272372256e-06,
+      "loss": 0.967,
+      "step": 2685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.934549764814744e-06,
+      "loss": 1.0677,
+      "step": 2686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.933798049389998e-06,
+      "loss": 0.9838,
+      "step": 2687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.933046126178788e-06,
+      "loss": 0.8914,
+      "step": 2688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.932293995261911e-06,
+      "loss": 0.8832,
+      "step": 2689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.931541656720181e-06,
+      "loss": 1.1334,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9307891106344395e-06,
+      "loss": 1.0899,
+      "step": 2691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.930036357085544e-06,
+      "loss": 1.0349,
+      "step": 2692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.929283396154381e-06,
+      "loss": 0.7373,
+      "step": 2693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.928530227921854e-06,
+      "loss": 0.9768,
+      "step": 2694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.927776852468892e-06,
+      "loss": 1.0718,
+      "step": 2695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9270232698764445e-06,
+      "loss": 0.7544,
+      "step": 2696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.926269480225484e-06,
+      "loss": 1.0599,
+      "step": 2697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.925515483597007e-06,
+      "loss": 0.7094,
+      "step": 2698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.924761280072026e-06,
+      "loss": 0.9542,
+      "step": 2699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.924006869731584e-06,
+      "loss": 0.8975,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.923252252656742e-06,
+      "loss": 1.0897,
+      "step": 2701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.922497428928581e-06,
+      "loss": 1.0104,
+      "step": 2702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.921742398628209e-06,
+      "loss": 0.9648,
+      "step": 2703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9209871618367545e-06,
+      "loss": 0.9775,
+      "step": 2704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.920231718635365e-06,
+      "loss": 0.9733,
+      "step": 2705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.919476069105217e-06,
+      "loss": 0.8774,
+      "step": 2706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9187202133275e-06,
+      "loss": 1.1875,
+      "step": 2707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9179641513834334e-06,
+      "loss": 0.8037,
+      "step": 2708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9172078833542555e-06,
+      "loss": 0.8601,
+      "step": 2709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.916451409321228e-06,
+      "loss": 0.9621,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.915694729365633e-06,
+      "loss": 0.7833,
+      "step": 2711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914937843568775e-06,
+      "loss": 0.7765,
+      "step": 2712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914180752011983e-06,
+      "loss": 0.9161,
+      "step": 2713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913423454776607e-06,
+      "loss": 0.8773,
+      "step": 2714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912665951944016e-06,
+      "loss": 0.8775,
+      "step": 2715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9119082435956055e-06,
+      "loss": 1.0859,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.91115032981279e-06,
+      "loss": 1.0935,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.910392210677009e-06,
+      "loss": 0.8797,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9096338862697206e-06,
+      "loss": 0.8954,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9088753566724075e-06,
+      "loss": 1.0249,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9081166219665736e-06,
+      "loss": 1.0245,
+      "step": 2721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907357682233745e-06,
+      "loss": 0.8705,
+      "step": 2722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906598537555469e-06,
+      "loss": 0.9589,
+      "step": 2723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905839188013316e-06,
+      "loss": 0.9523,
+      "step": 2724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905079633688879e-06,
+      "loss": 0.9988,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904319874663771e-06,
+      "loss": 0.7488,
+      "step": 2726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.903559911019628e-06,
+      "loss": 0.9469,
+      "step": 2727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9027997428381095e-06,
+      "loss": 0.7759,
+      "step": 2728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9020393702008935e-06,
+      "loss": 0.9342,
+      "step": 2729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.901278793189683e-06,
+      "loss": 1.0867,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.900518011886202e-06,
+      "loss": 1.0903,
+      "step": 2731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.899757026372197e-06,
+      "loss": 0.8959,
+      "step": 2732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.898995836729434e-06,
+      "loss": 0.8601,
+      "step": 2733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.898234443039704e-06,
+      "loss": 1.0701,
+      "step": 2734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.897472845384819e-06,
+      "loss": 0.9708,
+      "step": 2735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8967110438466115e-06,
+      "loss": 0.9711,
+      "step": 2736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.895949038506938e-06,
+      "loss": 0.9794,
+      "step": 2737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.895186829447676e-06,
+      "loss": 0.8646,
+      "step": 2738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.894424416750723e-06,
+      "loss": 0.9558,
+      "step": 2739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8936618004980035e-06,
+      "loss": 0.9935,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.892898980771458e-06,
+      "loss": 1.0338,
+      "step": 2741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.892135957653051e-06,
+      "loss": 1.0308,
+      "step": 2742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.891372731224772e-06,
+      "loss": 0.8039,
+      "step": 2743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.890609301568627e-06,
+      "loss": 0.9429,
+      "step": 2744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.889845668766648e-06,
+      "loss": 0.9663,
+      "step": 2745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.889081832900886e-06,
+      "loss": 0.9829,
+      "step": 2746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.888317794053417e-06,
+      "loss": 0.8141,
+      "step": 2747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.887553552306336e-06,
+      "loss": 0.7318,
+      "step": 2748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8867891077417585e-06,
+      "loss": 0.7234,
+      "step": 2749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.886024460441827e-06,
+      "loss": 1.0528,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.885259610488702e-06,
+      "loss": 0.8824,
+      "step": 2751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8844945579645664e-06,
+      "loss": 0.9795,
+      "step": 2752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.883729302951625e-06,
+      "loss": 0.9262,
+      "step": 2753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.882963845532104e-06,
+      "loss": 0.9753,
+      "step": 2754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.882198185788253e-06,
+      "loss": 1.0247,
+      "step": 2755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.881432323802341e-06,
+      "loss": 0.8908,
+      "step": 2756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.88066625965666e-06,
+      "loss": 0.9432,
+      "step": 2757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.879899993433523e-06,
+      "loss": 0.9784,
+      "step": 2758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.879133525215266e-06,
+      "loss": 1.0825,
+      "step": 2759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.878366855084246e-06,
+      "loss": 0.7847,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.877599983122841e-06,
+      "loss": 0.9108,
+      "step": 2761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.876832909413453e-06,
+      "loss": 0.9932,
+      "step": 2762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.876065634038502e-06,
+      "loss": 0.9881,
+      "step": 2763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.875298157080432e-06,
+      "loss": 0.8183,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.87453047862171e-06,
+      "loss": 0.8591,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.873762598744821e-06,
+      "loss": 0.8704,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.872994517532276e-06,
+      "loss": 1.0732,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8722262350666015e-06,
+      "loss": 0.9458,
+      "step": 2768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.871457751430353e-06,
+      "loss": 1.1114,
+      "step": 2769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8706890667061025e-06,
+      "loss": 0.971,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.869920180976445e-06,
+      "loss": 1.1274,
+      "step": 2771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8691510943239975e-06,
+      "loss": 0.7605,
+      "step": 2772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8683818068313985e-06,
+      "loss": 0.9043,
+      "step": 2773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8676123185813075e-06,
+      "loss": 1.077,
+      "step": 2774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.866842629656406e-06,
+      "loss": 1.096,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.866072740139398e-06,
+      "loss": 1.0177,
+      "step": 2776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.865302650113006e-06,
+      "loss": 0.9357,
+      "step": 2777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.864532359659978e-06,
+      "loss": 0.903,
+      "step": 2778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.863761868863082e-06,
+      "loss": 0.8957,
+      "step": 2779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.862991177805104e-06,
+      "loss": 1.1514,
+      "step": 2780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.862220286568858e-06,
+      "loss": 0.8428,
+      "step": 2781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8614491952371756e-06,
+      "loss": 1.1242,
+      "step": 2782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.86067790389291e-06,
+      "loss": 0.8054,
+      "step": 2783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.859906412618935e-06,
+      "loss": 1.2093,
+      "step": 2784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.85913472149815e-06,
+      "loss": 0.9945,
+      "step": 2785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.858362830613472e-06,
+      "loss": 0.7778,
+      "step": 2786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.857590740047839e-06,
+      "loss": 1.0078,
+      "step": 2787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8568184498842144e-06,
+      "loss": 1.1242,
+      "step": 2788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8560459602055804e-06,
+      "loss": 0.9869,
+      "step": 2789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.855273271094941e-06,
+      "loss": 0.9784,
+      "step": 2790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8545003826353185e-06,
+      "loss": 1.0542,
+      "step": 2791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.853727294909763e-06,
+      "loss": 1.139,
+      "step": 2792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.852954008001343e-06,
+      "loss": 0.9511,
+      "step": 2793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.852180521993147e-06,
+      "loss": 0.6632,
+      "step": 2794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.851406836968286e-06,
+      "loss": 0.9383,
+      "step": 2795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.850632953009893e-06,
+      "loss": 0.9786,
+      "step": 2796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8498588702011195e-06,
+      "loss": 1.1394,
+      "step": 2797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.849084588625144e-06,
+      "loss": 0.9034,
+      "step": 2798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.848310108365161e-06,
+      "loss": 1.0086,
+      "step": 2799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.847535429504387e-06,
+      "loss": 0.9096,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8467605521260656e-06,
+      "loss": 0.8642,
+      "step": 2801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.845985476313453e-06,
+      "loss": 1.0171,
+      "step": 2802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8452102021498335e-06,
+      "loss": 0.9324,
+      "step": 2803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.844434729718509e-06,
+      "loss": 0.9764,
+      "step": 2804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.843659059102805e-06,
+      "loss": 0.9629,
+      "step": 2805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.842883190386066e-06,
+      "loss": 0.9794,
+      "step": 2806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.84210712365166e-06,
+      "loss": 1.0965,
+      "step": 2807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.841330858982975e-06,
+      "loss": 1.0717,
+      "step": 2808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.840554396463421e-06,
+      "loss": 1.1481,
+      "step": 2809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.839777736176427e-06,
+      "loss": 1.0776,
+      "step": 2810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.839000878205448e-06,
+      "loss": 0.7658,
+      "step": 2811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.838223822633955e-06,
+      "loss": 1.0859,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.837446569545444e-06,
+      "loss": 0.9488,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.836669119023431e-06,
+      "loss": 0.8919,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.83589147115145e-06,
+      "loss": 0.9451,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.835113626013063e-06,
+      "loss": 1.0162,
+      "step": 2816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.834335583691848e-06,
+      "loss": 1.0735,
+      "step": 2817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.833557344271405e-06,
+      "loss": 1.0893,
+      "step": 2818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.832778907835355e-06,
+      "loss": 0.8911,
+      "step": 2819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.832000274467344e-06,
+      "loss": 0.748,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.831221444251033e-06,
+      "loss": 1.0105,
+      "step": 2821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.830442417270109e-06,
+      "loss": 1.0361,
+      "step": 2822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.829663193608278e-06,
+      "loss": 1.0661,
+      "step": 2823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.828883773349268e-06,
+      "loss": 1.1904,
+      "step": 2824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.828104156576826e-06,
+      "loss": 0.9388,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.827324343374723e-06,
+      "loss": 1.235,
+      "step": 2826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.826544333826751e-06,
+      "loss": 0.8404,
+      "step": 2827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.825764128016719e-06,
+      "loss": 0.8038,
+      "step": 2828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.824983726028462e-06,
+      "loss": 1.1004,
+      "step": 2829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.824203127945834e-06,
+      "loss": 1.1069,
+      "step": 2830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8234223338527105e-06,
+      "loss": 0.8658,
+      "step": 2831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.822641343832987e-06,
+      "loss": 0.7181,
+      "step": 2832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.821860157970581e-06,
+      "loss": 1.1232,
+      "step": 2833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.821078776349431e-06,
+      "loss": 0.7647,
+      "step": 2834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.820297199053497e-06,
+      "loss": 0.9346,
+      "step": 2835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.819515426166759e-06,
+      "loss": 0.8797,
+      "step": 2836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.818733457773217e-06,
+      "loss": 0.8636,
+      "step": 2837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.817951293956895e-06,
+      "loss": 1.0507,
+      "step": 2838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.817168934801838e-06,
+      "loss": 0.8214,
+      "step": 2839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.816386380392107e-06,
+      "loss": 0.7938,
+      "step": 2840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8156036308117895e-06,
+      "loss": 0.7879,
+      "step": 2841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.814820686144992e-06,
+      "loss": 0.9677,
+      "step": 2842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.81403754647584e-06,
+      "loss": 0.8837,
+      "step": 2843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8132542118884835e-06,
+      "loss": 0.7743,
+      "step": 2844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8124706824670915e-06,
+      "loss": 1.0627,
+      "step": 2845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8116869582958546e-06,
+      "loss": 0.8837,
+      "step": 2846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.810903039458982e-06,
+      "loss": 0.9299,
+      "step": 2847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8101189260407086e-06,
+      "loss": 0.8153,
+      "step": 2848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.809334618125285e-06,
+      "loss": 0.7995,
+      "step": 2849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.808550115796986e-06,
+      "loss": 1.1485,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.807765419140105e-06,
+      "loss": 0.869,
+      "step": 2851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.80698052823896e-06,
+      "loss": 0.8123,
+      "step": 2852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.806195443177886e-06,
+      "loss": 0.9202,
+      "step": 2853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.80541016404124e-06,
+      "loss": 1.0557,
+      "step": 2854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.804624690913401e-06,
+      "loss": 0.9507,
+      "step": 2855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.803839023878767e-06,
+      "loss": 1.1463,
+      "step": 2856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.80305316302176e-06,
+      "loss": 1.1706,
+      "step": 2857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8022671084268185e-06,
+      "loss": 0.8909,
+      "step": 2858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.801480860178406e-06,
+      "loss": 0.9442,
+      "step": 2859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.800694418361002e-06,
+      "loss": 0.8815,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.799907783059111e-06,
+      "loss": 0.9049,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.799120954357258e-06,
+      "loss": 1.0631,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.798333932339987e-06,
+      "loss": 0.8125,
+      "step": 2863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.797546717091863e-06,
+      "loss": 0.8176,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.796759308697473e-06,
+      "loss": 0.9898,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.795971707241423e-06,
+      "loss": 0.9235,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.795183912808341e-06,
+      "loss": 1.1545,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.794395925482876e-06,
+      "loss": 0.8292,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.793607745349697e-06,
+      "loss": 1.105,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.792819372493494e-06,
+      "loss": 0.9857,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7920308069989764e-06,
+      "loss": 1.1532,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7912420489508775e-06,
+      "loss": 1.0268,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.790453098433947e-06,
+      "loss": 0.8412,
+      "step": 2873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.789663955532961e-06,
+      "loss": 0.9191,
+      "step": 2874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.788874620332708e-06,
+      "loss": 1.122,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.788085092918007e-06,
+      "loss": 1.0056,
+      "step": 2876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.787295373373688e-06,
+      "loss": 1.0216,
+      "step": 2877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.786505461784611e-06,
+      "loss": 0.7896,
+      "step": 2878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.785715358235649e-06,
+      "loss": 0.9848,
+      "step": 2879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.784925062811698e-06,
+      "loss": 0.8343,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.784134575597677e-06,
+      "loss": 1.0993,
+      "step": 2881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.783343896678524e-06,
+      "loss": 0.9762,
+      "step": 2882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.782553026139195e-06,
+      "loss": 0.7929,
+      "step": 2883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.781761964064672e-06,
+      "loss": 1.1848,
+      "step": 2884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7809707105399525e-06,
+      "loss": 0.9826,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.780179265650057e-06,
+      "loss": 1.0781,
+      "step": 2886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7793876294800265e-06,
+      "loss": 1.1268,
+      "step": 2887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.778595802114923e-06,
+      "loss": 0.9998,
+      "step": 2888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.777803783639827e-06,
+      "loss": 0.92,
+      "step": 2889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.777011574139842e-06,
+      "loss": 1.0166,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.776219173700088e-06,
+      "loss": 0.9029,
+      "step": 2891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.775426582405713e-06,
+      "loss": 0.999,
+      "step": 2892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.774633800341877e-06,
+      "loss": 1.1518,
+      "step": 2893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.773840827593765e-06,
+      "loss": 0.9275,
+      "step": 2894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.773047664246584e-06,
+      "loss": 0.9437,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.772254310385557e-06,
+      "loss": 1.0086,
+      "step": 2896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7714607660959326e-06,
+      "loss": 0.7944,
+      "step": 2897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.770667031462974e-06,
+      "loss": 0.9558,
+      "step": 2898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7698731065719695e-06,
+      "loss": 1.0307,
+      "step": 2899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.769078991508226e-06,
+      "loss": 0.9667,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.76828468635707e-06,
+      "loss": 0.9969,
+      "step": 2901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.767490191203853e-06,
+      "loss": 1.0142,
+      "step": 2902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7666955061339405e-06,
+      "loss": 0.8911,
+      "step": 2903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7659006312327204e-06,
+      "loss": 1.0494,
+      "step": 2904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.765105566585604e-06,
+      "loss": 1.0229,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.76431031227802e-06,
+      "loss": 1.0967,
+      "step": 2906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.763514868395421e-06,
+      "loss": 0.8342,
+      "step": 2907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.762719235023274e-06,
+      "loss": 1.0053,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.76192341224707e-06,
+      "loss": 0.8375,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.761127400152321e-06,
+      "loss": 0.9901,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.76033119882456e-06,
+      "loss": 0.9149,
+      "step": 2911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.759534808349337e-06,
+      "loss": 1.0135,
+      "step": 2912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.758738228812224e-06,
+      "loss": 0.8248,
+      "step": 2913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7579414602988144e-06,
+      "loss": 0.9157,
+      "step": 2914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.75714450289472e-06,
+      "loss": 0.9965,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.756347356685573e-06,
+      "loss": 0.8312,
+      "step": 2916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7555500217570295e-06,
+      "loss": 0.9384,
+      "step": 2917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.75475249819476e-06,
+      "loss": 0.8476,
+      "step": 2918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.753954786084462e-06,
+      "loss": 1.0674,
+      "step": 2919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.753156885511845e-06,
+      "loss": 0.8742,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.752358796562647e-06,
+      "loss": 0.8825,
+      "step": 2921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.751560519322621e-06,
+      "loss": 0.956,
+      "step": 2922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.750762053877543e-06,
+      "loss": 0.9224,
+      "step": 2923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.749963400313206e-06,
+      "loss": 0.9427,
+      "step": 2924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.749164558715427e-06,
+      "loss": 1.0677,
+      "step": 2925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.748365529170041e-06,
+      "loss": 0.8543,
+      "step": 2926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7475663117629035e-06,
+      "loss": 0.8991,
+      "step": 2927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.746766906579891e-06,
+      "loss": 0.94,
+      "step": 2928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.745967313706898e-06,
+      "loss": 0.7597,
+      "step": 2929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.745167533229843e-06,
+      "loss": 0.8865,
+      "step": 2930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.74436756523466e-06,
+      "loss": 0.9761,
+      "step": 2931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7435674098073076e-06,
+      "loss": 0.7734,
+      "step": 2932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.742767067033761e-06,
+      "loss": 0.9891,
+      "step": 2933
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7419665370000175e-06,
+      "loss": 0.8091,
+      "step": 2934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.741165819792094e-06,
+      "loss": 0.9532,
+      "step": 2935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.740364915496026e-06,
+      "loss": 1.1199,
+      "step": 2936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.739563824197874e-06,
+      "loss": 0.8885,
+      "step": 2937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.738762545983712e-06,
+      "loss": 0.7341,
+      "step": 2938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7379610809396395e-06,
+      "loss": 0.7823,
+      "step": 2939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.737159429151772e-06,
+      "loss": 1.0084,
+      "step": 2940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.736357590706248e-06,
+      "loss": 0.9854,
+      "step": 2941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7355555656892244e-06,
+      "loss": 0.846,
+      "step": 2942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.734753354186879e-06,
+      "loss": 1.0073,
+      "step": 2943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.733950956285411e-06,
+      "loss": 0.9657,
+      "step": 2944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.733148372071036e-06,
+      "loss": 0.8099,
+      "step": 2945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.732345601629991e-06,
+      "loss": 1.008,
+      "step": 2946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.731542645048536e-06,
+      "loss": 0.9804,
+      "step": 2947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.730739502412946e-06,
+      "loss": 0.8989,
+      "step": 2948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7299361738095205e-06,
+      "loss": 1.0996,
+      "step": 2949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.729132659324577e-06,
+      "loss": 0.8881,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.728328959044453e-06,
+      "loss": 1.1627,
+      "step": 2951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.727525073055505e-06,
+      "loss": 0.9498,
+      "step": 2952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.726721001444111e-06,
+      "loss": 0.7894,
+      "step": 2953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.725916744296669e-06,
+      "loss": 0.8205,
+      "step": 2954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.725112301699595e-06,
+      "loss": 0.905,
+      "step": 2955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7243076737393285e-06,
+      "loss": 0.9272,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.723502860502325e-06,
+      "loss": 1.0368,
+      "step": 2957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.722697862075062e-06,
+      "loss": 0.7315,
+      "step": 2958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.721892678544038e-06,
+      "loss": 0.8794,
+      "step": 2959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.721087309995766e-06,
+      "loss": 1.0733,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.720281756516787e-06,
+      "loss": 0.9513,
+      "step": 2961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.719476018193657e-06,
+      "loss": 0.9955,
+      "step": 2962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.71867009511295e-06,
+      "loss": 0.8348,
+      "step": 2963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.717863987361265e-06,
+      "loss": 1.0419,
+      "step": 2964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.717057695025215e-06,
+      "loss": 0.9154,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.71625121819144e-06,
+      "loss": 0.913,
+      "step": 2966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.715444556946594e-06,
+      "loss": 0.9571,
+      "step": 2967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.714637711377352e-06,
+      "loss": 1.1672,
+      "step": 2968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.713830681570411e-06,
+      "loss": 1.0206,
+      "step": 2969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.713023467612485e-06,
+      "loss": 0.8163,
+      "step": 2970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.71221606959031e-06,
+      "loss": 0.7672,
+      "step": 2971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.711408487590641e-06,
+      "loss": 0.9236,
+      "step": 2972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.710600721700251e-06,
+      "loss": 0.8967,
+      "step": 2973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.709792772005937e-06,
+      "loss": 1.0023,
+      "step": 2974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.708984638594509e-06,
+      "loss": 1.0547,
+      "step": 2975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.708176321552806e-06,
+      "loss": 0.9954,
+      "step": 2976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.707367820967678e-06,
+      "loss": 0.8708,
+      "step": 2977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.706559136926001e-06,
+      "loss": 0.8108,
+      "step": 2978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.705750269514665e-06,
+      "loss": 0.8816,
+      "step": 2979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.704941218820585e-06,
+      "loss": 0.8904,
+      "step": 2980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.704131984930691e-06,
+      "loss": 1.1285,
+      "step": 2981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.703322567931939e-06,
+      "loss": 0.8329,
+      "step": 2982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.702512967911298e-06,
+      "loss": 1.0239,
+      "step": 2983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7017031849557606e-06,
+      "loss": 0.927,
+      "step": 2984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700893219152336e-06,
+      "loss": 1.2445,
+      "step": 2985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700083070588057e-06,
+      "loss": 1.2013,
+      "step": 2986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699272739349973e-06,
+      "loss": 0.9241,
+      "step": 2987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698462225525156e-06,
+      "loss": 1.0359,
+      "step": 2988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.697651529200692e-06,
+      "loss": 0.9663,
+      "step": 2989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696840650463693e-06,
+      "loss": 0.9415,
+      "step": 2990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696029589401288e-06,
+      "loss": 0.9096,
+      "step": 2991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695218346100625e-06,
+      "loss": 1.1242,
+      "step": 2992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69440692064887e-06,
+      "loss": 0.8452,
+      "step": 2993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693595313133213e-06,
+      "loss": 0.95,
+      "step": 2994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692783523640861e-06,
+      "loss": 1.119,
+      "step": 2995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69197155225904e-06,
+      "loss": 0.8879,
+      "step": 2996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.691159399074995e-06,
+      "loss": 1.0193,
+      "step": 2997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690347064175995e-06,
+      "loss": 0.9862,
+      "step": 2998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6895345476493226e-06,
+      "loss": 0.9465,
+      "step": 2999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.688721849582285e-06,
+      "loss": 0.8747,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0279,
+      "eval_samples_per_second": 352.344,
+      "eval_steps_per_second": 44.051,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.687908970062204e-06,
+      "loss": 0.9518,
+      "step": 3001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.687095909176424e-06,
+      "loss": 0.9898,
+      "step": 3002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.68628266701231e-06,
+      "loss": 0.8518,
+      "step": 3003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.685469243657244e-06,
+      "loss": 0.7503,
+      "step": 3004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.684655639198627e-06,
+      "loss": 0.7467,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683841853723883e-06,
+      "loss": 0.8044,
+      "step": 3006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68302788732045e-06,
+      "loss": 0.8377,
+      "step": 3007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6822137400757935e-06,
+      "loss": 0.8644,
+      "step": 3008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.681399412077388e-06,
+      "loss": 0.8762,
+      "step": 3009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.680584903412738e-06,
+      "loss": 1.2594,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.679770214169358e-06,
+      "loss": 0.986,
+      "step": 3011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.67895534443479e-06,
+      "loss": 0.9514,
+      "step": 3012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.678140294296588e-06,
+      "loss": 0.9434,
+      "step": 3013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.677325063842332e-06,
+      "loss": 0.9443,
+      "step": 3014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6765096531596185e-06,
+      "loss": 0.968,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.675694062336062e-06,
+      "loss": 1.0099,
+      "step": 3016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.674878291459297e-06,
+      "loss": 1.0249,
+      "step": 3017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6740623406169805e-06,
+      "loss": 0.8514,
+      "step": 3018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.673246209896784e-06,
+      "loss": 1.0592,
+      "step": 3019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6724298993864025e-06,
+      "loss": 0.9291,
+      "step": 3020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.671613409173547e-06,
+      "loss": 0.8889,
+      "step": 3021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.670796739345951e-06,
+      "loss": 0.9185,
+      "step": 3022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.669979889991363e-06,
+      "loss": 0.8975,
+      "step": 3023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.669162861197556e-06,
+      "loss": 0.9568,
+      "step": 3024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.668345653052318e-06,
+      "loss": 0.937,
+      "step": 3025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.667528265643459e-06,
+      "loss": 0.8358,
+      "step": 3026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.666710699058807e-06,
+      "loss": 0.9835,
+      "step": 3027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.665892953386209e-06,
+      "loss": 0.9471,
+      "step": 3028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.665075028713532e-06,
+      "loss": 1.1176,
+      "step": 3029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6642569251286625e-06,
+      "loss": 0.9811,
+      "step": 3030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.663438642719505e-06,
+      "loss": 1.0107,
+      "step": 3031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.662620181573984e-06,
+      "loss": 0.9262,
+      "step": 3032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.661801541780043e-06,
+      "loss": 1.0861,
+      "step": 3033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.660982723425644e-06,
+      "loss": 0.8239,
+      "step": 3034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.660163726598772e-06,
+      "loss": 1.2042,
+      "step": 3035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.659344551387424e-06,
+      "loss": 1.0402,
+      "step": 3036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6585251978796244e-06,
+      "loss": 1.019,
+      "step": 3037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.657705666163408e-06,
+      "loss": 1.0939,
+      "step": 3038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.656885956326839e-06,
+      "loss": 0.9064,
+      "step": 3039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.65606606845799e-06,
+      "loss": 0.9934,
+      "step": 3040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.655246002644962e-06,
+      "loss": 1.0363,
+      "step": 3041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.654425758975868e-06,
+      "loss": 0.9093,
+      "step": 3042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.653605337538843e-06,
+      "loss": 1.0595,
+      "step": 3043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.652784738422044e-06,
+      "loss": 0.9243,
+      "step": 3044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.651963961713643e-06,
+      "loss": 1.0185,
+      "step": 3045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.651143007501832e-06,
+      "loss": 0.8653,
+      "step": 3046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.650321875874822e-06,
+      "loss": 0.958,
+      "step": 3047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.649500566920845e-06,
+      "loss": 0.8905,
+      "step": 3048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.64867908072815e-06,
+      "loss": 0.9626,
+      "step": 3049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.647857417385004e-06,
+      "loss": 0.8212,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.647035576979698e-06,
+      "loss": 0.8393,
+      "step": 3051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.646213559600535e-06,
+      "loss": 0.9107,
+      "step": 3052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.645391365335843e-06,
+      "loss": 0.8557,
+      "step": 3053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.644568994273967e-06,
+      "loss": 1.0617,
+      "step": 3054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.643746446503269e-06,
+      "loss": 0.9377,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.642923722112133e-06,
+      "loss": 1.049,
+      "step": 3056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.642100821188961e-06,
+      "loss": 0.8781,
+      "step": 3057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.641277743822171e-06,
+      "loss": 0.8521,
+      "step": 3058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.640454490100206e-06,
+      "loss": 0.9214,
+      "step": 3059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.639631060111523e-06,
+      "loss": 0.9106,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6388074539445995e-06,
+      "loss": 1.089,
+      "step": 3061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.637983671687931e-06,
+      "loss": 0.891,
+      "step": 3062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.637159713430035e-06,
+      "loss": 0.915,
+      "step": 3063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.636335579259445e-06,
+      "loss": 1.1608,
+      "step": 3064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.635511269264714e-06,
+      "loss": 1.0218,
+      "step": 3065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.634686783534412e-06,
+      "loss": 1.1135,
+      "step": 3066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.633862122157133e-06,
+      "loss": 0.9779,
+      "step": 3067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.633037285221485e-06,
+      "loss": 0.9657,
+      "step": 3068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.632212272816098e-06,
+      "loss": 0.8419,
+      "step": 3069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.631387085029618e-06,
+      "loss": 0.9165,
+      "step": 3070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.630561721950713e-06,
+      "loss": 0.8944,
+      "step": 3071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.629736183668067e-06,
+      "loss": 0.9364,
+      "step": 3072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.628910470270384e-06,
+      "loss": 0.8593,
+      "step": 3073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.628084581846387e-06,
+      "loss": 0.9705,
+      "step": 3074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.627258518484819e-06,
+      "loss": 0.9321,
+      "step": 3075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.626432280274438e-06,
+      "loss": 0.7933,
+      "step": 3076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.625605867304026e-06,
+      "loss": 1.0441,
+      "step": 3077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.624779279662378e-06,
+      "loss": 1.0429,
+      "step": 3078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.623952517438313e-06,
+      "loss": 0.9669,
+      "step": 3079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.623125580720666e-06,
+      "loss": 0.9869,
+      "step": 3080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.622298469598292e-06,
+      "loss": 0.7866,
+      "step": 3081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6214711841600615e-06,
+      "loss": 0.875,
+      "step": 3082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.620643724494869e-06,
+      "loss": 0.8939,
+      "step": 3083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6198160906916235e-06,
+      "loss": 1.1856,
+      "step": 3084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6189882828392544e-06,
+      "loss": 0.9544,
+      "step": 3085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.61816030102671e-06,
+      "loss": 0.7741,
+      "step": 3086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6173321453429576e-06,
+      "loss": 0.9803,
+      "step": 3087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.61650381587698e-06,
+      "loss": 1.0927,
+      "step": 3088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.615675312717784e-06,
+      "loss": 0.7975,
+      "step": 3089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.614846635954391e-06,
+      "loss": 1.1274,
+      "step": 3090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.614017785675842e-06,
+      "loss": 0.8602,
+      "step": 3091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6131887619711965e-06,
+      "loss": 0.9268,
+      "step": 3092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.612359564929534e-06,
+      "loss": 0.9117,
+      "step": 3093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.611530194639952e-06,
+      "loss": 0.8501,
+      "step": 3094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.610700651191564e-06,
+      "loss": 0.8825,
+      "step": 3095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.609870934673507e-06,
+      "loss": 0.9283,
+      "step": 3096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.609041045174933e-06,
+      "loss": 0.7831,
+      "step": 3097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.608210982785014e-06,
+      "loss": 1.0002,
+      "step": 3098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.607380747592938e-06,
+      "loss": 0.898,
+      "step": 3099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.606550339687917e-06,
+      "loss": 0.9183,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.605719759159176e-06,
+      "loss": 0.8782,
+      "step": 3101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.604889006095962e-06,
+      "loss": 1.0765,
+      "step": 3102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.604058080587538e-06,
+      "loss": 0.7357,
+      "step": 3103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.603226982723188e-06,
+      "loss": 1.0128,
+      "step": 3104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.602395712592213e-06,
+      "loss": 0.9319,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.601564270283933e-06,
+      "loss": 1.1112,
+      "step": 3106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.600732655887687e-06,
+      "loss": 0.8437,
+      "step": 3107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.599900869492831e-06,
+      "loss": 0.9436,
+      "step": 3108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.59906891118874e-06,
+      "loss": 0.7562,
+      "step": 3109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.59823678106481e-06,
+      "loss": 0.9718,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.597404479210451e-06,
+      "loss": 1.0037,
+      "step": 3111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.596572005715095e-06,
+      "loss": 0.8704,
+      "step": 3112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.595739360668191e-06,
+      "loss": 0.8347,
+      "step": 3113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5949065441592065e-06,
+      "loss": 0.9201,
+      "step": 3114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.594073556277627e-06,
+      "loss": 0.7518,
+      "step": 3115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.59324039711296e-06,
+      "loss": 1.0214,
+      "step": 3116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.592407066754723e-06,
+      "loss": 0.898,
+      "step": 3117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.591573565292462e-06,
+      "loss": 0.8909,
+      "step": 3118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.590739892815734e-06,
+      "loss": 1.0544,
+      "step": 3119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.589906049414119e-06,
+      "loss": 0.9706,
+      "step": 3120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.589072035177211e-06,
+      "loss": 1.0674,
+      "step": 3121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.588237850194627e-06,
+      "loss": 0.9353,
+      "step": 3122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5874034945560006e-06,
+      "loss": 0.9377,
+      "step": 3123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.58656896835098e-06,
+      "loss": 0.9675,
+      "step": 3124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.585734271669238e-06,
+      "loss": 1.1243,
+      "step": 3125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5848994046004615e-06,
+      "loss": 1.1148,
+      "step": 3126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.584064367234358e-06,
+      "loss": 1.0847,
+      "step": 3127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.58322915966065e-06,
+      "loss": 0.8938,
+      "step": 3128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5823937819690825e-06,
+      "loss": 0.8224,
+      "step": 3129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.581558234249416e-06,
+      "loss": 0.8603,
+      "step": 3130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.580722516591429e-06,
+      "loss": 1.0142,
+      "step": 3131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.579886629084921e-06,
+      "loss": 0.8956,
+      "step": 3132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.579050571819707e-06,
+      "loss": 1.2492,
+      "step": 3133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.578214344885623e-06,
+      "loss": 0.9579,
+      "step": 3134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.57737794837252e-06,
+      "loss": 1.1407,
+      "step": 3135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.576541382370268e-06,
+      "loss": 0.8865,
+      "step": 3136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5757046469687586e-06,
+      "loss": 0.8387,
+      "step": 3137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.574867742257895e-06,
+      "loss": 0.9164,
+      "step": 3138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.5740306683276065e-06,
+      "loss": 0.9939,
+      "step": 3139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.573193425267835e-06,
+      "loss": 0.7047,
+      "step": 3140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.572356013168542e-06,
+      "loss": 0.9988,
+      "step": 3141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.571518432119707e-06,
+      "loss": 1.0334,
+      "step": 3142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.57068068221133e-06,
+      "loss": 1.0043,
+      "step": 3143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.569842763533424e-06,
+      "loss": 1.0353,
+      "step": 3144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.569004676176026e-06,
+      "loss": 0.8973,
+      "step": 3145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.568166420229188e-06,
+      "loss": 0.7469,
+      "step": 3146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.567327995782979e-06,
+      "loss": 0.8095,
+      "step": 3147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.56648940292749e-06,
+      "loss": 0.8342,
+      "step": 3148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.565650641752825e-06,
+      "loss": 1.0895,
+      "step": 3149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5648117123491114e-06,
+      "loss": 0.8801,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.56397261480649e-06,
+      "loss": 0.7843,
+      "step": 3151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.563133349215124e-06,
+      "loss": 0.8371,
+      "step": 3152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.56229391566519e-06,
+      "loss": 0.8594,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.561454314246887e-06,
+      "loss": 0.952,
+      "step": 3154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.560614545050428e-06,
+      "loss": 0.926,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.559774608166049e-06,
+      "loss": 1.0369,
+      "step": 3156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.558934503683999e-06,
+      "loss": 0.9788,
+      "step": 3157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.558094231694548e-06,
+      "loss": 0.8388,
+      "step": 3158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.557253792287983e-06,
+      "loss": 0.9064,
+      "step": 3159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.556413185554609e-06,
+      "loss": 0.938,
+      "step": 3160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.55557241158475e-06,
+      "loss": 0.8971,
+      "step": 3161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.554731470468746e-06,
+      "loss": 0.7372,
+      "step": 3162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5538903622969555e-06,
+      "loss": 1.1798,
+      "step": 3163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.553049087159757e-06,
+      "loss": 0.7614,
+      "step": 3164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.552207645147545e-06,
+      "loss": 1.1323,
+      "step": 3165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.551366036350733e-06,
+      "loss": 0.8946,
+      "step": 3166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.55052426085975e-06,
+      "loss": 0.9426,
+      "step": 3167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.549682318765046e-06,
+      "loss": 1.2226,
+      "step": 3168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.548840210157088e-06,
+      "loss": 1.2983,
+      "step": 3169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.54799793512636e-06,
+      "loss": 1.0525,
+      "step": 3170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5471554937633635e-06,
+      "loss": 0.9032,
+      "step": 3171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.54631288615862e-06,
+      "loss": 1.0312,
+      "step": 3172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.545470112402667e-06,
+      "loss": 0.8113,
+      "step": 3173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.544627172586062e-06,
+      "loss": 0.9054,
+      "step": 3174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5437840667993755e-06,
+      "loss": 1.1193,
+      "step": 3175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.542940795133202e-06,
+      "loss": 0.7728,
+      "step": 3176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.54209735767815e-06,
+      "loss": 1.144,
+      "step": 3177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5412537545248485e-06,
+      "loss": 1.0828,
+      "step": 3178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.54040998576394e-06,
+      "loss": 0.7095,
+      "step": 3179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.539566051486088e-06,
+      "loss": 0.9468,
+      "step": 3180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.538721951781974e-06,
+      "loss": 0.7854,
+      "step": 3181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.537877686742296e-06,
+      "loss": 0.8796,
+      "step": 3182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.53703325645777e-06,
+      "loss": 0.9828,
+      "step": 3183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.536188661019131e-06,
+      "loss": 0.8951,
+      "step": 3184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.535343900517129e-06,
+      "loss": 0.997,
+      "step": 3185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.534498975042535e-06,
+      "loss": 0.8277,
+      "step": 3186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.533653884686136e-06,
+      "loss": 0.9716,
+      "step": 3187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.5328086295387366e-06,
+      "loss": 0.924,
+      "step": 3188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.531963209691158e-06,
+      "loss": 0.9054,
+      "step": 3189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.531117625234244e-06,
+      "loss": 0.8828,
+      "step": 3190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.530271876258849e-06,
+      "loss": 0.8708,
+      "step": 3191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.52942596285585e-06,
+      "loss": 0.9937,
+      "step": 3192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.528579885116141e-06,
+      "loss": 1.0573,
+      "step": 3193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.527733643130633e-06,
+      "loss": 0.8786,
+      "step": 3194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.526887236990252e-06,
+      "loss": 0.9917,
+      "step": 3195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.526040666785948e-06,
+      "loss": 0.8868,
+      "step": 3196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.525193932608684e-06,
+      "loss": 0.8511,
+      "step": 3197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.524347034549438e-06,
+      "loss": 0.8072,
+      "step": 3198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.523499972699214e-06,
+      "loss": 0.9129,
+      "step": 3199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.522652747149026e-06,
+      "loss": 0.8953,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.521805357989909e-06,
+      "loss": 1.1848,
+      "step": 3201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.520957805312916e-06,
+      "loss": 0.7492,
+      "step": 3202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5201100892091144e-06,
+      "loss": 1.0002,
+      "step": 3203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.519262209769591e-06,
+      "loss": 0.7325,
+      "step": 3204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.518414167085454e-06,
+      "loss": 0.8325,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.517565961247821e-06,
+      "loss": 0.9757,
+      "step": 3206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.516717592347835e-06,
+      "loss": 0.9462,
+      "step": 3207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.515869060476651e-06,
+      "loss": 1.0734,
+      "step": 3208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.515020365725446e-06,
+      "loss": 1.0478,
+      "step": 3209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.51417150818541e-06,
+      "loss": 1.0423,
+      "step": 3210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.513322487947752e-06,
+      "loss": 1.0398,
+      "step": 3211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.512473305103703e-06,
+      "loss": 0.9736,
+      "step": 3212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.511623959744504e-06,
+      "loss": 0.8756,
+      "step": 3213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.510774451961419e-06,
+      "loss": 0.7763,
+      "step": 3214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.5099247818457275e-06,
+      "loss": 0.9699,
+      "step": 3215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.509074949488724e-06,
+      "loss": 0.8681,
+      "step": 3216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.508224954981727e-06,
+      "loss": 1.0273,
+      "step": 3217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.507374798416065e-06,
+      "loss": 0.989,
+      "step": 3218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.506524479883089e-06,
+      "loss": 1.0806,
+      "step": 3219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.505673999474164e-06,
+      "loss": 0.835,
+      "step": 3220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.504823357280676e-06,
+      "loss": 0.966,
+      "step": 3221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.503972553394026e-06,
+      "loss": 0.9249,
+      "step": 3222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.503121587905632e-06,
+      "loss": 0.9214,
+      "step": 3223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.50227046090693e-06,
+      "loss": 0.8363,
+      "step": 3224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.501419172489376e-06,
+      "loss": 0.9762,
+      "step": 3225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.500567722744438e-06,
+      "loss": 0.9471,
+      "step": 3226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.499716111763604e-06,
+      "loss": 0.9023,
+      "step": 3227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.498864339638382e-06,
+      "loss": 0.7317,
+      "step": 3228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.498012406460294e-06,
+      "loss": 1.0489,
+      "step": 3229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.49716031232088e-06,
+      "loss": 0.9339,
+      "step": 3230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.496308057311697e-06,
+      "loss": 0.8915,
+      "step": 3231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.49545564152432e-06,
+      "loss": 0.8793,
+      "step": 3232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4946030650503406e-06,
+      "loss": 0.9988,
+      "step": 3233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.493750327981369e-06,
+      "loss": 0.7372,
+      "step": 3234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4928974304090305e-06,
+      "loss": 1.1039,
+      "step": 3235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.492044372424971e-06,
+      "loss": 1.1117,
+      "step": 3236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.491191154120849e-06,
+      "loss": 1.0306,
+      "step": 3237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.490337775588344e-06,
+      "loss": 0.7722,
+      "step": 3238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.489484236919152e-06,
+      "loss": 0.811,
+      "step": 3239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.488630538204985e-06,
+      "loss": 1.3126,
+      "step": 3240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.487776679537572e-06,
+      "loss": 0.8606,
+      "step": 3241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.486922661008662e-06,
+      "loss": 0.9031,
+      "step": 3242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.486068482710017e-06,
+      "loss": 0.9875,
+      "step": 3243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.485214144733421e-06,
+      "loss": 1.0799,
+      "step": 3244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.48435964717067e-06,
+      "loss": 0.9219,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.483504990113581e-06,
+      "loss": 0.7849,
+      "step": 3246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.482650173653987e-06,
+      "loss": 0.9007,
+      "step": 3247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.481795197883738e-06,
+      "loss": 0.8961,
+      "step": 3248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.480940062894699e-06,
+      "loss": 1.0088,
+      "step": 3249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.480084768778758e-06,
+      "loss": 1.1692,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.479229315627813e-06,
+      "loss": 1.1109,
+      "step": 3251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.478373703533784e-06,
+      "loss": 0.778,
+      "step": 3252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.477517932588606e-06,
+      "loss": 0.8897,
+      "step": 3253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.476662002884232e-06,
+      "loss": 1.0023,
+      "step": 3254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.475805914512631e-06,
+      "loss": 1.1579,
+      "step": 3255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.474949667565789e-06,
+      "loss": 0.966,
+      "step": 3256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.474093262135711e-06,
+      "loss": 0.8485,
+      "step": 3257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4732366983144175e-06,
+      "loss": 1.0312,
+      "step": 3258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.472379976193945e-06,
+      "loss": 0.9535,
+      "step": 3259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.47152309586635e-06,
+      "loss": 1.0468,
+      "step": 3260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.470666057423703e-06,
+      "loss": 0.9334,
+      "step": 3261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.469808860958094e-06,
+      "loss": 0.9394,
+      "step": 3262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.468951506561626e-06,
+      "loss": 0.8189,
+      "step": 3263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.468093994326424e-06,
+      "loss": 1.0347,
+      "step": 3264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.467236324344628e-06,
+      "loss": 0.8667,
+      "step": 3265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.466378496708393e-06,
+      "loss": 1.1199,
+      "step": 3266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.465520511509893e-06,
+      "loss": 1.2093,
+      "step": 3267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.464662368841319e-06,
+      "loss": 0.8026,
+      "step": 3268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.463804068794878e-06,
+      "loss": 1.0439,
+      "step": 3269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.462945611462794e-06,
+      "loss": 0.8969,
+      "step": 3270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.462086996937309e-06,
+      "loss": 1.3904,
+      "step": 3271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.46122822531068e-06,
+      "loss": 1.0246,
+      "step": 3272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.460369296675182e-06,
+      "loss": 0.7448,
+      "step": 3273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.459510211123108e-06,
+      "loss": 0.9264,
+      "step": 3274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.458650968746766e-06,
+      "loss": 0.8438,
+      "step": 3275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.457791569638481e-06,
+      "loss": 0.9985,
+      "step": 3276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.456932013890596e-06,
+      "loss": 0.8425,
+      "step": 3277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.456072301595471e-06,
+      "loss": 0.9602,
+      "step": 3278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.45521243284548e-06,
+      "loss": 0.791,
+      "step": 3279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.454352407733017e-06,
+      "loss": 1.0895,
+      "step": 3280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.453492226350491e-06,
+      "loss": 0.7769,
+      "step": 3281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.452631888790331e-06,
+      "loss": 1.0126,
+      "step": 3282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4517713951449764e-06,
+      "loss": 0.8093,
+      "step": 3283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4509107455068894e-06,
+      "loss": 1.1267,
+      "step": 3284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.450049939968547e-06,
+      "loss": 0.8313,
+      "step": 3285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4491889786224415e-06,
+      "loss": 0.9571,
+      "step": 3286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4483278615610835e-06,
+      "loss": 0.9878,
+      "step": 3287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.447466588877001e-06,
+      "loss": 0.9375,
+      "step": 3288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4466051606627355e-06,
+      "loss": 1.0182,
+      "step": 3289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.445743577010849e-06,
+      "loss": 0.749,
+      "step": 3290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.444881838013918e-06,
+      "loss": 1.0061,
+      "step": 3291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.444019943764538e-06,
+      "loss": 0.8873,
+      "step": 3292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.443157894355318e-06,
+      "loss": 0.8782,
+      "step": 3293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.442295689878884e-06,
+      "loss": 1.0894,
+      "step": 3294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.441433330427881e-06,
+      "loss": 0.83,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4405708160949705e-06,
+      "loss": 1.0744,
+      "step": 3296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.439708146972828e-06,
+      "loss": 1.0598,
+      "step": 3297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.438845323154149e-06,
+      "loss": 0.8161,
+      "step": 3298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.437982344731642e-06,
+      "loss": 0.9913,
+      "step": 3299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4371192117980335e-06,
+      "loss": 1.111,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4362559244460696e-06,
+      "loss": 1.0002,
+      "step": 3301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.435392482768508e-06,
+      "loss": 0.8097,
+      "step": 3302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.434528886858128e-06,
+      "loss": 1.0539,
+      "step": 3303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.433665136807721e-06,
+      "loss": 0.9239,
+      "step": 3304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.432801232710097e-06,
+      "loss": 1.0484,
+      "step": 3305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.431937174658083e-06,
+      "loss": 0.8123,
+      "step": 3306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.431072962744521e-06,
+      "loss": 1.0213,
+      "step": 3307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.430208597062272e-06,
+      "loss": 0.8136,
+      "step": 3308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.429344077704212e-06,
+      "loss": 0.8728,
+      "step": 3309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.428479404763231e-06,
+      "loss": 1.0472,
+      "step": 3310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.427614578332242e-06,
+      "loss": 0.8552,
+      "step": 3311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.426749598504167e-06,
+      "loss": 0.8032,
+      "step": 3312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.42588446537195e-06,
+      "loss": 0.868,
+      "step": 3313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.425019179028549e-06,
+      "loss": 1.0481,
+      "step": 3314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4241537395669386e-06,
+      "loss": 0.997,
+      "step": 3315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.423288147080111e-06,
+      "loss": 1.0476,
+      "step": 3316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.422422401661072e-06,
+      "loss": 0.7697,
+      "step": 3317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4215565034028485e-06,
+      "loss": 0.8413,
+      "step": 3318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.420690452398481e-06,
+      "loss": 0.9296,
+      "step": 3319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4198242487410245e-06,
+      "loss": 0.9671,
+      "step": 3320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.418957892523555e-06,
+      "loss": 1.1275,
+      "step": 3321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.41809138383916e-06,
+      "loss": 1.0889,
+      "step": 3322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.417224722780948e-06,
+      "loss": 0.86,
+      "step": 3323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.41635790944204e-06,
+      "loss": 0.9845,
+      "step": 3324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.415490943915576e-06,
+      "loss": 0.9264,
+      "step": 3325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.414623826294711e-06,
+      "loss": 1.1099,
+      "step": 3326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4137565566726165e-06,
+      "loss": 1.023,
+      "step": 3327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.412889135142481e-06,
+      "loss": 1.0623,
+      "step": 3328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4120215617975095e-06,
+      "loss": 0.9416,
+      "step": 3329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.411153836730921e-06,
+      "loss": 0.9628,
+      "step": 3330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.410285960035953e-06,
+      "loss": 0.9549,
+      "step": 3331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.40941793180586e-06,
+      "loss": 0.8794,
+      "step": 3332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.408549752133912e-06,
+      "loss": 0.8903,
+      "step": 3333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.407681421113391e-06,
+      "loss": 1.0391,
+      "step": 3334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.406812938837604e-06,
+      "loss": 0.8691,
+      "step": 3335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.405944305399866e-06,
+      "loss": 0.9104,
+      "step": 3336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4050755208935136e-06,
+      "loss": 1.0622,
+      "step": 3337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.404206585411896e-06,
+      "loss": 0.8936,
+      "step": 3338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.403337499048381e-06,
+      "loss": 0.8666,
+      "step": 3339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.402468261896352e-06,
+      "loss": 0.9042,
+      "step": 3340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.401598874049209e-06,
+      "loss": 1.068,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400729335600366e-06,
+      "loss": 0.9277,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399859646643256e-06,
+      "loss": 0.8869,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398989807271327e-06,
+      "loss": 0.6673,
+      "step": 3344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398119817578043e-06,
+      "loss": 0.7153,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397249677656884e-06,
+      "loss": 1.0675,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3963793876013475e-06,
+      "loss": 0.973,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395508947504945e-06,
+      "loss": 0.9984,
+      "step": 3348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394638357461207e-06,
+      "loss": 1.1425,
+      "step": 3349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393767617563676e-06,
+      "loss": 0.9081,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392896727905913e-06,
+      "loss": 1.1408,
+      "step": 3351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392025688581497e-06,
+      "loss": 0.867,
+      "step": 3352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39115449968402e-06,
+      "loss": 1.1025,
+      "step": 3353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390283161307091e-06,
+      "loss": 0.9981,
+      "step": 3354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3894116735443366e-06,
+      "loss": 1.0786,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388540036489396e-06,
+      "loss": 0.8672,
+      "step": 3356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387668250235928e-06,
+      "loss": 1.0751,
+      "step": 3357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386796314877605e-06,
+      "loss": 0.9135,
+      "step": 3358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385924230508117e-06,
+      "loss": 0.8685,
+      "step": 3359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38505199722117e-06,
+      "loss": 0.9796,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384179615110484e-06,
+      "loss": 0.9074,
+      "step": 3361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383307084269797e-06,
+      "loss": 1.0492,
+      "step": 3362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382434404792863e-06,
+      "loss": 0.9291,
+      "step": 3363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381561576773451e-06,
+      "loss": 0.8773,
+      "step": 3364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380688600305346e-06,
+      "loss": 0.9222,
+      "step": 3365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379815475482349e-06,
+      "loss": 0.8438,
+      "step": 3366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378942202398278e-06,
+      "loss": 0.902,
+      "step": 3367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378068781146966e-06,
+      "loss": 0.9851,
+      "step": 3368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377195211822261e-06,
+      "loss": 0.8909,
+      "step": 3369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37632149451803e-06,
+      "loss": 0.8314,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375447629328151e-06,
+      "loss": 0.8936,
+      "step": 3371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374573616346524e-06,
+      "loss": 1.0566,
+      "step": 3372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373699455667058e-06,
+      "loss": 0.9836,
+      "step": 3373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372825147383686e-06,
+      "loss": 0.9083,
+      "step": 3374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371950691590348e-06,
+      "loss": 0.9043,
+      "step": 3375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371076088381007e-06,
+      "loss": 0.8193,
+      "step": 3376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370201337849638e-06,
+      "loss": 1.0721,
+      "step": 3377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369326440090234e-06,
+      "loss": 1.0828,
+      "step": 3378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368451395196799e-06,
+      "loss": 1.0586,
+      "step": 3379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3675762032633625e-06,
+      "loss": 1.0943,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366700864383958e-06,
+      "loss": 0.918,
+      "step": 3381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365825378652645e-06,
+      "loss": 0.85,
+      "step": 3382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364949746163491e-06,
+      "loss": 0.9193,
+      "step": 3383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364073967010585e-06,
+      "loss": 0.8768,
+      "step": 3384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363198041288028e-06,
+      "loss": 0.8088,
+      "step": 3385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36232196908994e-06,
+      "loss": 1.0142,
+      "step": 3386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361445750510452e-06,
+      "loss": 1.0501,
+      "step": 3387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360569385643716e-06,
+      "loss": 0.9089,
+      "step": 3388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359692874583897e-06,
+      "loss": 1.133,
+      "step": 3389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.358816217425176e-06,
+      "loss": 0.8858,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.357939414261748e-06,
+      "loss": 1.0208,
+      "step": 3391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.357062465187828e-06,
+      "loss": 0.9572,
+      "step": 3392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3561853702976414e-06,
+      "loss": 0.8947,
+      "step": 3393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.355308129685435e-06,
+      "loss": 0.9601,
+      "step": 3394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.354430743445466e-06,
+      "loss": 0.7621,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.353553211672011e-06,
+      "loss": 0.747,
+      "step": 3396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.35267553445936e-06,
+      "loss": 0.808,
+      "step": 3397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3517977119018185e-06,
+      "loss": 0.8968,
+      "step": 3398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3509197440937096e-06,
+      "loss": 0.8673,
+      "step": 3399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.350041631129371e-06,
+      "loss": 0.8697,
+      "step": 3400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.349163373103155e-06,
+      "loss": 1.1449,
+      "step": 3401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.348284970109432e-06,
+      "loss": 0.8641,
+      "step": 3402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.347406422242584e-06,
+      "loss": 0.8337,
+      "step": 3403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3465277295970125e-06,
+      "loss": 0.8524,
+      "step": 3404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3456488922671324e-06,
+      "loss": 1.139,
+      "step": 3405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.344769910347375e-06,
+      "loss": 0.9511,
+      "step": 3406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.343890783932186e-06,
+      "loss": 1.0521,
+      "step": 3407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3430115131160294e-06,
+      "loss": 1.0845,
+      "step": 3408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.34213209799338e-06,
+      "loss": 1.194,
+      "step": 3409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.341252538658734e-06,
+      "loss": 1.0167,
+      "step": 3410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.340372835206598e-06,
+      "loss": 0.8838,
+      "step": 3411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.339492987731496e-06,
+      "loss": 0.7539,
+      "step": 3412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.338612996327967e-06,
+      "loss": 1.0874,
+      "step": 3413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.33773286109057e-06,
+      "loss": 0.8938,
+      "step": 3414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.33685258211387e-06,
+      "loss": 1.0346,
+      "step": 3415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3359721594924565e-06,
+      "loss": 1.1296,
+      "step": 3416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.33509159332093e-06,
+      "loss": 1.029,
+      "step": 3417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.334210883693907e-06,
+      "loss": 0.9882,
+      "step": 3418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.333330030706019e-06,
+      "loss": 0.9229,
+      "step": 3419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.332449034451915e-06,
+      "loss": 1.1449,
+      "step": 3420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.331567895026257e-06,
+      "loss": 1.0202,
+      "step": 3421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.330686612523724e-06,
+      "loss": 1.0483,
+      "step": 3422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.329805187039009e-06,
+      "loss": 1.0702,
+      "step": 3423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.328923618666822e-06,
+      "loss": 0.891,
+      "step": 3424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.328041907501886e-06,
+      "loss": 0.8772,
+      "step": 3425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.327160053638944e-06,
+      "loss": 0.9633,
+      "step": 3426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.326278057172746e-06,
+      "loss": 1.0303,
+      "step": 3427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3253959181980676e-06,
+      "loss": 0.9097,
+      "step": 3428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.324513636809691e-06,
+      "loss": 0.9014,
+      "step": 3429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.32363121310242e-06,
+      "loss": 0.992,
+      "step": 3430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.322748647171069e-06,
+      "loss": 1.0107,
+      "step": 3431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.32186593911047e-06,
+      "loss": 0.9996,
+      "step": 3432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.320983089015471e-06,
+      "loss": 0.9945,
+      "step": 3433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.320100096980932e-06,
+      "loss": 0.8982,
+      "step": 3434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.319216963101734e-06,
+      "loss": 0.9617,
+      "step": 3435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.318333687472765e-06,
+      "loss": 0.848,
+      "step": 3436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.317450270188937e-06,
+      "loss": 0.8862,
+      "step": 3437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.31656671134517e-06,
+      "loss": 0.8906,
+      "step": 3438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315683011036405e-06,
+      "loss": 0.9274,
+      "step": 3439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.314799169357594e-06,
+      "loss": 0.8619,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.313915186403706e-06,
+      "loss": 0.8059,
+      "step": 3441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.313031062269724e-06,
+      "loss": 0.9113,
+      "step": 3442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.31214679705065e-06,
+      "loss": 1.0043,
+      "step": 3443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.311262390841496e-06,
+      "loss": 0.9531,
+      "step": 3444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.31037784373729e-06,
+      "loss": 1.35,
+      "step": 3445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.30949315583308e-06,
+      "loss": 0.9556,
+      "step": 3446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.308608327223923e-06,
+      "loss": 1.0562,
+      "step": 3447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.307723358004895e-06,
+      "loss": 0.8761,
+      "step": 3448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.306838248271084e-06,
+      "loss": 0.7873,
+      "step": 3449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.305952998117598e-06,
+      "loss": 1.1943,
+      "step": 3450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.305067607639556e-06,
+      "loss": 0.9983,
+      "step": 3451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.304182076932091e-06,
+      "loss": 0.9429,
+      "step": 3452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.303296406090356e-06,
+      "loss": 0.9906,
+      "step": 3453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.302410595209515e-06,
+      "loss": 0.8486,
+      "step": 3454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.301524644384749e-06,
+      "loss": 0.8666,
+      "step": 3455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.300638553711251e-06,
+      "loss": 0.9506,
+      "step": 3456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.299752323284235e-06,
+      "loss": 1.065,
+      "step": 3457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.298865953198924e-06,
+      "loss": 0.8908,
+      "step": 3458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29797944355056e-06,
+      "loss": 0.8107,
+      "step": 3459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2970927944343964e-06,
+      "loss": 1.1452,
+      "step": 3460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.296206005945705e-06,
+      "loss": 1.0273,
+      "step": 3461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.295319078179771e-06,
+      "loss": 1.123,
+      "step": 3462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.294432011231893e-06,
+      "loss": 1.1995,
+      "step": 3463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2935448051973894e-06,
+      "loss": 0.907,
+      "step": 3464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2926574601715875e-06,
+      "loss": 1.0855,
+      "step": 3465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.291769976249835e-06,
+      "loss": 0.7324,
+      "step": 3466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29088235352749e-06,
+      "loss": 0.7538,
+      "step": 3467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.289994592099928e-06,
+      "loss": 0.9504,
+      "step": 3468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.289106692062539e-06,
+      "loss": 1.101,
+      "step": 3469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.288218653510729e-06,
+      "loss": 0.8395,
+      "step": 3470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.287330476539916e-06,
+      "loss": 0.7937,
+      "step": 3471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2864421612455355e-06,
+      "loss": 1.0348,
+      "step": 3472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285553707723036e-06,
+      "loss": 0.8542,
+      "step": 3473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.284665116067883e-06,
+      "loss": 1.0183,
+      "step": 3474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.283776386375554e-06,
+      "loss": 0.7283,
+      "step": 3475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.282887518741544e-06,
+      "loss": 0.9496,
+      "step": 3476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.281998513261363e-06,
+      "loss": 0.9132,
+      "step": 3477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2811093700305324e-06,
+      "loss": 0.7566,
+      "step": 3478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.280220089144591e-06,
+      "loss": 0.8266,
+      "step": 3479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.279330670699093e-06,
+      "loss": 0.8204,
+      "step": 3480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.278441114789604e-06,
+      "loss": 0.9413,
+      "step": 3481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2775514215117114e-06,
+      "loss": 0.9314,
+      "step": 3482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.276661590961007e-06,
+      "loss": 0.9408,
+      "step": 3483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2757716232331076e-06,
+      "loss": 0.8982,
+      "step": 3484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.274881518423637e-06,
+      "loss": 1.1474,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2739912766282394e-06,
+      "loss": 0.8295,
+      "step": 3486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.27310089794257e-06,
+      "loss": 0.8983,
+      "step": 3487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.272210382462299e-06,
+      "loss": 0.9493,
+      "step": 3488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2713197302831134e-06,
+      "loss": 0.9503,
+      "step": 3489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.270428941500714e-06,
+      "loss": 0.7448,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.269538016210815e-06,
+      "loss": 0.9328,
+      "step": 3491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.268646954509148e-06,
+      "loss": 0.8227,
+      "step": 3492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2677557564914545e-06,
+      "loss": 1.1788,
+      "step": 3493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.266864422253496e-06,
+      "loss": 0.8258,
+      "step": 3494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.265972951891046e-06,
+      "loss": 1.0879,
+      "step": 3495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.265081345499893e-06,
+      "loss": 0.8698,
+      "step": 3496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.264189603175839e-06,
+      "loss": 0.9619,
+      "step": 3497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.263297725014702e-06,
+      "loss": 0.7746,
+      "step": 3498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.262405711112316e-06,
+      "loss": 1.0532,
+      "step": 3499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.261513561564526e-06,
+      "loss": 0.9368,
+      "step": 3500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.260621276467193e-06,
+      "loss": 0.7647,
+      "step": 3501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.259728855916194e-06,
+      "loss": 0.9242,
+      "step": 3502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.25883630000742e-06,
+      "loss": 0.9647,
+      "step": 3503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2579436088367755e-06,
+      "loss": 0.8475,
+      "step": 3504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.257050782500179e-06,
+      "loss": 0.758,
+      "step": 3505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.256157821093566e-06,
+      "loss": 0.7758,
+      "step": 3506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.255264724712884e-06,
+      "loss": 0.7878,
+      "step": 3507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.254371493454099e-06,
+      "loss": 0.9981,
+      "step": 3508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.253478127413185e-06,
+      "loss": 0.9878,
+      "step": 3509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.252584626686136e-06,
+      "loss": 1.1134,
+      "step": 3510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.251690991368958e-06,
+      "loss": 0.8319,
+      "step": 3511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.250797221557673e-06,
+      "loss": 0.8534,
+      "step": 3512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.249903317348315e-06,
+      "loss": 0.9037,
+      "step": 3513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.249009278836935e-06,
+      "loss": 0.8691,
+      "step": 3514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.248115106119597e-06,
+      "loss": 0.9276,
+      "step": 3515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.24722079929238e-06,
+      "loss": 1.0728,
+      "step": 3516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.246326358451377e-06,
+      "loss": 1.0653,
+      "step": 3517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.245431783692697e-06,
+      "loss": 0.9523,
+      "step": 3518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.244537075112459e-06,
+      "loss": 0.7537,
+      "step": 3519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2436422328068025e-06,
+      "loss": 0.8829,
+      "step": 3520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.242747256871875e-06,
+      "loss": 1.1453,
+      "step": 3521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.241852147403845e-06,
+      "loss": 0.8894,
+      "step": 3522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.24095690449889e-06,
+      "loss": 0.8774,
+      "step": 3523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.240061528253204e-06,
+      "loss": 1.1253,
+      "step": 3524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.239166018762995e-06,
+      "loss": 1.0195,
+      "step": 3525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2382703761244866e-06,
+      "loss": 1.0923,
+      "step": 3526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2373746004339134e-06,
+      "loss": 1.292,
+      "step": 3527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.236478691787529e-06,
+      "loss": 0.9297,
+      "step": 3528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.235582650281596e-06,
+      "loss": 0.9208,
+      "step": 3529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.234686476012396e-06,
+      "loss": 0.8704,
+      "step": 3530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.233790169076222e-06,
+      "loss": 0.7931,
+      "step": 3531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.232893729569383e-06,
+      "loss": 0.9236,
+      "step": 3532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.231997157588201e-06,
+      "loss": 1.1514,
+      "step": 3533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.231100453229012e-06,
+      "loss": 0.9594,
+      "step": 3534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.230203616588167e-06,
+      "loss": 0.8509,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.229306647762031e-06,
+      "loss": 0.8754,
+      "step": 3536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.228409546846983e-06,
+      "loss": 1.0972,
+      "step": 3537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.227512313939418e-06,
+      "loss": 1.0483,
+      "step": 3538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2266149491357415e-06,
+      "loss": 0.9213,
+      "step": 3539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2257174525323766e-06,
+      "loss": 0.894,
+      "step": 3540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.224819824225759e-06,
+      "loss": 1.017,
+      "step": 3541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.223922064312339e-06,
+      "loss": 0.9297,
+      "step": 3542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.22302417288858e-06,
+      "loss": 0.799,
+      "step": 3543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.222126150050962e-06,
+      "loss": 1.1207,
+      "step": 3544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.221227995895974e-06,
+      "loss": 0.7632,
+      "step": 3545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.220329710520127e-06,
+      "loss": 0.853,
+      "step": 3546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.219431294019938e-06,
+      "loss": 0.847,
+      "step": 3547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.218532746491946e-06,
+      "loss": 0.9976,
+      "step": 3548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.217634068032696e-06,
+      "loss": 1.0239,
+      "step": 3549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.216735258738751e-06,
+      "loss": 0.8153,
+      "step": 3550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.215836318706691e-06,
+      "loss": 0.9319,
+      "step": 3551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.214937248033105e-06,
+      "loss": 0.916,
+      "step": 3552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2140380468145975e-06,
+      "loss": 0.9854,
+      "step": 3553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.21313871514779e-06,
+      "loss": 1.111,
+      "step": 3554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.212239253129313e-06,
+      "loss": 0.8464,
+      "step": 3555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.211339660855815e-06,
+      "loss": 0.8019,
+      "step": 3556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.210439938423956e-06,
+      "loss": 1.1104,
+      "step": 3557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.209540085930414e-06,
+      "loss": 1.0256,
+      "step": 3558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2086401034718746e-06,
+      "loss": 0.9971,
+      "step": 3559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.207739991145043e-06,
+      "loss": 1.137,
+      "step": 3560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.206839749046635e-06,
+      "loss": 0.9002,
+      "step": 3561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.205939377273382e-06,
+      "loss": 0.798,
+      "step": 3562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.205038875922029e-06,
+      "loss": 0.8058,
+      "step": 3563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2041382450893356e-06,
+      "loss": 1.1301,
+      "step": 3564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.203237484872073e-06,
+      "loss": 0.7505,
+      "step": 3565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.202336595367029e-06,
+      "loss": 0.7886,
+      "step": 3566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.201435576671004e-06,
+      "loss": 1.0576,
+      "step": 3567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.200534428880811e-06,
+      "loss": 0.8532,
+      "step": 3568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.199633152093279e-06,
+      "loss": 0.9086,
+      "step": 3569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.198731746405251e-06,
+      "loss": 1.0006,
+      "step": 3570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.197830211913583e-06,
+      "loss": 1.075,
+      "step": 3571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1969285487151425e-06,
+      "loss": 0.8186,
+      "step": 3572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.196026756906817e-06,
+      "loss": 1.0068,
+      "step": 3573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1951248365855e-06,
+      "loss": 0.8313,
+      "step": 3574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.194222787848106e-06,
+      "loss": 1.0249,
+      "step": 3575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1933206107915575e-06,
+      "loss": 0.9919,
+      "step": 3576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.192418305512796e-06,
+      "loss": 1.1974,
+      "step": 3577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1915158721087705e-06,
+      "loss": 0.9332,
+      "step": 3578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.190613310676451e-06,
+      "loss": 0.8314,
+      "step": 3579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.189710621312816e-06,
+      "loss": 0.9587,
+      "step": 3580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.18880780411486e-06,
+      "loss": 0.8235,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.18790485917959e-06,
+      "loss": 1.1722,
+      "step": 3582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.187001786604028e-06,
+      "loss": 1.0222,
+      "step": 3583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.186098586485208e-06,
+      "loss": 0.7984,
+      "step": 3584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.18519525892018e-06,
+      "loss": 1.0661,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.184291804006006e-06,
+      "loss": 1.0798,
+      "step": 3586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.183388221839762e-06,
+      "loss": 1.1587,
+      "step": 3587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.182484512518537e-06,
+      "loss": 0.9398,
+      "step": 3588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.181580676139437e-06,
+      "loss": 0.7604,
+      "step": 3589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.180676712799577e-06,
+      "loss": 0.9776,
+      "step": 3590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.179772622596089e-06,
+      "loss": 0.8762,
+      "step": 3591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.178868405626116e-06,
+      "loss": 0.9815,
+      "step": 3592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.177964061986818e-06,
+      "loss": 0.9589,
+      "step": 3593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1770595917753645e-06,
+      "loss": 0.8085,
+      "step": 3594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.176154995088943e-06,
+      "loss": 1.0293,
+      "step": 3595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.17525027202475e-06,
+      "loss": 0.8238,
+      "step": 3596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.17434542268e-06,
+      "loss": 0.8501,
+      "step": 3597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.173440447151918e-06,
+      "loss": 0.8039,
+      "step": 3598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.172535345537744e-06,
+      "loss": 0.8847,
+      "step": 3599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.17163011793473e-06,
+      "loss": 0.8696,
+      "step": 3600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.170724764440144e-06,
+      "loss": 0.9525,
+      "step": 3601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1698192851512655e-06,
+      "loss": 1.0049,
+      "step": 3602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.168913680165389e-06,
+      "loss": 0.9802,
+      "step": 3603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.168007949579819e-06,
+      "loss": 0.8999,
+      "step": 3604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.16710209349188e-06,
+      "loss": 0.9112,
+      "step": 3605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.166196111998903e-06,
+      "loss": 1.0111,
+      "step": 3606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.165290005198238e-06,
+      "loss": 0.7999,
+      "step": 3607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.164383773187245e-06,
+      "loss": 1.0389,
+      "step": 3608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.163477416063298e-06,
+      "loss": 0.9433,
+      "step": 3609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.162570933923786e-06,
+      "loss": 1.09,
+      "step": 3610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1616643268661105e-06,
+      "loss": 0.83,
+      "step": 3611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1607575949876855e-06,
+      "loss": 0.8015,
+      "step": 3612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.15985073838594e-06,
+      "loss": 1.2488,
+      "step": 3613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.158943757158315e-06,
+      "loss": 0.963,
+      "step": 3614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.158036651402267e-06,
+      "loss": 0.7842,
+      "step": 3615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.157129421215263e-06,
+      "loss": 0.9218,
+      "step": 3616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.156222066694785e-06,
+      "loss": 1.0587,
+      "step": 3617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.155314587938328e-06,
+      "loss": 1.1454,
+      "step": 3618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.154406985043402e-06,
+      "loss": 0.9877,
+      "step": 3619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.153499258107527e-06,
+      "loss": 1.1088,
+      "step": 3620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1525914072282395e-06,
+      "loss": 0.8986,
+      "step": 3621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.151683432503087e-06,
+      "loss": 0.9411,
+      "step": 3622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.150775334029632e-06,
+      "loss": 1.0465,
+      "step": 3623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.14986711190545e-06,
+      "loss": 0.9748,
+      "step": 3624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.148958766228128e-06,
+      "loss": 0.8212,
+      "step": 3625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.14805029709527e-06,
+      "loss": 0.8207,
+      "step": 3626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.147141704604488e-06,
+      "loss": 1.0977,
+      "step": 3627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.146232988853413e-06,
+      "loss": 0.8563,
+      "step": 3628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.145324149939684e-06,
+      "loss": 0.9618,
+      "step": 3629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.144415187960958e-06,
+      "loss": 0.9256,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1435061030149e-06,
+      "loss": 1.0686,
+      "step": 3631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1425968951991936e-06,
+      "loss": 0.9414,
+      "step": 3632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.141687564611532e-06,
+      "loss": 1.1524,
+      "step": 3633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.140778111349625e-06,
+      "loss": 0.8236,
+      "step": 3634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.139868535511189e-06,
+      "loss": 0.8404,
+      "step": 3635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.138958837193961e-06,
+      "loss": 0.9354,
+      "step": 3636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.138049016495688e-06,
+      "loss": 1.031,
+      "step": 3637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.137139073514129e-06,
+      "loss": 1.1,
+      "step": 3638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.136229008347057e-06,
+      "loss": 0.9411,
+      "step": 3639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.13531882109226e-06,
+      "loss": 0.7959,
+      "step": 3640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.134408511847536e-06,
+      "loss": 1.122,
+      "step": 3641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.133498080710699e-06,
+      "loss": 0.9287,
+      "step": 3642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.132587527779574e-06,
+      "loss": 1.1266,
+      "step": 3643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.131676853151999e-06,
+      "loss": 0.9217,
+      "step": 3644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.130766056925827e-06,
+      "loss": 0.881,
+      "step": 3645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.129855139198924e-06,
+      "loss": 1.043,
+      "step": 3646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.128944100069166e-06,
+      "loss": 0.8369,
+      "step": 3647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.128032939634447e-06,
+      "loss": 0.9613,
+      "step": 3648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.127121657992667e-06,
+      "loss": 1.081,
+      "step": 3649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.126210255241745e-06,
+      "loss": 0.9826,
+      "step": 3650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.125298731479613e-06,
+      "loss": 0.9601,
+      "step": 3651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.124387086804212e-06,
+      "loss": 1.0152,
+      "step": 3652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.123475321313498e-06,
+      "loss": 0.8308,
+      "step": 3653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.122563435105442e-06,
+      "loss": 0.9653,
+      "step": 3654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.121651428278024e-06,
+      "loss": 0.8847,
+      "step": 3655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.120739300929241e-06,
+      "loss": 0.8779,
+      "step": 3656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.119827053157099e-06,
+      "loss": 0.7514,
+      "step": 3657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.118914685059621e-06,
+      "loss": 0.9541,
+      "step": 3658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.118002196734839e-06,
+      "loss": 0.8419,
+      "step": 3659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.117089588280802e-06,
+      "loss": 0.9234,
+      "step": 3660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.116176859795568e-06,
+      "loss": 0.9204,
+      "step": 3661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.11526401137721e-06,
+      "loss": 1.042,
+      "step": 3662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.114351043123813e-06,
+      "loss": 1.0253,
+      "step": 3663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.113437955133477e-06,
+      "loss": 1.1657,
+      "step": 3664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.112524747504311e-06,
+      "loss": 0.8177,
+      "step": 3665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.111611420334441e-06,
+      "loss": 0.8944,
+      "step": 3666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.110697973722003e-06,
+      "loss": 1.0394,
+      "step": 3667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.109784407765147e-06,
+      "loss": 1.0076,
+      "step": 3668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.108870722562035e-06,
+      "loss": 1.4124,
+      "step": 3669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.107956918210844e-06,
+      "loss": 1.0489,
+      "step": 3670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.107042994809761e-06,
+      "loss": 0.8722,
+      "step": 3671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1061289524569865e-06,
+      "loss": 0.9246,
+      "step": 3672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.105214791250736e-06,
+      "loss": 1.1704,
+      "step": 3673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.104300511289234e-06,
+      "loss": 1.0774,
+      "step": 3674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.103386112670721e-06,
+      "loss": 0.7417,
+      "step": 3675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1024715954934506e-06,
+      "loss": 0.8253,
+      "step": 3676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.101556959855684e-06,
+      "loss": 0.8101,
+      "step": 3677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.100642205855702e-06,
+      "loss": 0.7658,
+      "step": 3678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.099727333591792e-06,
+      "loss": 0.9541,
+      "step": 3679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.09881234316226e-06,
+      "loss": 0.9408,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.097897234665419e-06,
+      "loss": 0.9176,
+      "step": 3681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.096982008199599e-06,
+      "loss": 0.9748,
+      "step": 3682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0960666638631405e-06,
+      "loss": 0.919,
+      "step": 3683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.095151201754397e-06,
+      "loss": 0.8735,
+      "step": 3684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.094235621971735e-06,
+      "loss": 1.0619,
+      "step": 3685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.093319924613534e-06,
+      "loss": 0.9872,
+      "step": 3686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.092404109778184e-06,
+      "loss": 1.1064,
+      "step": 3687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.091488177564091e-06,
+      "loss": 0.9663,
+      "step": 3688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.090572128069672e-06,
+      "loss": 0.8888,
+      "step": 3689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.089655961393355e-06,
+      "loss": 0.9134,
+      "step": 3690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0887396776335825e-06,
+      "loss": 1.0994,
+      "step": 3691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.08782327688881e-06,
+      "loss": 0.8769,
+      "step": 3692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.086906759257503e-06,
+      "loss": 0.8805,
+      "step": 3693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.085990124838143e-06,
+      "loss": 0.9009,
+      "step": 3694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.085073373729221e-06,
+      "loss": 0.9048,
+      "step": 3695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.084156506029243e-06,
+      "loss": 1.0515,
+      "step": 3696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.083239521836726e-06,
+      "loss": 0.9042,
+      "step": 3697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.082322421250199e-06,
+      "loss": 1.1959,
+      "step": 3698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0814052043682055e-06,
+      "loss": 0.8233,
+      "step": 3699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.080487871289299e-06,
+      "loss": 0.9878,
+      "step": 3700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.079570422112049e-06,
+      "loss": 1.077,
+      "step": 3701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.078652856935034e-06,
+      "loss": 0.776,
+      "step": 3702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.077735175856847e-06,
+      "loss": 0.85,
+      "step": 3703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.076817378976091e-06,
+      "loss": 0.8337,
+      "step": 3704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.075899466391385e-06,
+      "loss": 0.8942,
+      "step": 3705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.074981438201358e-06,
+      "loss": 0.9372,
+      "step": 3706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.074063294504653e-06,
+      "loss": 1.0944,
+      "step": 3707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.073145035399922e-06,
+      "loss": 0.9716,
+      "step": 3708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.072226660985835e-06,
+      "loss": 0.9435,
+      "step": 3709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.071308171361069e-06,
+      "loss": 0.9947,
+      "step": 3710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.070389566624317e-06,
+      "loss": 1.0681,
+      "step": 3711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.069470846874282e-06,
+      "loss": 0.8998,
+      "step": 3712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.068552012209682e-06,
+      "loss": 0.8082,
+      "step": 3713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0676330627292425e-06,
+      "loss": 0.7926,
+      "step": 3714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.066713998531708e-06,
+      "loss": 0.8379,
+      "step": 3715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.065794819715829e-06,
+      "loss": 0.8172,
+      "step": 3716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.064875526380376e-06,
+      "loss": 0.8494,
+      "step": 3717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.06395611862412e-06,
+      "loss": 0.9464,
+      "step": 3718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.063036596545857e-06,
+      "loss": 0.8772,
+      "step": 3719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.062116960244387e-06,
+      "loss": 1.1453,
+      "step": 3720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.061197209818525e-06,
+      "loss": 0.9824,
+      "step": 3721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.060277345367099e-06,
+      "loss": 0.7891,
+      "step": 3722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0593573669889476e-06,
+      "loss": 1.0241,
+      "step": 3723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.058437274782923e-06,
+      "loss": 0.9529,
+      "step": 3724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.057517068847889e-06,
+      "loss": 0.7969,
+      "step": 3725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.056596749282721e-06,
+      "loss": 0.9884,
+      "step": 3726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0556763161863075e-06,
+      "loss": 0.9444,
+      "step": 3727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.05475576965755e-06,
+      "loss": 0.8787,
+      "step": 3728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.05383510979536e-06,
+      "loss": 0.959,
+      "step": 3729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.052914336698662e-06,
+      "loss": 0.8991,
+      "step": 3730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.051993450466394e-06,
+      "loss": 0.9443,
+      "step": 3731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0510724511975045e-06,
+      "loss": 0.962,
+      "step": 3732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.050151338990956e-06,
+      "loss": 0.8835,
+      "step": 3733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.04923011394572e-06,
+      "loss": 0.8811,
+      "step": 3734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.048308776160785e-06,
+      "loss": 0.851,
+      "step": 3735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.047387325735144e-06,
+      "loss": 1.0219,
+      "step": 3736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.046465762767812e-06,
+      "loss": 0.8954,
+      "step": 3737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.045544087357807e-06,
+      "loss": 0.8965,
+      "step": 3738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.044622299604166e-06,
+      "loss": 1.2523,
+      "step": 3739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.043700399605931e-06,
+      "loss": 0.9719,
+      "step": 3740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.042778387462165e-06,
+      "loss": 0.876,
+      "step": 3741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.041856263271934e-06,
+      "loss": 0.8843,
+      "step": 3742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.040934027134323e-06,
+      "loss": 0.8996,
+      "step": 3743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.040011679148425e-06,
+      "loss": 0.9244,
+      "step": 3744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.039089219413345e-06,
+      "loss": 1.0528,
+      "step": 3745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.038166648028204e-06,
+      "loss": 1.1299,
+      "step": 3746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.03724396509213e-06,
+      "loss": 0.9025,
+      "step": 3747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.036321170704266e-06,
+      "loss": 0.94,
+      "step": 3748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.035398264963767e-06,
+      "loss": 1.0245,
+      "step": 3749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.034475247969797e-06,
+      "loss": 0.8154,
+      "step": 3750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.033552119821536e-06,
+      "loss": 0.7817,
+      "step": 3751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.032628880618173e-06,
+      "loss": 0.9285,
+      "step": 3752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.031705530458911e-06,
+      "loss": 0.8617,
+      "step": 3753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.030782069442963e-06,
+      "loss": 0.9429,
+      "step": 3754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0298584976695565e-06,
+      "loss": 0.9416,
+      "step": 3755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.028934815237927e-06,
+      "loss": 1.0098,
+      "step": 3756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.028011022247324e-06,
+      "loss": 0.8809,
+      "step": 3757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0270871187970114e-06,
+      "loss": 0.9568,
+      "step": 3758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.026163104986261e-06,
+      "loss": 0.9239,
+      "step": 3759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.025238980914357e-06,
+      "loss": 0.882,
+      "step": 3760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.024314746680599e-06,
+      "loss": 0.965,
+      "step": 3761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.023390402384295e-06,
+      "loss": 0.9596,
+      "step": 3762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.022465948124764e-06,
+      "loss": 0.8378,
+      "step": 3763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.02154138400134e-06,
+      "loss": 1.1284,
+      "step": 3764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.020616710113368e-06,
+      "loss": 0.9128,
+      "step": 3765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0196919265602026e-06,
+      "loss": 1.0149,
+      "step": 3766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.018767033441213e-06,
+      "loss": 1.0,
+      "step": 3767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.017842030855778e-06,
+      "loss": 0.958,
+      "step": 3768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.016916918903289e-06,
+      "loss": 0.9632,
+      "step": 3769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.01599169768315e-06,
+      "loss": 1.1191,
+      "step": 3770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.015066367294776e-06,
+      "loss": 1.1033,
+      "step": 3771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.014140927837593e-06,
+      "loss": 0.8101,
+      "step": 3772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.01321537941104e-06,
+      "loss": 0.7795,
+      "step": 3773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.012289722114565e-06,
+      "loss": 0.9427,
+      "step": 3774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.011363956047634e-06,
+      "loss": 0.9194,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.010438081309716e-06,
+      "loss": 0.8964,
+      "step": 3776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.009512098000301e-06,
+      "loss": 0.9051,
+      "step": 3777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.00858600621888e-06,
+      "loss": 0.8975,
+      "step": 3778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0076598060649664e-06,
+      "loss": 1.1255,
+      "step": 3779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0067334976380775e-06,
+      "loss": 1.0583,
+      "step": 3780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.005807081037748e-06,
+      "loss": 0.8134,
+      "step": 3781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0048805563635175e-06,
+      "loss": 0.911,
+      "step": 3782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.003953923714944e-06,
+      "loss": 0.8437,
+      "step": 3783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.003027183191593e-06,
+      "loss": 0.9268,
+      "step": 3784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.002100334893042e-06,
+      "loss": 1.0368,
+      "step": 3785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.001173378918881e-06,
+      "loss": 0.8529,
+      "step": 3786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.000246315368714e-06,
+      "loss": 0.9483,
+      "step": 3787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9993191443421505e-06,
+      "loss": 0.7421,
+      "step": 3788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.998391865938816e-06,
+      "loss": 1.0262,
+      "step": 3789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.997464480258347e-06,
+      "loss": 0.9229,
+      "step": 3790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.996536987400391e-06,
+      "loss": 0.9232,
+      "step": 3791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9956093874646075e-06,
+      "loss": 0.7749,
+      "step": 3792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.994681680550665e-06,
+      "loss": 1.0376,
+      "step": 3793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.993753866758248e-06,
+      "loss": 0.811,
+      "step": 3794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.992825946187048e-06,
+      "loss": 0.8191,
+      "step": 3795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.991897918936771e-06,
+      "loss": 1.2268,
+      "step": 3796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.990969785107134e-06,
+      "loss": 0.9575,
+      "step": 3797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.990041544797864e-06,
+      "loss": 1.2313,
+      "step": 3798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9891131981087e-06,
+      "loss": 1.0532,
+      "step": 3799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.988184745139395e-06,
+      "loss": 1.0772,
+      "step": 3800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.987256185989708e-06,
+      "loss": 0.9932,
+      "step": 3801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.986327520759415e-06,
+      "loss": 0.8713,
+      "step": 3802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.985398749548301e-06,
+      "loss": 1.06,
+      "step": 3803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.984469872456161e-06,
+      "loss": 1.0111,
+      "step": 3804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.983540889582802e-06,
+      "loss": 1.0816,
+      "step": 3805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9826118010280475e-06,
+      "loss": 0.9222,
+      "step": 3806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9816826068917226e-06,
+      "loss": 0.9359,
+      "step": 3807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.980753307273673e-06,
+      "loss": 1.2298,
+      "step": 3808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.97982390227375e-06,
+      "loss": 0.9393,
+      "step": 3809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9788943919918195e-06,
+      "loss": 0.8787,
+      "step": 3810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.977964776527756e-06,
+      "loss": 0.7547,
+      "step": 3811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.977035055981446e-06,
+      "loss": 0.9754,
+      "step": 3812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.97610523045279e-06,
+      "loss": 0.9537,
+      "step": 3813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.975175300041696e-06,
+      "loss": 1.0461,
+      "step": 3814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9742452648480865e-06,
+      "loss": 0.8756,
+      "step": 3815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.973315124971892e-06,
+      "loss": 0.9661,
+      "step": 3816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.972384880513057e-06,
+      "loss": 0.8536,
+      "step": 3817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.971454531571537e-06,
+      "loss": 0.9163,
+      "step": 3818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.970524078247295e-06,
+      "loss": 0.8901,
+      "step": 3819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9695935206403115e-06,
+      "loss": 0.9448,
+      "step": 3820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.968662858850573e-06,
+      "loss": 1.0035,
+      "step": 3821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.967732092978079e-06,
+      "loss": 0.892,
+      "step": 3822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.966801223122841e-06,
+      "loss": 1.0656,
+      "step": 3823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.965870249384881e-06,
+      "loss": 0.8404,
+      "step": 3824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.964939171864231e-06,
+      "loss": 1.0061,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.964007990660936e-06,
+      "loss": 0.8025,
+      "step": 3826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.963076705875051e-06,
+      "loss": 1.002,
+      "step": 3827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.962145317606643e-06,
+      "loss": 0.9586,
+      "step": 3828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.961213825955789e-06,
+      "loss": 0.8764,
+      "step": 3829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.960282231022577e-06,
+      "loss": 0.8868,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.95935053290711e-06,
+      "loss": 1.0285,
+      "step": 3831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9584187317094955e-06,
+      "loss": 1.0641,
+      "step": 3832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.957486827529858e-06,
+      "loss": 1.0307,
+      "step": 3833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.956554820468327e-06,
+      "loss": 1.1348,
+      "step": 3834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.95562271062505e-06,
+      "loss": 0.9925,
+      "step": 3835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.954690498100181e-06,
+      "loss": 0.9428,
+      "step": 3836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953758182993886e-06,
+      "loss": 0.8575,
+      "step": 3837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952825765406342e-06,
+      "loss": 0.8825,
+      "step": 3838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9518932454377375e-06,
+      "loss": 0.8805,
+      "step": 3839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950960623188272e-06,
+      "loss": 0.8299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950027898758155e-06,
+      "loss": 0.7966,
+      "step": 3841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.949095072247608e-06,
+      "loss": 0.9059,
+      "step": 3842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.948162143756864e-06,
+      "loss": 0.8639,
+      "step": 3843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.947229113386164e-06,
+      "loss": 1.1212,
+      "step": 3844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.946295981235765e-06,
+      "loss": 0.8818,
+      "step": 3845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.945362747405929e-06,
+      "loss": 0.8763,
+      "step": 3846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.944429411996935e-06,
+      "loss": 1.0639,
+      "step": 3847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9434959751090664e-06,
+      "loss": 0.8457,
+      "step": 3848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.942562436842623e-06,
+      "loss": 0.9986,
+      "step": 3849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.941628797297913e-06,
+      "loss": 1.0204,
+      "step": 3850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.940695056575257e-06,
+      "loss": 0.8287,
+      "step": 3851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.939761214774982e-06,
+      "loss": 1.0085,
+      "step": 3852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.938827271997434e-06,
+      "loss": 1.1025,
+      "step": 3853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.937893228342961e-06,
+      "loss": 0.7536,
+      "step": 3854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936959083911928e-06,
+      "loss": 0.9094,
+      "step": 3855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936024838804708e-06,
+      "loss": 0.9964,
+      "step": 3856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.935090493121686e-06,
+      "loss": 1.1451,
+      "step": 3857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.934156046963257e-06,
+      "loss": 1.0174,
+      "step": 3858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.933221500429828e-06,
+      "loss": 0.8157,
+      "step": 3859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.932286853621816e-06,
+      "loss": 1.1481,
+      "step": 3860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.931352106639647e-06,
+      "loss": 1.1517,
+      "step": 3861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.930417259583761e-06,
+      "loss": 1.0351,
+      "step": 3862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.929482312554607e-06,
+      "loss": 0.9193,
+      "step": 3863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.928547265652644e-06,
+      "loss": 1.0855,
+      "step": 3864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.927612118978345e-06,
+      "loss": 0.9827,
+      "step": 3865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.926676872632188e-06,
+      "loss": 0.9447,
+      "step": 3866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.92574152671467e-06,
+      "loss": 0.8476,
+      "step": 3867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9248060813262886e-06,
+      "loss": 1.1827,
+      "step": 3868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.923870536567561e-06,
+      "loss": 1.0924,
+      "step": 3869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9229348925390095e-06,
+      "loss": 1.013,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.92199914934117e-06,
+      "loss": 0.7901,
+      "step": 3871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.921063307074587e-06,
+      "loss": 0.9834,
+      "step": 3872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.920127365839818e-06,
+      "loss": 1.0297,
+      "step": 3873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.919191325737429e-06,
+      "loss": 0.8885,
+      "step": 3874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.918255186867998e-06,
+      "loss": 1.0376,
+      "step": 3875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.917318949332112e-06,
+      "loss": 0.8614,
+      "step": 3876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.916382613230371e-06,
+      "loss": 0.9026,
+      "step": 3877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.915446178663383e-06,
+      "loss": 1.0544,
+      "step": 3878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.914509645731769e-06,
+      "loss": 1.0426,
+      "step": 3879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.913573014536159e-06,
+      "loss": 0.829,
+      "step": 3880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.912636285177194e-06,
+      "loss": 0.8553,
+      "step": 3881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.911699457755524e-06,
+      "loss": 0.9856,
+      "step": 3882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9107625323718135e-06,
+      "loss": 0.7343,
+      "step": 3883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.909825509126733e-06,
+      "loss": 0.9011,
+      "step": 3884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.908888388120968e-06,
+      "loss": 1.358,
+      "step": 3885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.907951169455208e-06,
+      "loss": 0.7838,
+      "step": 3886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.907013853230162e-06,
+      "loss": 0.9331,
+      "step": 3887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.906076439546541e-06,
+      "loss": 0.8857,
+      "step": 3888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.905138928505072e-06,
+      "loss": 0.817,
+      "step": 3889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.904201320206488e-06,
+      "loss": 0.8566,
+      "step": 3890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.903263614751539e-06,
+      "loss": 0.8787,
+      "step": 3891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.902325812240977e-06,
+      "loss": 0.7788,
+      "step": 3892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.901387912775572e-06,
+      "loss": 0.8107,
+      "step": 3893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9004499164561e-06,
+      "loss": 0.8029,
+      "step": 3894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.899511823383348e-06,
+      "loss": 0.9557,
+      "step": 3895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.898573633658116e-06,
+      "loss": 1.012,
+      "step": 3896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.897635347381211e-06,
+      "loss": 0.9658,
+      "step": 3897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.896696964653451e-06,
+      "loss": 1.1055,
+      "step": 3898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8957584855756666e-06,
+      "loss": 0.8522,
+      "step": 3899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.894819910248697e-06,
+      "loss": 0.8496,
+      "step": 3900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.893881238773392e-06,
+      "loss": 0.9057,
+      "step": 3901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.892942471250612e-06,
+      "loss": 0.8052,
+      "step": 3902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.892003607781226e-06,
+      "loss": 0.9694,
+      "step": 3903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.891064648466117e-06,
+      "loss": 0.8927,
+      "step": 3904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8901255934061745e-06,
+      "loss": 0.8468,
+      "step": 3905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.889186442702301e-06,
+      "loss": 0.8252,
+      "step": 3906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.888247196455407e-06,
+      "loss": 0.9465,
+      "step": 3907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8873078547664165e-06,
+      "loss": 0.8557,
+      "step": 3908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.886368417736259e-06,
+      "loss": 0.8883,
+      "step": 3909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.885428885465879e-06,
+      "loss": 0.8395,
+      "step": 3910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.884489258056227e-06,
+      "loss": 0.7982,
+      "step": 3911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.88354953560827e-06,
+      "loss": 1.0204,
+      "step": 3912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.882609718222976e-06,
+      "loss": 1.2195,
+      "step": 3913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.881669806001332e-06,
+      "loss": 0.8069,
+      "step": 3914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.88072979904433e-06,
+      "loss": 0.8882,
+      "step": 3915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8797896974529745e-06,
+      "loss": 0.8832,
+      "step": 3916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.878849501328277e-06,
+      "loss": 0.8312,
+      "step": 3917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.877909210771265e-06,
+      "loss": 0.981,
+      "step": 3918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.87696882588297e-06,
+      "loss": 1.0386,
+      "step": 3919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.876028346764438e-06,
+      "loss": 1.0622,
+      "step": 3920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.875087773516722e-06,
+      "loss": 0.8422,
+      "step": 3921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.874147106240888e-06,
+      "loss": 1.1257,
+      "step": 3922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.873206345038009e-06,
+      "loss": 1.1398,
+      "step": 3923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.872265490009171e-06,
+      "loss": 1.0428,
+      "step": 3924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.871324541255468e-06,
+      "loss": 0.8139,
+      "step": 3925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.870383498878006e-06,
+      "loss": 0.836,
+      "step": 3926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.869442362977898e-06,
+      "loss": 1.0325,
+      "step": 3927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.868501133656271e-06,
+      "loss": 1.0455,
+      "step": 3928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.867559811014258e-06,
+      "loss": 0.904,
+      "step": 3929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.866618395153007e-06,
+      "loss": 1.1511,
+      "step": 3930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.86567688617367e-06,
+      "loss": 1.0719,
+      "step": 3931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.864735284177414e-06,
+      "loss": 0.792,
+      "step": 3932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.863793589265412e-06,
+      "loss": 1.0233,
+      "step": 3933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8628518015388526e-06,
+      "loss": 1.089,
+      "step": 3934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.861909921098928e-06,
+      "loss": 0.7632,
+      "step": 3935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.860967948046843e-06,
+      "loss": 1.0903,
+      "step": 3936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.860025882483814e-06,
+      "loss": 0.9723,
+      "step": 3937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.859083724511066e-06,
+      "loss": 0.8795,
+      "step": 3938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.858141474229831e-06,
+      "loss": 0.8388,
+      "step": 3939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8571991317413576e-06,
+      "loss": 0.9238,
+      "step": 3940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.856256697146898e-06,
+      "loss": 1.0554,
+      "step": 3941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.855314170547718e-06,
+      "loss": 0.9815,
+      "step": 3942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.854371552045092e-06,
+      "loss": 1.0006,
+      "step": 3943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.853428841740302e-06,
+      "loss": 0.9339,
+      "step": 3944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.852486039734644e-06,
+      "loss": 1.2821,
+      "step": 3945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.851543146129423e-06,
+      "loss": 0.791,
+      "step": 3946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8506001610259515e-06,
+      "loss": 0.9255,
+      "step": 3947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.849657084525554e-06,
+      "loss": 0.8988,
+      "step": 3948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8487139167295615e-06,
+      "loss": 1.0086,
+      "step": 3949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8477706577393204e-06,
+      "loss": 0.847,
+      "step": 3950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.846827307656182e-06,
+      "loss": 0.7646,
+      "step": 3951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8458838665815114e-06,
+      "loss": 0.921,
+      "step": 3952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.844940334616678e-06,
+      "loss": 0.8931,
+      "step": 3953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.843996711863067e-06,
+      "loss": 0.9235,
+      "step": 3954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8430529984220685e-06,
+      "loss": 0.9882,
+      "step": 3955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.842109194395087e-06,
+      "loss": 1.102,
+      "step": 3956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.84116529988353e-06,
+      "loss": 1.0022,
+      "step": 3957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.840221314988824e-06,
+      "loss": 0.9483,
+      "step": 3958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.839277239812396e-06,
+      "loss": 0.8593,
+      "step": 3959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.83833307445569e-06,
+      "loss": 0.8057,
+      "step": 3960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.837388819020154e-06,
+      "loss": 1.1992,
+      "step": 3961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.83644447360725e-06,
+      "loss": 1.0125,
+      "step": 3962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8355000383184466e-06,
+      "loss": 0.8311,
+      "step": 3963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8345555132552224e-06,
+      "loss": 0.6868,
+      "step": 3964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.83361089851907e-06,
+      "loss": 0.8084,
+      "step": 3965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.832666194211484e-06,
+      "loss": 0.8894,
+      "step": 3966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.831721400433976e-06,
+      "loss": 0.9127,
+      "step": 3967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.830776517288063e-06,
+      "loss": 0.7486,
+      "step": 3968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.829831544875273e-06,
+      "loss": 1.0235,
+      "step": 3969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.828886483297142e-06,
+      "loss": 0.9726,
+      "step": 3970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.827941332655217e-06,
+      "loss": 0.997,
+      "step": 3971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.826996093051056e-06,
+      "loss": 1.0046,
+      "step": 3972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.826050764586223e-06,
+      "loss": 1.1332,
+      "step": 3973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.825105347362295e-06,
+      "loss": 0.9745,
+      "step": 3974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.824159841480856e-06,
+      "loss": 0.7514,
+      "step": 3975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8232142470435e-06,
+      "loss": 1.0914,
+      "step": 3976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.822268564151833e-06,
+      "loss": 0.7727,
+      "step": 3977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.821322792907466e-06,
+      "loss": 1.1581,
+      "step": 3978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.820376933412026e-06,
+      "loss": 1.0128,
+      "step": 3979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.81943098576714e-06,
+      "loss": 0.9484,
+      "step": 3980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.818484950074456e-06,
+      "loss": 1.034,
+      "step": 3981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.817538826435621e-06,
+      "loss": 1.0288,
+      "step": 3982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.816592614952298e-06,
+      "loss": 0.8392,
+      "step": 3983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.815646315726155e-06,
+      "loss": 1.0208,
+      "step": 3984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.814699928858875e-06,
+      "loss": 1.0332,
+      "step": 3985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8137534544521463e-06,
+      "loss": 0.9151,
+      "step": 3986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8128068926076663e-06,
+      "loss": 0.8192,
+      "step": 3987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.811860243427144e-06,
+      "loss": 0.9774,
+      "step": 3988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.810913507012296e-06,
+      "loss": 0.8152,
+      "step": 3989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8099666834648503e-06,
+      "loss": 1.0455,
+      "step": 3990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8090197728865415e-06,
+      "loss": 0.9135,
+      "step": 3991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.808072775379116e-06,
+      "loss": 0.8247,
+      "step": 3992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.807125691044329e-06,
+      "loss": 1.0436,
+      "step": 3993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8061785199839433e-06,
+      "loss": 0.8997,
+      "step": 3994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8052312622997337e-06,
+      "loss": 1.2254,
+      "step": 3995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8042839180934824e-06,
+      "loss": 0.8059,
+      "step": 3996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8033364874669815e-06,
+      "loss": 0.8674,
+      "step": 3997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8023889705220313e-06,
+      "loss": 1.0595,
+      "step": 3998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.801441367360445e-06,
+      "loss": 0.8738,
+      "step": 3999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8004936780840405e-06,
+      "loss": 0.8604,
+      "step": 4000
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0493,
+      "eval_samples_per_second": 351.843,
+      "eval_steps_per_second": 43.989,
+      "step": 4000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7995459027946474e-06,
+      "loss": 1.0068,
+      "step": 4001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7985980415941033e-06,
+      "loss": 1.1087,
+      "step": 4002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7976500945842576e-06,
+      "loss": 0.9803,
+      "step": 4003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7967020618669655e-06,
+      "loss": 0.982,
+      "step": 4004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7957539435440946e-06,
+      "loss": 0.9607,
+      "step": 4005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7948057397175185e-06,
+      "loss": 0.9886,
+      "step": 4006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7938574504891224e-06,
+      "loss": 0.9852,
+      "step": 4007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7929090759608e-06,
+      "loss": 0.9313,
+      "step": 4008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.791960616234454e-06,
+      "loss": 0.7548,
+      "step": 4009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.791012071411996e-06,
+      "loss": 1.0853,
+      "step": 4010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7900634415953483e-06,
+      "loss": 0.8809,
+      "step": 4011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7891147268864387e-06,
+      "loss": 0.9649,
+      "step": 4012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.78816592738721e-06,
+      "loss": 0.9716,
+      "step": 4013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7872170431996067e-06,
+      "loss": 0.9801,
+      "step": 4014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.78626807442559e-06,
+      "loss": 0.8498,
+      "step": 4015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7853190211671237e-06,
+      "loss": 0.9078,
+      "step": 4016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7843698835261848e-06,
+      "loss": 1.0606,
+      "step": 4017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7834206616047573e-06,
+      "loss": 0.9788,
+      "step": 4018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.782471355504837e-06,
+      "loss": 0.8211,
+      "step": 4019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.781521965328424e-06,
+      "loss": 1.0115,
+      "step": 4020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7805724911775327e-06,
+      "loss": 0.912,
+      "step": 4021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.779622933154182e-06,
+      "loss": 0.9372,
+      "step": 4022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7786732913604038e-06,
+      "loss": 0.9856,
+      "step": 4023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7777235658982345e-06,
+      "loss": 0.8821,
+      "step": 4024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.776773756869725e-06,
+      "loss": 0.9869,
+      "step": 4025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7758238643769297e-06,
+      "loss": 0.8875,
+      "step": 4026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7748738885219157e-06,
+      "loss": 0.8873,
+      "step": 4027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7739238294067587e-06,
+      "loss": 0.87,
+      "step": 4028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.77297368713354e-06,
+      "loss": 1.078,
+      "step": 4029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.772023461804355e-06,
+      "loss": 0.9641,
+      "step": 4030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7710731535213036e-06,
+      "loss": 0.8619,
+      "step": 4031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.770122762386497e-06,
+      "loss": 0.9749,
+      "step": 4032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7691722885020543e-06,
+      "loss": 0.7779,
+      "step": 4033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7682217319701042e-06,
+      "loss": 1.0465,
+      "step": 4034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7672710928927834e-06,
+      "loss": 0.8022,
+      "step": 4035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7663203713722394e-06,
+      "loss": 1.1332,
+      "step": 4036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.765369567510626e-06,
+      "loss": 1.1939,
+      "step": 4037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.764418681410107e-06,
+      "loss": 1.034,
+      "step": 4038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.763467713172854e-06,
+      "loss": 1.0067,
+      "step": 4039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7625166629010517e-06,
+      "loss": 0.8228,
+      "step": 4040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.761565530696887e-06,
+      "loss": 0.7287,
+      "step": 4041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.760614316662561e-06,
+      "loss": 1.0984,
+      "step": 4042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7596630209002807e-06,
+      "loss": 1.0487,
+      "step": 4043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7587116435122636e-06,
+      "loss": 0.8251,
+      "step": 4044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.757760184600734e-06,
+      "loss": 1.0126,
+      "step": 4045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.756808644267927e-06,
+      "loss": 1.045,
+      "step": 4046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7558570226160844e-06,
+      "loss": 0.9922,
+      "step": 4047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.754905319747459e-06,
+      "loss": 1.0089,
+      "step": 4048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7539535357643105e-06,
+      "loss": 1.1247,
+      "step": 4049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7530016707689096e-06,
+      "loss": 1.032,
+      "step": 4050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.752049724863532e-06,
+      "loss": 0.8223,
+      "step": 4051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7510976981504655e-06,
+      "loss": 1.0056,
+      "step": 4052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7501455907320045e-06,
+      "loss": 0.9726,
+      "step": 4053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7491934027104536e-06,
+      "loss": 0.9594,
+      "step": 4054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7482411341881246e-06,
+      "loss": 0.7893,
+      "step": 4055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7472887852673397e-06,
+      "loss": 0.7602,
+      "step": 4056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.746336356050428e-06,
+      "loss": 0.8603,
+      "step": 4057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7453838466397286e-06,
+      "loss": 0.9778,
+      "step": 4058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.744431257137587e-06,
+      "loss": 0.8884,
+      "step": 4059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7434785876463604e-06,
+      "loss": 0.9505,
+      "step": 4060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.742525838268413e-06,
+      "loss": 0.8615,
+      "step": 4061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7415730091061173e-06,
+      "loss": 0.9678,
+      "step": 4062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.740620100261854e-06,
+      "loss": 0.7841,
+      "step": 4063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.739667111838014e-06,
+      "loss": 1.0094,
+      "step": 4064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.738714043936996e-06,
+      "loss": 0.8817,
+      "step": 4065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7377608966612063e-06,
+      "loss": 0.9775,
+      "step": 4066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7368076701130607e-06,
+      "loss": 1.0554,
+      "step": 4067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.735854364394984e-06,
+      "loss": 0.8733,
+      "step": 4068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7349009796094087e-06,
+      "loss": 0.9476,
+      "step": 4069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.733947515858774e-06,
+      "loss": 0.9747,
+      "step": 4070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.732993973245533e-06,
+      "loss": 1.1685,
+      "step": 4071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7320403518721402e-06,
+      "loss": 0.9597,
+      "step": 4072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7310866518410653e-06,
+      "loss": 0.7094,
+      "step": 4073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7301328732547814e-06,
+      "loss": 0.9585,
+      "step": 4074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7291790162157722e-06,
+      "loss": 1.2543,
+      "step": 4075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7282250808265293e-06,
+      "loss": 0.8419,
+      "step": 4076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7272710671895546e-06,
+      "loss": 0.8111,
+      "step": 4077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.726316975407354e-06,
+      "loss": 0.8578,
+      "step": 4078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7253628055824474e-06,
+      "loss": 1.2262,
+      "step": 4079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.724408557817359e-06,
+      "loss": 1.0303,
+      "step": 4080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.723454232214623e-06,
+      "loss": 0.9475,
+      "step": 4081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.72249982887678e-06,
+      "loss": 0.9208,
+      "step": 4082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7215453479063834e-06,
+      "loss": 0.9444,
+      "step": 4083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.72059078940599e-06,
+      "loss": 0.9563,
+      "step": 4084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.719636153478168e-06,
+      "loss": 0.8827,
+      "step": 4085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.718681440225492e-06,
+      "loss": 0.9106,
+      "step": 4086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7177266497505474e-06,
+      "loss": 0.7753,
+      "step": 4087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.716771782155925e-06,
+      "loss": 0.8751,
+      "step": 4088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.715816837544226e-06,
+      "loss": 0.9886,
+      "step": 4089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7148618160180585e-06,
+      "loss": 1.1248,
+      "step": 4090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7139067176800402e-06,
+      "loss": 0.9552,
+      "step": 4091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.712951542632796e-06,
+      "loss": 0.8236,
+      "step": 4092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7119962909789593e-06,
+      "loss": 0.9933,
+      "step": 4093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.711040962821172e-06,
+      "loss": 0.9507,
+      "step": 4094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7100855582620845e-06,
+      "loss": 0.7273,
+      "step": 4095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.709130077404353e-06,
+      "loss": 0.9527,
+      "step": 4096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.708174520350647e-06,
+      "loss": 1.0169,
+      "step": 4097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.707218887203638e-06,
+      "loss": 0.7894,
+      "step": 4098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.706263178066011e-06,
+      "loss": 0.9018,
+      "step": 4099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7053073930404547e-06,
+      "loss": 0.9593,
+      "step": 4100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7043515322296704e-06,
+      "loss": 1.0466,
+      "step": 4101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.703395595736364e-06,
+      "loss": 0.973,
+      "step": 4102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7024395836632514e-06,
+      "loss": 0.7311,
+      "step": 4103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.701483496113055e-06,
+      "loss": 0.7305,
+      "step": 4104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7005273331885075e-06,
+      "loss": 1.1698,
+      "step": 4105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6995710949923475e-06,
+      "loss": 0.9429,
+      "step": 4106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.698614781627324e-06,
+      "loss": 0.9108,
+      "step": 4107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.697658393196191e-06,
+      "loss": 1.0381,
+      "step": 4108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6967019298017144e-06,
+      "loss": 0.8929,
+      "step": 4109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.695745391546665e-06,
+      "loss": 0.8746,
+      "step": 4110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6947887785338225e-06,
+      "loss": 0.9581,
+      "step": 4111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6938320908659754e-06,
+      "loss": 0.7451,
+      "step": 4112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.69287532864592e-06,
+      "loss": 0.9989,
+      "step": 4113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.69191849197646e-06,
+      "loss": 0.9757,
+      "step": 4114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6909615809604077e-06,
+      "loss": 1.0695,
+      "step": 4115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6900045957005815e-06,
+      "loss": 0.7498,
+      "step": 4116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6890475362998118e-06,
+      "loss": 0.8496,
+      "step": 4117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6880904028609336e-06,
+      "loss": 1.0019,
+      "step": 4118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6871331954867903e-06,
+      "loss": 0.7593,
+      "step": 4119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6861759142802337e-06,
+      "loss": 0.7554,
+      "step": 4120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.685218559344125e-06,
+      "loss": 1.0455,
+      "step": 4121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.68426113078133e-06,
+      "loss": 1.0572,
+      "step": 4122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6833036286947267e-06,
+      "loss": 1.035,
+      "step": 4123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.682346053187196e-06,
+      "loss": 0.7394,
+      "step": 4124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6813884043616313e-06,
+      "loss": 1.1489,
+      "step": 4125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.680430682320931e-06,
+      "loss": 1.1311,
+      "step": 4126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6794728871680023e-06,
+      "loss": 1.0837,
+      "step": 4127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6785150190057602e-06,
+      "loss": 0.82,
+      "step": 4128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6775570779371287e-06,
+      "loss": 0.9707,
+      "step": 4129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6765990640650365e-06,
+      "loss": 1.0821,
+      "step": 4130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6756409774924238e-06,
+      "loss": 0.9289,
+      "step": 4131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6746828183222353e-06,
+      "loss": 0.9753,
+      "step": 4132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6737245866574273e-06,
+      "loss": 0.9821,
+      "step": 4133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.67276628260096e-06,
+      "loss": 1.0676,
+      "step": 4134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6718079062558033e-06,
+      "loss": 1.0897,
+      "step": 4135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.670849457724935e-06,
+      "loss": 0.7526,
+      "step": 4136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6698909371113415e-06,
+      "loss": 0.8059,
+      "step": 4137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.668932344518013e-06,
+      "loss": 0.9817,
+      "step": 4138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.667973680047953e-06,
+      "loss": 0.8009,
+      "step": 4139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.667014943804168e-06,
+      "loss": 1.2015,
+      "step": 4140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.666056135889676e-06,
+      "loss": 0.9084,
+      "step": 4141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6650972564074983e-06,
+      "loss": 1.1919,
+      "step": 4142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.664138305460669e-06,
+      "loss": 1.0359,
+      "step": 4143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6631792831522252e-06,
+      "loss": 0.9713,
+      "step": 4144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6622201895852156e-06,
+      "loss": 0.9381,
+      "step": 4145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.661261024862693e-06,
+      "loss": 0.9056,
+      "step": 4146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.660301789087722e-06,
+      "loss": 0.8577,
+      "step": 4147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.659342482363369e-06,
+      "loss": 0.8956,
+      "step": 4148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6583831047927145e-06,
+      "loss": 0.9183,
+      "step": 4149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6574236564788423e-06,
+      "loss": 0.8903,
+      "step": 4150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6564641375248448e-06,
+      "loss": 0.7841,
+      "step": 4151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.655504548033822e-06,
+      "loss": 0.9013,
+      "step": 4152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.654544888108884e-06,
+      "loss": 0.6859,
+      "step": 4153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6535851578531425e-06,
+      "loss": 1.1309,
+      "step": 4154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6526253573697243e-06,
+      "loss": 0.8477,
+      "step": 4155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6516654867617563e-06,
+      "loss": 0.7977,
+      "step": 4156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.65070554613238e-06,
+      "loss": 0.8256,
+      "step": 4157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6497455355847374e-06,
+      "loss": 1.0216,
+      "step": 4158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.648785455221984e-06,
+      "loss": 0.9339,
+      "step": 4159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.64782530514728e-06,
+      "loss": 0.8966,
+      "step": 4160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6468650854637925e-06,
+      "loss": 0.8709,
+      "step": 4161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.645904796274699e-06,
+      "loss": 0.9412,
+      "step": 4162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.64494443768318e-06,
+      "loss": 0.9297,
+      "step": 4163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.643984009792428e-06,
+      "loss": 0.8369,
+      "step": 4164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6430235127056397e-06,
+      "loss": 0.9043,
+      "step": 4165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6420629465260217e-06,
+      "loss": 0.7783,
+      "step": 4166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.641102311356785e-06,
+      "loss": 0.8405,
+      "step": 4167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6401416073011506e-06,
+      "loss": 1.0366,
+      "step": 4168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.639180834462346e-06,
+      "loss": 0.8467,
+      "step": 4169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6382199929436077e-06,
+      "loss": 0.8341,
+      "step": 4170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.637259082848175e-06,
+      "loss": 0.9806,
+      "step": 4171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6362981042793008e-06,
+      "loss": 0.9171,
+      "step": 4172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.63533705734024e-06,
+      "loss": 0.847,
+      "step": 4173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6343759421342578e-06,
+      "loss": 0.987,
+      "step": 4174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6334147587646253e-06,
+      "loss": 0.9263,
+      "step": 4175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6324535073346233e-06,
+      "loss": 1.1115,
+      "step": 4176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6314921879475364e-06,
+      "loss": 0.9814,
+      "step": 4177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.630530800706659e-06,
+      "loss": 1.1112,
+      "step": 4178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6295693457152922e-06,
+      "loss": 0.8,
+      "step": 4179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6286078230767445e-06,
+      "loss": 0.8919,
+      "step": 4180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6276462328943298e-06,
+      "loss": 1.0514,
+      "step": 4181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6266845752713733e-06,
+      "loss": 1.1498,
+      "step": 4182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6257228503112037e-06,
+      "loss": 0.9557,
+      "step": 4183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.624761058117158e-06,
+      "loss": 1.0406,
+      "step": 4184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6237991987925816e-06,
+      "loss": 1.0133,
+      "step": 4185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6228372724408264e-06,
+      "loss": 0.8782,
+      "step": 4186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6218752791652496e-06,
+      "loss": 0.9218,
+      "step": 4187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6209132190692197e-06,
+      "loss": 0.8872,
+      "step": 4188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.619951092256107e-06,
+      "loss": 1.013,
+      "step": 4189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.618988898829295e-06,
+      "loss": 1.0264,
+      "step": 4190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6180266388921694e-06,
+      "loss": 0.7283,
+      "step": 4191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6170643125481262e-06,
+      "loss": 1.0086,
+      "step": 4192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.616101919900566e-06,
+      "loss": 0.7666,
+      "step": 4193
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.615139461052899e-06,
+      "loss": 0.9891,
+      "step": 4194
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.61417693610854e-06,
+      "loss": 0.9806,
+      "step": 4195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6132143451709137e-06,
+      "loss": 0.836,
+      "step": 4196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.612251688343449e-06,
+      "loss": 1.1176,
+      "step": 4197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.611288965729585e-06,
+      "loss": 0.9631,
+      "step": 4198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6103261774327648e-06,
+      "loss": 0.8893,
+      "step": 4199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6093633235564415e-06,
+      "loss": 1.0076,
+      "step": 4200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6084004042040704e-06,
+      "loss": 0.9367,
+      "step": 4201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.607437419479121e-06,
+      "loss": 1.0653,
+      "step": 4202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6064743694850634e-06,
+      "loss": 1.0415,
+      "step": 4203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.605511254325379e-06,
+      "loss": 0.8028,
+      "step": 4204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.604548074103552e-06,
+      "loss": 1.0535,
+      "step": 4205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.603584828923079e-06,
+      "loss": 1.0591,
+      "step": 4206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6026215188874584e-06,
+      "loss": 0.9426,
+      "step": 4207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.601658144100199e-06,
+      "loss": 1.0444,
+      "step": 4208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.600694704664815e-06,
+      "loss": 0.9884,
+      "step": 4209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.599731200684828e-06,
+      "loss": 1.0701,
+      "step": 4210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5987676322637653e-06,
+      "loss": 0.9111,
+      "step": 4211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5978039995051644e-06,
+      "loss": 1.1379,
+      "step": 4212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5968403025125654e-06,
+      "loss": 0.9609,
+      "step": 4213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5958765413895196e-06,
+      "loss": 0.8224,
+      "step": 4214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5949127162395804e-06,
+      "loss": 0.7783,
+      "step": 4215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5939488271663143e-06,
+      "loss": 0.8531,
+      "step": 4216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5929848742732877e-06,
+      "loss": 0.9436,
+      "step": 4217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5920208576640787e-06,
+      "loss": 0.9011,
+      "step": 4218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5910567774422715e-06,
+      "loss": 0.9087,
+      "step": 4219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.590092633711455e-06,
+      "loss": 1.1456,
+      "step": 4220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5891284265752273e-06,
+      "loss": 0.9713,
+      "step": 4221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.588164156137193e-06,
+      "loss": 1.1566,
+      "step": 4222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5871998225009613e-06,
+      "loss": 0.9939,
+      "step": 4223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.586235425770151e-06,
+      "loss": 0.9316,
+      "step": 4224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.585270966048385e-06,
+      "loss": 0.9518,
+      "step": 4225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5843064434392958e-06,
+      "loss": 0.9397,
+      "step": 4226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.583341858046522e-06,
+      "loss": 1.0128,
+      "step": 4227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.582377209973706e-06,
+      "loss": 0.862,
+      "step": 4228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5814124993245016e-06,
+      "loss": 1.0805,
+      "step": 4229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5804477262025643e-06,
+      "loss": 0.9244,
+      "step": 4230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.579482890711561e-06,
+      "loss": 0.7953,
+      "step": 4231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.578517992955162e-06,
+      "loss": 0.8448,
+      "step": 4232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.577553033037047e-06,
+      "loss": 0.9229,
+      "step": 4233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5765880110608983e-06,
+      "loss": 0.9596,
+      "step": 4234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.57562292713041e-06,
+      "loss": 0.833,
+      "step": 4235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5746577813492786e-06,
+      "loss": 0.7614,
+      "step": 4236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.57369257382121e-06,
+      "loss": 0.8277,
+      "step": 4237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5727273046499153e-06,
+      "loss": 1.0229,
+      "step": 4238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.571761973939113e-06,
+      "loss": 1.1311,
+      "step": 4239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5707965817925268e-06,
+      "loss": 0.9243,
+      "step": 4240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.569831128313889e-06,
+      "loss": 0.766,
+      "step": 4241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5688656136069363e-06,
+      "loss": 0.9055,
+      "step": 4242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.567900037775415e-06,
+      "loss": 0.9399,
+      "step": 4243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5669344009230744e-06,
+      "loss": 0.9336,
+      "step": 4244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5659687031536737e-06,
+      "loss": 0.8251,
+      "step": 4245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5650029445709754e-06,
+      "loss": 0.7857,
+      "step": 4246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5640371252787518e-06,
+      "loss": 0.9072,
+      "step": 4247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.563071245380778e-06,
+      "loss": 0.9087,
+      "step": 4248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5621053049808408e-06,
+      "loss": 1.0383,
+      "step": 4249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5611393041827274e-06,
+      "loss": 0.8522,
+      "step": 4250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5601732430902372e-06,
+      "loss": 0.9792,
+      "step": 4251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5592071218071704e-06,
+      "loss": 1.1862,
+      "step": 4252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5582409404373396e-06,
+      "loss": 0.7801,
+      "step": 4253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5572746990845585e-06,
+      "loss": 1.1832,
+      "step": 4254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5563083978526513e-06,
+      "loss": 0.7778,
+      "step": 4255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.555342036845446e-06,
+      "loss": 0.994,
+      "step": 4256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.554375616166779e-06,
+      "loss": 1.1453,
+      "step": 4257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5534091359204914e-06,
+      "loss": 0.8163,
+      "step": 4258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5524425962104322e-06,
+      "loss": 0.9811,
+      "step": 4259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.551475997140454e-06,
+      "loss": 0.8216,
+      "step": 4260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5505093388144204e-06,
+      "loss": 0.9545,
+      "step": 4261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.549542621336197e-06,
+      "loss": 1.1106,
+      "step": 4262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.548575844809659e-06,
+      "loss": 0.9955,
+      "step": 4263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5476090093386854e-06,
+      "loss": 0.9691,
+      "step": 4264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5466421150271627e-06,
+      "loss": 1.0153,
+      "step": 4265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5456751619789835e-06,
+      "loss": 0.8249,
+      "step": 4266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.544708150298047e-06,
+      "loss": 0.8161,
+      "step": 4267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5437410800882587e-06,
+      "loss": 0.7219,
+      "step": 4268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5427739514535305e-06,
+      "loss": 0.9553,
+      "step": 4269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5418067644977798e-06,
+      "loss": 0.9718,
+      "step": 4270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.540839519324931e-06,
+      "loss": 0.9072,
+      "step": 4271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.539872216038914e-06,
+      "loss": 0.9234,
+      "step": 4272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5389048547436663e-06,
+      "loss": 1.3678,
+      "step": 4273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5379374355431302e-06,
+      "loss": 0.8501,
+      "step": 4274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5369699585412556e-06,
+      "loss": 0.9107,
+      "step": 4275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.536002423841996e-06,
+      "loss": 0.9554,
+      "step": 4276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.535034831549316e-06,
+      "loss": 0.9027,
+      "step": 4277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.53406718176718e-06,
+      "loss": 0.7721,
+      "step": 4278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5330994745995643e-06,
+      "loss": 1.1429,
+      "step": 4279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5321317101504475e-06,
+      "loss": 0.8344,
+      "step": 4280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531163888523817e-06,
+      "loss": 1.0361,
+      "step": 4281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5301960098236645e-06,
+      "loss": 0.867,
+      "step": 4282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.529228074153988e-06,
+      "loss": 1.0368,
+      "step": 4283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.528260081618793e-06,
+      "loss": 0.7578,
+      "step": 4284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5272920323220906e-06,
+      "loss": 1.0517,
+      "step": 4285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526323926367896e-06,
+      "loss": 0.9647,
+      "step": 4286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5253557638602334e-06,
+      "loss": 0.9328,
+      "step": 4287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524387544903131e-06,
+      "loss": 0.901,
+      "step": 4288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5234192696006255e-06,
+      "loss": 1.1862,
+      "step": 4289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5224509380567557e-06,
+      "loss": 1.2642,
+      "step": 4290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5214825503755707e-06,
+      "loss": 0.9363,
+      "step": 4291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520514106661122e-06,
+      "loss": 0.9245,
+      "step": 4292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51954560701747e-06,
+      "loss": 0.8686,
+      "step": 4293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5185770515486795e-06,
+      "loss": 0.8547,
+      "step": 4294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5176084403588223e-06,
+      "loss": 1.0648,
+      "step": 4295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516639773551975e-06,
+      "loss": 0.7917,
+      "step": 4296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51567105123222e-06,
+      "loss": 0.8984,
+      "step": 4297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514702273503649e-06,
+      "loss": 0.8965,
+      "step": 4298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513733440470354e-06,
+      "loss": 1.1048,
+      "step": 4299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5127645522364385e-06,
+      "loss": 0.7478,
+      "step": 4300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5117956089060078e-06,
+      "loss": 1.0793,
+      "step": 4301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.510826610583176e-06,
+      "loss": 0.8547,
+      "step": 4302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5098575573720602e-06,
+      "loss": 0.8671,
+      "step": 4303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508888449376788e-06,
+      "loss": 0.8056,
+      "step": 4304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5079192867014866e-06,
+      "loss": 0.8318,
+      "step": 4305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5069500694502955e-06,
+      "loss": 0.9719,
+      "step": 4306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5059807977273545e-06,
+      "loss": 0.9243,
+      "step": 4307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5050114716368135e-06,
+      "loss": 0.9752,
+      "step": 4308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504042091282825e-06,
+      "loss": 0.9234,
+      "step": 4309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503072656769551e-06,
+      "loss": 0.9024,
+      "step": 4310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502103168201155e-06,
+      "loss": 0.8985,
+      "step": 4311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50113362568181e-06,
+      "loss": 0.7889,
+      "step": 4312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5001640293156924e-06,
+      "loss": 0.8111,
+      "step": 4313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499194379206986e-06,
+      "loss": 0.7634,
+      "step": 4314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498224675459878e-06,
+      "loss": 0.8502,
+      "step": 4315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4972549181785652e-06,
+      "loss": 0.8013,
+      "step": 4316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4962851074672464e-06,
+      "loss": 0.9301,
+      "step": 4317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495315243430129e-06,
+      "loss": 0.9348,
+      "step": 4318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494345326171423e-06,
+      "loss": 0.934,
+      "step": 4319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.493375355795348e-06,
+      "loss": 0.9167,
+      "step": 4320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4924053324061253e-06,
+      "loss": 0.9365,
+      "step": 4321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.491435256107986e-06,
+      "loss": 0.8678,
+      "step": 4322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4904651270051623e-06,
+      "loss": 1.2068,
+      "step": 4323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4894949452018962e-06,
+      "loss": 0.9386,
+      "step": 4324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4885247108024326e-06,
+      "loss": 0.7178,
+      "step": 4325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.487554423911025e-06,
+      "loss": 0.9339,
+      "step": 4326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4865840846319277e-06,
+      "loss": 0.856,
+      "step": 4327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4856136930694065e-06,
+      "loss": 0.9913,
+      "step": 4328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4846432493277278e-06,
+      "loss": 0.8755,
+      "step": 4329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.483672753511167e-06,
+      "loss": 0.9936,
+      "step": 4330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.482702205724003e-06,
+      "loss": 0.971,
+      "step": 4331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4817316060705216e-06,
+      "loss": 0.9336,
+      "step": 4332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4807609546550138e-06,
+      "loss": 0.8179,
+      "step": 4333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4797902515817757e-06,
+      "loss": 0.9706,
+      "step": 4334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.478819496955109e-06,
+      "loss": 0.8726,
+      "step": 4335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.477848690879322e-06,
+      "loss": 0.8802,
+      "step": 4336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4768778334587264e-06,
+      "loss": 1.0507,
+      "step": 4337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.475906924797642e-06,
+      "loss": 0.8361,
+      "step": 4338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.474935965000393e-06,
+      "loss": 0.7838,
+      "step": 4339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.473964954171308e-06,
+      "loss": 1.1069,
+      "step": 4340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4729938924147225e-06,
+      "loss": 0.9552,
+      "step": 4341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4720227798349772e-06,
+      "loss": 0.815,
+      "step": 4342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4710516165364174e-06,
+      "loss": 1.0446,
+      "step": 4343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4700804026233962e-06,
+      "loss": 1.0412,
+      "step": 4344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.469109138200268e-06,
+      "loss": 0.9222,
+      "step": 4345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.468137823371397e-06,
+      "loss": 0.9518,
+      "step": 4346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4671664582411495e-06,
+      "loss": 1.0171,
+      "step": 4347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4661950429139002e-06,
+      "loss": 0.7906,
+      "step": 4348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4652235774940266e-06,
+      "loss": 0.9499,
+      "step": 4349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4642520620859133e-06,
+      "loss": 0.9546,
+      "step": 4350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.463280496793948e-06,
+      "loss": 0.8993,
+      "step": 4351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4623088817225274e-06,
+      "loss": 1.0976,
+      "step": 4352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4613372169760497e-06,
+      "loss": 1.0296,
+      "step": 4353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4603655026589217e-06,
+      "loss": 1.0342,
+      "step": 4354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4593937388755524e-06,
+      "loss": 1.1103,
+      "step": 4355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4584219257303597e-06,
+      "loss": 0.8928,
+      "step": 4356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.457450063327763e-06,
+      "loss": 0.7671,
+      "step": 4357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4564781517721903e-06,
+      "loss": 0.9408,
+      "step": 4358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4555061911680716e-06,
+      "loss": 1.1083,
+      "step": 4359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.454534181619846e-06,
+      "loss": 0.8458,
+      "step": 4360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4535621232319547e-06,
+      "loss": 0.8631,
+      "step": 4361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4525900161088463e-06,
+      "loss": 1.0551,
+      "step": 4362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.451617860354972e-06,
+      "loss": 1.0771,
+      "step": 4363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.450645656074792e-06,
+      "loss": 1.0237,
+      "step": 4364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4496734033727665e-06,
+      "loss": 0.8029,
+      "step": 4365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.448701102353368e-06,
+      "loss": 0.7932,
+      "step": 4366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.447728753121067e-06,
+      "loss": 0.9814,
+      "step": 4367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.446756355780344e-06,
+      "loss": 0.9383,
+      "step": 4368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4457839104356815e-06,
+      "loss": 0.8339,
+      "step": 4369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4448114171915707e-06,
+      "loss": 0.9622,
+      "step": 4370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4438388761525036e-06,
+      "loss": 1.0681,
+      "step": 4371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.442866287422982e-06,
+      "loss": 1.1735,
+      "step": 4372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4418936511075088e-06,
+      "loss": 0.7363,
+      "step": 4373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4409209673105954e-06,
+      "loss": 0.849,
+      "step": 4374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.439948236136754e-06,
+      "loss": 0.9085,
+      "step": 4375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.438975457690507e-06,
+      "loss": 1.0114,
+      "step": 4376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.438002632076378e-06,
+      "loss": 1.0775,
+      "step": 4377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.437029759398898e-06,
+      "loss": 0.921,
+      "step": 4378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4360568397626005e-06,
+      "loss": 0.845,
+      "step": 4379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.435083873272028e-06,
+      "loss": 0.8634,
+      "step": 4380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4341108600317234e-06,
+      "loss": 0.8485,
+      "step": 4381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4331378001462382e-06,
+      "loss": 1.0521,
+      "step": 4382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4321646937201276e-06,
+      "loss": 1.1715,
+      "step": 4383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.431191540857951e-06,
+      "loss": 0.9237,
+      "step": 4384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.430218341664274e-06,
+      "loss": 0.9567,
+      "step": 4385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4292450962436675e-06,
+      "loss": 0.9474,
+      "step": 4386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4282718047007047e-06,
+      "loss": 1.0085,
+      "step": 4387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4272984671399686e-06,
+      "loss": 0.97,
+      "step": 4388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.426325083666042e-06,
+      "loss": 0.9688,
+      "step": 4389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4253516543835158e-06,
+      "loss": 0.9195,
+      "step": 4390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4243781793969848e-06,
+      "loss": 0.8541,
+      "step": 4391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4234046588110483e-06,
+      "loss": 0.8813,
+      "step": 4392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.422431092730312e-06,
+      "loss": 0.849,
+      "step": 4393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.421457481259385e-06,
+      "loss": 0.776,
+      "step": 4394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.420483824502881e-06,
+      "loss": 0.8508,
+      "step": 4395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4195101225654196e-06,
+      "loss": 0.9736,
+      "step": 4396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4185363755516267e-06,
+      "loss": 0.907,
+      "step": 4397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4175625835661293e-06,
+      "loss": 0.8645,
+      "step": 4398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4165887467135627e-06,
+      "loss": 0.9434,
+      "step": 4399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4156148650985646e-06,
+      "loss": 1.1672,
+      "step": 4400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.414640938825779e-06,
+      "loss": 1.0569,
+      "step": 4401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.413666967999854e-06,
+      "loss": 0.7587,
+      "step": 4402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4126929527254433e-06,
+      "loss": 0.911,
+      "step": 4403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4117188931072036e-06,
+      "loss": 1.1367,
+      "step": 4404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4107447892497984e-06,
+      "loss": 0.8833,
+      "step": 4405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.409770641257895e-06,
+      "loss": 0.7307,
+      "step": 4406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4087964492361655e-06,
+      "loss": 0.979,
+      "step": 4407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4078222132892865e-06,
+      "loss": 1.0441,
+      "step": 4408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4068479335219403e-06,
+      "loss": 0.8838,
+      "step": 4409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4058736100388115e-06,
+      "loss": 1.0993,
+      "step": 4410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4048992429445936e-06,
+      "loss": 0.8351,
+      "step": 4411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40392483234398e-06,
+      "loss": 0.9206,
+      "step": 4412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.402950378341673e-06,
+      "loss": 0.8863,
+      "step": 4413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4019758810423752e-06,
+      "loss": 1.0478,
+      "step": 4414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4010013405507988e-06,
+      "loss": 0.8681,
+      "step": 4415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.400026756971657e-06,
+      "loss": 0.8088,
+      "step": 4416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3990521304096683e-06,
+      "loss": 0.975,
+      "step": 4417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3980774609695564e-06,
+      "loss": 0.8362,
+      "step": 4418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3971027487560505e-06,
+      "loss": 0.9554,
+      "step": 4419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3961279938738815e-06,
+      "loss": 0.8413,
+      "step": 4420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3951531964277887e-06,
+      "loss": 1.2945,
+      "step": 4421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3941783565225125e-06,
+      "loss": 1.0559,
+      "step": 4422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.393203474262801e-06,
+      "loss": 0.8369,
+      "step": 4423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3922285497534036e-06,
+      "loss": 1.0012,
+      "step": 4424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.391253583099076e-06,
+      "loss": 1.0519,
+      "step": 4425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3902785744045797e-06,
+      "loss": 1.0642,
+      "step": 4426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3893035237746778e-06,
+      "loss": 1.0671,
+      "step": 4427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.38832843131414e-06,
+      "loss": 0.9359,
+      "step": 4428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3873532971277405e-06,
+      "loss": 0.9278,
+      "step": 4429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.386378121320257e-06,
+      "loss": 1.0131,
+      "step": 4430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3854029039964714e-06,
+      "loss": 0.7947,
+      "step": 4431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.384427645261172e-06,
+      "loss": 0.9108,
+      "step": 4432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3834523452191483e-06,
+      "loss": 1.0395,
+      "step": 4433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3824770039751987e-06,
+      "loss": 0.9562,
+      "step": 4434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.381501621634122e-06,
+      "loss": 0.9206,
+      "step": 4435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380526198300723e-06,
+      "loss": 0.7484,
+      "step": 4436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.379550734079811e-06,
+      "loss": 0.8488,
+      "step": 4437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3785752290762003e-06,
+      "loss": 0.9386,
+      "step": 4438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.377599683394707e-06,
+      "loss": 0.8523,
+      "step": 4439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3766240971401565e-06,
+      "loss": 1.1292,
+      "step": 4440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.375648470417372e-06,
+      "loss": 1.0143,
+      "step": 4441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.374672803331187e-06,
+      "loss": 0.86,
+      "step": 4442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.373697095986436e-06,
+      "loss": 0.8848,
+      "step": 4443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.372721348487959e-06,
+      "loss": 0.9335,
+      "step": 4444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3717455609405987e-06,
+      "loss": 1.0374,
+      "step": 4445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3707697334492055e-06,
+      "loss": 1.0602,
+      "step": 4446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.36979386611863e-06,
+      "loss": 0.9581,
+      "step": 4447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.368817959053731e-06,
+      "loss": 0.8601,
+      "step": 4448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.367842012359367e-06,
+      "loss": 1.0682,
+      "step": 4449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.366866026140407e-06,
+      "loss": 0.9576,
+      "step": 4450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.365890000501717e-06,
+      "loss": 0.8445,
+      "step": 4451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3649139355481735e-06,
+      "loss": 0.8509,
+      "step": 4452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.363937831384653e-06,
+      "loss": 0.8749,
+      "step": 4453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3629616881160395e-06,
+      "loss": 0.8143,
+      "step": 4454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3619855058472176e-06,
+      "loss": 0.9471,
+      "step": 4455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3610092846830794e-06,
+      "loss": 0.8725,
+      "step": 4456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.360033024728519e-06,
+      "loss": 1.1142,
+      "step": 4457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.359056726088436e-06,
+      "loss": 0.9696,
+      "step": 4458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3580803888677333e-06,
+      "loss": 1.1277,
+      "step": 4459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.357104013171319e-06,
+      "loss": 0.9571,
+      "step": 4460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3561275991041023e-06,
+      "loss": 1.0401,
+      "step": 4461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.355151146771002e-06,
+      "loss": 0.8539,
+      "step": 4462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3541746562769357e-06,
+      "loss": 0.7241,
+      "step": 4463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.353198127726828e-06,
+      "loss": 0.9206,
+      "step": 4464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.352221561225607e-06,
+      "loss": 1.1001,
+      "step": 4465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3512449568782044e-06,
+      "loss": 1.0632,
+      "step": 4466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3502683147895554e-06,
+      "loss": 0.8763,
+      "step": 4467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.349291635064602e-06,
+      "loss": 1.1072,
+      "step": 4468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3483149178082863e-06,
+      "loss": 0.8463,
+      "step": 4469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.347338163125558e-06,
+      "loss": 1.061,
+      "step": 4470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.346361371121369e-06,
+      "loss": 0.7292,
+      "step": 4471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3453845419006755e-06,
+      "loss": 0.9062,
+      "step": 4472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.344407675568437e-06,
+      "loss": 0.9437,
+      "step": 4473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3434307722296195e-06,
+      "loss": 1.0264,
+      "step": 4474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3424538319891885e-06,
+      "loss": 0.828,
+      "step": 4475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.341476854952119e-06,
+      "loss": 0.7326,
+      "step": 4476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3404998412233852e-06,
+      "loss": 0.8716,
+      "step": 4477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3395227909079683e-06,
+      "loss": 0.9095,
+      "step": 4478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.338545704110851e-06,
+      "loss": 1.0631,
+      "step": 4479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.337568580937022e-06,
+      "loss": 0.8915,
+      "step": 4480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3365914214914736e-06,
+      "loss": 1.0079,
+      "step": 4481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3356142258792013e-06,
+      "loss": 0.9936,
+      "step": 4482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.334636994205203e-06,
+      "loss": 0.8889,
+      "step": 4483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3336597265744846e-06,
+      "loss": 1.1444,
+      "step": 4484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3326824230920515e-06,
+      "loss": 1.206,
+      "step": 4485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3317050838629166e-06,
+      "loss": 1.022,
+      "step": 4486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.330727708992094e-06,
+      "loss": 0.9801,
+      "step": 4487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3297502985846033e-06,
+      "loss": 0.8814,
+      "step": 4488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.328772852745465e-06,
+      "loss": 0.9248,
+      "step": 4489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.327795371579708e-06,
+      "loss": 1.0082,
+      "step": 4490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.326817855192362e-06,
+      "loss": 0.7582,
+      "step": 4491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.32584030368846e-06,
+      "loss": 1.1828,
+      "step": 4492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.324862717173041e-06,
+      "loss": 1.2155,
+      "step": 4493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3238850957511466e-06,
+      "loss": 1.0136,
+      "step": 4494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3229074395278218e-06,
+      "loss": 0.9137,
+      "step": 4495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.321929748608115e-06,
+      "loss": 0.9663,
+      "step": 4496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3209520230970803e-06,
+      "loss": 0.9723,
+      "step": 4497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3199742630997723e-06,
+      "loss": 0.8419,
+      "step": 4498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3189964687212543e-06,
+      "loss": 0.9955,
+      "step": 4499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3180186400665875e-06,
+      "loss": 0.9702,
+      "step": 4500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3170407772408413e-06,
+      "loss": 1.127,
+      "step": 4501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3160628803490855e-06,
+      "loss": 0.9288,
+      "step": 4502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3150849494963958e-06,
+      "loss": 0.8358,
+      "step": 4503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.314106984787851e-06,
+      "loss": 0.9415,
+      "step": 4504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3131289863285336e-06,
+      "loss": 0.9213,
+      "step": 4505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.312150954223528e-06,
+      "loss": 1.0247,
+      "step": 4506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.311172888577925e-06,
+      "loss": 0.9155,
+      "step": 4507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3101947894968176e-06,
+      "loss": 1.0404,
+      "step": 4508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3092166570853023e-06,
+      "loss": 1.0482,
+      "step": 4509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.308238491448479e-06,
+      "loss": 0.8855,
+      "step": 4510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3072602926914517e-06,
+      "loss": 0.9378,
+      "step": 4511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3062820609193274e-06,
+      "loss": 0.8873,
+      "step": 4512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3053037962372185e-06,
+      "loss": 0.9625,
+      "step": 4513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3043254987502372e-06,
+      "loss": 0.831,
+      "step": 4514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3033471685635045e-06,
+      "loss": 1.1034,
+      "step": 4515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3023688057821385e-06,
+      "loss": 0.9047,
+      "step": 4516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.301390410511267e-06,
+      "loss": 0.9079,
+      "step": 4517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3004119828560173e-06,
+      "loss": 0.9455,
+      "step": 4518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.299433522921521e-06,
+      "loss": 0.9396,
+      "step": 4519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2984550308129136e-06,
+      "loss": 0.8139,
+      "step": 4520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2974765066353356e-06,
+      "loss": 1.1382,
+      "step": 4521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.296497950493928e-06,
+      "loss": 1.0385,
+      "step": 4522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2955193624938365e-06,
+      "loss": 1.0319,
+      "step": 4523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.29454074274021e-06,
+      "loss": 1.2297,
+      "step": 4524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.293562091338203e-06,
+      "loss": 0.8819,
+      "step": 4525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.292583408392969e-06,
+      "loss": 1.092,
+      "step": 4526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2916046940096696e-06,
+      "loss": 0.8909,
+      "step": 4527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.290625948293466e-06,
+      "loss": 0.9434,
+      "step": 4528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.289647171349525e-06,
+      "loss": 1.0408,
+      "step": 4529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2886683632830162e-06,
+      "loss": 0.996,
+      "step": 4530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.287689524199113e-06,
+      "loss": 0.8603,
+      "step": 4531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2867106542029895e-06,
+      "loss": 0.8225,
+      "step": 4532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.285731753399828e-06,
+      "loss": 0.9189,
+      "step": 4533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2847528218948096e-06,
+      "loss": 0.9443,
+      "step": 4534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.283773859793121e-06,
+      "loss": 0.9935,
+      "step": 4535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2827948671999504e-06,
+      "loss": 0.9593,
+      "step": 4536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.281815844220493e-06,
+      "loss": 0.9621,
+      "step": 4537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2808367909599423e-06,
+      "loss": 0.8421,
+      "step": 4538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.279857707523499e-06,
+      "loss": 0.9005,
+      "step": 4539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2788785940163644e-06,
+      "loss": 1.0368,
+      "step": 4540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.277899450543745e-06,
+      "loss": 0.8788,
+      "step": 4541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2769202772108495e-06,
+      "loss": 0.8493,
+      "step": 4542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2759410741228905e-06,
+      "loss": 0.7223,
+      "step": 4543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.274961841385082e-06,
+      "loss": 0.8452,
+      "step": 4544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.273982579102644e-06,
+      "loss": 0.9816,
+      "step": 4545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2730032873807964e-06,
+      "loss": 1.0151,
+      "step": 4546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2720239663247656e-06,
+      "loss": 0.8091,
+      "step": 4547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.271044616039779e-06,
+      "loss": 0.9972,
+      "step": 4548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2700652366310683e-06,
+      "loss": 1.3231,
+      "step": 4549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.269085828203866e-06,
+      "loss": 0.9313,
+      "step": 4550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.268106390863412e-06,
+      "loss": 0.9414,
+      "step": 4551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2671269247149444e-06,
+      "loss": 0.8726,
+      "step": 4552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.266147429863708e-06,
+      "loss": 0.9656,
+      "step": 4553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2651679064149495e-06,
+      "loss": 0.9921,
+      "step": 4554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2641883544739182e-06,
+      "loss": 0.8813,
+      "step": 4555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.263208774145867e-06,
+      "loss": 0.9237,
+      "step": 4556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.262229165536053e-06,
+      "loss": 0.9373,
+      "step": 4557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.261249528749732e-06,
+      "loss": 0.9263,
+      "step": 4558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2602698638921684e-06,
+      "loss": 1.2797,
+      "step": 4559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2592901710686265e-06,
+      "loss": 0.9703,
+      "step": 4560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.258310450384375e-06,
+      "loss": 0.9257,
+      "step": 4561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2573307019446836e-06,
+      "loss": 1.0086,
+      "step": 4562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2563509258548257e-06,
+      "loss": 0.7805,
+      "step": 4563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2553711222200806e-06,
+      "loss": 1.0459,
+      "step": 4564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2543912911457257e-06,
+      "loss": 0.8289,
+      "step": 4565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2534114327370453e-06,
+      "loss": 1.0702,
+      "step": 4566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2524315470993237e-06,
+      "loss": 0.849,
+      "step": 4567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.251451634337851e-06,
+      "loss": 1.0459,
+      "step": 4568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2504716945579182e-06,
+      "loss": 0.9575,
+      "step": 4569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.24949172786482e-06,
+      "loss": 1.195,
+      "step": 4570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2485117343638526e-06,
+      "loss": 0.8492,
+      "step": 4571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2475317141603174e-06,
+      "loss": 0.7998,
+      "step": 4572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.246551667359517e-06,
+      "loss": 0.9392,
+      "step": 4573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.245571594066758e-06,
+      "loss": 0.9397,
+      "step": 4574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2445914943873483e-06,
+      "loss": 1.1165,
+      "step": 4575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.243611368426601e-06,
+      "loss": 0.9677,
+      "step": 4576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.242631216289828e-06,
+      "loss": 1.1085,
+      "step": 4577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2416510380823493e-06,
+      "loss": 0.8094,
+      "step": 4578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2406708339094827e-06,
+      "loss": 1.2573,
+      "step": 4579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.239690603876553e-06,
+      "loss": 1.3212,
+      "step": 4580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.238710348088885e-06,
+      "loss": 0.742,
+      "step": 4581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2377300666518066e-06,
+      "loss": 1.0205,
+      "step": 4582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.23674975967065e-06,
+      "loss": 0.9621,
+      "step": 4583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.235769427250749e-06,
+      "loss": 0.7429,
+      "step": 4584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.234789069497439e-06,
+      "loss": 1.2789,
+      "step": 4585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2338086865160606e-06,
+      "loss": 0.9792,
+      "step": 4586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.232828278411955e-06,
+      "loss": 1.1268,
+      "step": 4587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2318478452904687e-06,
+      "loss": 1.0391,
+      "step": 4588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2308673872569473e-06,
+      "loss": 0.8298,
+      "step": 4589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2298869044167416e-06,
+      "loss": 1.0151,
+      "step": 4590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2289063968752044e-06,
+      "loss": 0.8503,
+      "step": 4591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2279258647376914e-06,
+      "loss": 1.0696,
+      "step": 4592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.22694530810956e-06,
+      "loss": 0.9313,
+      "step": 4593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.225964727096172e-06,
+      "loss": 0.9565,
+      "step": 4594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2249841218028893e-06,
+      "loss": 0.8305,
+      "step": 4595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.22400349233508e-06,
+      "loss": 0.7221,
+      "step": 4596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.223022838798111e-06,
+      "loss": 0.8977,
+      "step": 4597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2220421612973536e-06,
+      "loss": 0.9416,
+      "step": 4598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2210614599381818e-06,
+      "loss": 0.862,
+      "step": 4599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2200807348259724e-06,
+      "loss": 0.8866,
+      "step": 4600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2190999860661033e-06,
+      "loss": 0.9634,
+      "step": 4601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2181192137639567e-06,
+      "loss": 0.8615,
+      "step": 4602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2171384180249154e-06,
+      "loss": 1.0237,
+      "step": 4603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.216157598954367e-06,
+      "loss": 0.942,
+      "step": 4604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2151767566577e-06,
+      "loss": 0.7564,
+      "step": 4605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.214195891240306e-06,
+      "loss": 1.0299,
+      "step": 4606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.213215002807579e-06,
+      "loss": 0.9011,
+      "step": 4607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2122340914649152e-06,
+      "loss": 0.9008,
+      "step": 4608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2112531573177126e-06,
+      "loss": 0.8336,
+      "step": 4609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2102722004713738e-06,
+      "loss": 0.965,
+      "step": 4610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.209291221031302e-06,
+      "loss": 0.8471,
+      "step": 4611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2083102191029037e-06,
+      "loss": 0.8491,
+      "step": 4612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.207329194791587e-06,
+      "loss": 0.8146,
+      "step": 4613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2063481482027626e-06,
+      "loss": 0.878,
+      "step": 4614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2053670794418454e-06,
+      "loss": 1.0154,
+      "step": 4615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.20438598861425e-06,
+      "loss": 0.9615,
+      "step": 4616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.203404875825394e-06,
+      "loss": 1.0437,
+      "step": 4617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2024237411806993e-06,
+      "loss": 1.2525,
+      "step": 4618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.201442584785587e-06,
+      "loss": 0.8439,
+      "step": 4619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2004614067454844e-06,
+      "loss": 0.9196,
+      "step": 4620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1994802071658176e-06,
+      "loss": 0.8912,
+      "step": 4621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1984989861520173e-06,
+      "loss": 0.8749,
+      "step": 4622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.197517743809514e-06,
+      "loss": 0.9248,
+      "step": 4623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1965364802437443e-06,
+      "loss": 0.932,
+      "step": 4624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.195555195560143e-06,
+      "loss": 0.7811,
+      "step": 4625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1945738898641497e-06,
+      "loss": 1.1132,
+      "step": 4626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.193592563261207e-06,
+      "loss": 0.914,
+      "step": 4627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1926112158567563e-06,
+      "loss": 1.0554,
+      "step": 4628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.191629847756245e-06,
+      "loss": 0.7158,
+      "step": 4629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.190648459065119e-06,
+      "loss": 1.035,
+      "step": 4630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.189667049888831e-06,
+      "loss": 0.8726,
+      "step": 4631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.188685620332831e-06,
+      "loss": 1.157,
+      "step": 4632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1877041705025752e-06,
+      "loss": 0.9441,
+      "step": 4633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1867227005035193e-06,
+      "loss": 0.9222,
+      "step": 4634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1857412104411234e-06,
+      "loss": 1.0051,
+      "step": 4635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.184759700420846e-06,
+      "loss": 1.0019,
+      "step": 4636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1837781705481537e-06,
+      "loss": 0.9522,
+      "step": 4637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1827966209285085e-06,
+      "loss": 0.9113,
+      "step": 4638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1818150516673813e-06,
+      "loss": 0.853,
+      "step": 4639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1808334628702384e-06,
+      "loss": 1.0399,
+      "step": 4640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1798518546425536e-06,
+      "loss": 1.086,
+      "step": 4641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1788702270897985e-06,
+      "loss": 1.1529,
+      "step": 4642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1778885803174516e-06,
+      "loss": 0.8409,
+      "step": 4643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.176906914430989e-06,
+      "loss": 0.8463,
+      "step": 4644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.175925229535891e-06,
+      "loss": 0.7614,
+      "step": 4645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1749435257376396e-06,
+      "loss": 0.8355,
+      "step": 4646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1739618031417193e-06,
+      "loss": 0.8856,
+      "step": 4647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.172980061853615e-06,
+      "loss": 0.7738,
+      "step": 4648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.171998301978816e-06,
+      "loss": 1.0726,
+      "step": 4649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.171016523622811e-06,
+      "loss": 0.8953,
+      "step": 4650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.170034726891094e-06,
+      "loss": 0.8518,
+      "step": 4651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.169052911889156e-06,
+      "loss": 0.8041,
+      "step": 4652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1680710787224968e-06,
+      "loss": 0.7901,
+      "step": 4653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1670892274966105e-06,
+      "loss": 1.113,
+      "step": 4654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1661073583169994e-06,
+      "loss": 0.9278,
+      "step": 4655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1651254712891637e-06,
+      "loss": 1.2456,
+      "step": 4656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.164143566518609e-06,
+      "loss": 0.8414,
+      "step": 4657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1631616441108388e-06,
+      "loss": 0.8997,
+      "step": 4658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1621797041713624e-06,
+      "loss": 0.8359,
+      "step": 4659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1611977468056884e-06,
+      "loss": 1.0168,
+      "step": 4660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.160215772119327e-06,
+      "loss": 0.8203,
+      "step": 4661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1592337802177937e-06,
+      "loss": 1.0731,
+      "step": 4662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1582517712066e-06,
+      "loss": 0.9924,
+      "step": 4663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.157269745191267e-06,
+      "loss": 0.9955,
+      "step": 4664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.15628770227731e-06,
+      "loss": 0.7654,
+      "step": 4665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1553056425702506e-06,
+      "loss": 1.0389,
+      "step": 4666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.154323566175611e-06,
+      "loss": 0.8118,
+      "step": 4667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1533414731989155e-06,
+      "loss": 0.8972,
+      "step": 4668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.152359363745689e-06,
+      "loss": 0.8256,
+      "step": 4669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.151377237921461e-06,
+      "loss": 1.015,
+      "step": 4670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.150395095831758e-06,
+      "loss": 0.7969,
+      "step": 4671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1494129375821138e-06,
+      "loss": 1.1691,
+      "step": 4672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1484307632780594e-06,
+      "loss": 0.7663,
+      "step": 4673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.147448573025131e-06,
+      "loss": 0.8382,
+      "step": 4674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.146466366928863e-06,
+      "loss": 0.8289,
+      "step": 4675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1454841450947955e-06,
+      "loss": 0.7489,
+      "step": 4676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1445019076284654e-06,
+      "loss": 0.8735,
+      "step": 4677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.143519654635417e-06,
+      "loss": 0.8679,
+      "step": 4678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.142537386221191e-06,
+      "loss": 0.9369,
+      "step": 4679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1415551024913334e-06,
+      "loss": 1.1702,
+      "step": 4680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1405728035513903e-06,
+      "loss": 0.9207,
+      "step": 4681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1395904895069095e-06,
+      "loss": 0.8999,
+      "step": 4682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1386081604634404e-06,
+      "loss": 0.7872,
+      "step": 4683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1376258165265346e-06,
+      "loss": 0.9259,
+      "step": 4684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1366434578017444e-06,
+      "loss": 0.8587,
+      "step": 4685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1356610843946244e-06,
+      "loss": 0.7489,
+      "step": 4686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1346786964107297e-06,
+      "loss": 0.8976,
+      "step": 4687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1336962939556206e-06,
+      "loss": 0.8747,
+      "step": 4688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1327138771348527e-06,
+      "loss": 0.9891,
+      "step": 4689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.131731446053989e-06,
+      "loss": 0.9875,
+      "step": 4690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1307490008185902e-06,
+      "loss": 1.2291,
+      "step": 4691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1297665415342213e-06,
+      "loss": 1.2263,
+      "step": 4692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.128784068306446e-06,
+      "loss": 0.8335,
+      "step": 4693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1278015812408323e-06,
+      "loss": 1.0208,
+      "step": 4694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.126819080442948e-06,
+      "loss": 1.2896,
+      "step": 4695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.125836566018362e-06,
+      "loss": 0.9659,
+      "step": 4696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1248540380726456e-06,
+      "loss": 0.8773,
+      "step": 4697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1238714967113732e-06,
+      "loss": 0.8349,
+      "step": 4698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1228889420401156e-06,
+      "loss": 0.8122,
+      "step": 4699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1219063741644508e-06,
+      "loss": 0.959,
+      "step": 4700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.120923793189955e-06,
+      "loss": 0.8049,
+      "step": 4701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.119941199222206e-06,
+      "loss": 1.2085,
+      "step": 4702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1189585923667833e-06,
+      "loss": 0.8594,
+      "step": 4703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1179759727292692e-06,
+      "loss": 0.9976,
+      "step": 4704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1169933404152443e-06,
+      "loss": 0.7043,
+      "step": 4705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.116010695530294e-06,
+      "loss": 0.7456,
+      "step": 4706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1150280381800026e-06,
+      "loss": 0.7913,
+      "step": 4707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.114045368469957e-06,
+      "loss": 0.8431,
+      "step": 4708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1130626865057437e-06,
+      "loss": 0.8438,
+      "step": 4709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.112079992392954e-06,
+      "loss": 1.0413,
+      "step": 4710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.111097286237176e-06,
+      "loss": 0.9316,
+      "step": 4711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1101145681440036e-06,
+      "loss": 0.9792,
+      "step": 4712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.109131838219028e-06,
+      "loss": 0.9227,
+      "step": 4713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1081490965678447e-06,
+      "loss": 1.0233,
+      "step": 4714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.107166343296048e-06,
+      "loss": 1.0246,
+      "step": 4715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1061835785092357e-06,
+      "loss": 0.7918,
+      "step": 4716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.105200802313005e-06,
+      "loss": 1.0635,
+      "step": 4717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1042180148129566e-06,
+      "loss": 0.8636,
+      "step": 4718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.103235216114689e-06,
+      "loss": 1.1091,
+      "step": 4719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.102252406323806e-06,
+      "loss": 1.1483,
+      "step": 4720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1012695855459076e-06,
+      "loss": 0.7466,
+      "step": 4721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1002867538866007e-06,
+      "loss": 0.9469,
+      "step": 4722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0993039114514885e-06,
+      "loss": 0.9171,
+      "step": 4723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0983210583461794e-06,
+      "loss": 0.7735,
+      "step": 4724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0973381946762777e-06,
+      "loss": 0.6681,
+      "step": 4725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0963553205473957e-06,
+      "loss": 0.8426,
+      "step": 4726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0953724360651402e-06,
+      "loss": 0.8872,
+      "step": 4727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0943895413351244e-06,
+      "loss": 1.2329,
+      "step": 4728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0934066364629575e-06,
+      "loss": 0.994,
+      "step": 4729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0924237215542554e-06,
+      "loss": 0.868,
+      "step": 4730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0914407967146306e-06,
+      "loss": 1.073,
+      "step": 4731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.090457862049699e-06,
+      "loss": 1.0568,
+      "step": 4732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0894749176650756e-06,
+      "loss": 1.0508,
+      "step": 4733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.08849196366638e-06,
+      "loss": 0.8911,
+      "step": 4734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0875090001592283e-06,
+      "loss": 0.9374,
+      "step": 4735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0865260272492412e-06,
+      "loss": 0.8447,
+      "step": 4736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.085543045042038e-06,
+      "loss": 0.9113,
+      "step": 4737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.084560053643242e-06,
+      "loss": 0.6478,
+      "step": 4738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.083577053158473e-06,
+      "loss": 1.1263,
+      "step": 4739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.082594043693356e-06,
+      "loss": 0.749,
+      "step": 4740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.081611025353515e-06,
+      "loss": 1.0218,
+      "step": 4741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.080627998244576e-06,
+      "loss": 0.9243,
+      "step": 4742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0796449624721632e-06,
+      "loss": 0.7994,
+      "step": 4743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.078661918141905e-06,
+      "loss": 0.8497,
+      "step": 4744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.07767886535943e-06,
+      "loss": 0.9142,
+      "step": 4745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.076695804230366e-06,
+      "loss": 0.7573,
+      "step": 4746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0757127348603427e-06,
+      "loss": 0.9221,
+      "step": 4747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0747296573549922e-06,
+      "loss": 1.0157,
+      "step": 4748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.073746571819945e-06,
+      "loss": 0.8952,
+      "step": 4749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0727634783608347e-06,
+      "loss": 1.0426,
+      "step": 4750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.071780377083293e-06,
+      "loss": 1.0076,
+      "step": 4751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0707972680929552e-06,
+      "loss": 0.8209,
+      "step": 4752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.069814151495456e-06,
+      "loss": 0.9645,
+      "step": 4753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0688310273964315e-06,
+      "loss": 1.032,
+      "step": 4754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.067847895901517e-06,
+      "loss": 0.8727,
+      "step": 4755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.066864757116352e-06,
+      "loss": 0.9957,
+      "step": 4756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0658816111465727e-06,
+      "loss": 0.9256,
+      "step": 4757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.064898458097819e-06,
+      "loss": 0.8241,
+      "step": 4758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0639152980757318e-06,
+      "loss": 1.0442,
+      "step": 4759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0629321311859493e-06,
+      "loss": 0.8581,
+      "step": 4760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0619489575341147e-06,
+      "loss": 0.7346,
+      "step": 4761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.060965777225868e-06,
+      "loss": 0.8647,
+      "step": 4762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.059982590366854e-06,
+      "loss": 0.8274,
+      "step": 4763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0589993970627136e-06,
+      "loss": 1.1737,
+      "step": 4764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.058016197419093e-06,
+      "loss": 1.129,
+      "step": 4765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0570329915416357e-06,
+      "loss": 1.0323,
+      "step": 4766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0560497795359884e-06,
+      "loss": 0.8479,
+      "step": 4767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0550665615077953e-06,
+      "loss": 1.0665,
+      "step": 4768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.054083337562705e-06,
+      "loss": 0.7783,
+      "step": 4769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.053100107806363e-06,
+      "loss": 0.7721,
+      "step": 4770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0521168723444194e-06,
+      "loss": 1.1013,
+      "step": 4771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0511336312825207e-06,
+      "loss": 0.8508,
+      "step": 4772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.050150384726317e-06,
+      "loss": 0.8233,
+      "step": 4773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.049167132781458e-06,
+      "loss": 0.8748,
+      "step": 4774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.048183875553595e-06,
+      "loss": 1.1237,
+      "step": 4775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.047200613148377e-06,
+      "loss": 1.0575,
+      "step": 4776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.046217345671457e-06,
+      "loss": 0.861,
+      "step": 4777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0452340732284858e-06,
+      "loss": 1.0142,
+      "step": 4778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0442507959251172e-06,
+      "loss": 0.7891,
+      "step": 4779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.043267513867004e-06,
+      "loss": 0.8309,
+      "step": 4780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0422842271597997e-06,
+      "loss": 0.949,
+      "step": 4781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.041300935909157e-06,
+      "loss": 0.9989,
+      "step": 4782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0403176402207325e-06,
+      "loss": 0.8259,
+      "step": 4783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0393343402001805e-06,
+      "loss": 0.9012,
+      "step": 4784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.038351035953156e-06,
+      "loss": 1.0597,
+      "step": 4785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.037367727585316e-06,
+      "loss": 0.9383,
+      "step": 4786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0363844152023165e-06,
+      "loss": 0.8889,
+      "step": 4787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0354010989098135e-06,
+      "loss": 0.8427,
+      "step": 4788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.034417778813465e-06,
+      "loss": 0.9037,
+      "step": 4789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0334344550189284e-06,
+      "loss": 0.8397,
+      "step": 4790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.032451127631863e-06,
+      "loss": 1.0384,
+      "step": 4791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.031467796757925e-06,
+      "loss": 0.9856,
+      "step": 4792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0304844625027757e-06,
+      "loss": 0.9437,
+      "step": 4793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.029501124972072e-06,
+      "loss": 1.0169,
+      "step": 4794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0285177842714758e-06,
+      "loss": 0.7272,
+      "step": 4795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.027534440506644e-06,
+      "loss": 0.8877,
+      "step": 4796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0265510937832402e-06,
+      "loss": 1.0549,
+      "step": 4797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.025567744206922e-06,
+      "loss": 0.8514,
+      "step": 4798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.024584391883353e-06,
+      "loss": 0.7945,
+      "step": 4799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0236010369181914e-06,
+      "loss": 0.7633,
+      "step": 4800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0226176794171006e-06,
+      "loss": 0.9479,
+      "step": 4801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0216343194857417e-06,
+      "loss": 0.8198,
+      "step": 4802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0206509572297768e-06,
+      "loss": 0.71,
+      "step": 4803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.019667592754868e-06,
+      "loss": 0.8367,
+      "step": 4804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.018684226166678e-06,
+      "loss": 0.9796,
+      "step": 4805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0177008575708684e-06,
+      "loss": 0.7916,
+      "step": 4806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.016717487073104e-06,
+      "loss": 1.0066,
+      "step": 4807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.015734114779046e-06,
+      "loss": 0.8551,
+      "step": 4808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0147507407943593e-06,
+      "loss": 1.0087,
+      "step": 4809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.013767365224706e-06,
+      "loss": 0.8467,
+      "step": 4810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.01278398817575e-06,
+      "loss": 0.7397,
+      "step": 4811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0118006097531555e-06,
+      "loss": 0.801,
+      "step": 4812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.010817230062587e-06,
+      "loss": 1.0023,
+      "step": 4813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.009833849209707e-06,
+      "loss": 0.8045,
+      "step": 4814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0088504673001814e-06,
+      "loss": 0.8411,
+      "step": 4815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0078670844396733e-06,
+      "loss": 0.7356,
+      "step": 4816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.006883700733848e-06,
+      "loss": 0.8737,
+      "step": 4817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0059003162883683e-06,
+      "loss": 0.7862,
+      "step": 4818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.004916931208901e-06,
+      "loss": 0.8311,
+      "step": 4819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.00393354560111e-06,
+      "loss": 1.0427,
+      "step": 4820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0029501595706593e-06,
+      "loss": 0.9415,
+      "step": 4821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0019667732232146e-06,
+      "loss": 0.5958,
+      "step": 4822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.00098338666444e-06,
+      "loss": 0.9194,
+      "step": 4823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3e-06,
+      "loss": 0.878,
+      "step": 4824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.999016613335561e-06,
+      "loss": 0.9068,
+      "step": 4825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9980332267767855e-06,
+      "loss": 0.6651,
+      "step": 4826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.997049840429341e-06,
+      "loss": 0.8941,
+      "step": 4827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9960664543988908e-06,
+      "loss": 0.9203,
+      "step": 4828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9950830687911e-06,
+      "loss": 0.8563,
+      "step": 4829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9940996837116314e-06,
+      "loss": 0.8812,
+      "step": 4830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.993116299266153e-06,
+      "loss": 1.1905,
+      "step": 4831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9921329155603277e-06,
+      "loss": 0.8708,
+      "step": 4832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.99114953269982e-06,
+      "loss": 0.8801,
+      "step": 4833
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.990166150790293e-06,
+      "loss": 0.7808,
+      "step": 4834
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9891827699374137e-06,
+      "loss": 0.8423,
+      "step": 4835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.988199390246845e-06,
+      "loss": 0.9464,
+      "step": 4836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.987216011824251e-06,
+      "loss": 0.9671,
+      "step": 4837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9862326347752942e-06,
+      "loss": 1.0859,
+      "step": 4838
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9852492592056417e-06,
+      "loss": 0.9065,
+      "step": 4839
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9842658852209542e-06,
+      "loss": 1.1341,
+      "step": 4840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9832825129268975e-06,
+      "loss": 0.9484,
+      "step": 4841
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9822991424291313e-06,
+      "loss": 0.8426,
+      "step": 4842
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9813157738333225e-06,
+      "loss": 1.0961,
+      "step": 4843
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9803324072451326e-06,
+      "loss": 1.1208,
+      "step": 4844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9793490427702242e-06,
+      "loss": 0.8845,
+      "step": 4845
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.978365680514258e-06,
+      "loss": 0.8944,
+      "step": 4846
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9773823205828996e-06,
+      "loss": 0.9701,
+      "step": 4847
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9763989630818096e-06,
+      "loss": 0.7674,
+      "step": 4848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9754156081166485e-06,
+      "loss": 0.9616,
+      "step": 4849
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9744322557930774e-06,
+      "loss": 1.1004,
+      "step": 4850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9734489062167603e-06,
+      "loss": 1.137,
+      "step": 4851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.972465559493356e-06,
+      "loss": 0.9499,
+      "step": 4852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9714822157285256e-06,
+      "loss": 0.9114,
+      "step": 4853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9704988750279275e-06,
+      "loss": 1.1018,
+      "step": 4854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.969515537497225e-06,
+      "loss": 0.8753,
+      "step": 4855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9685322032420753e-06,
+      "loss": 0.6706,
+      "step": 4856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9675488723681385e-06,
+      "loss": 0.9373,
+      "step": 4857
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9665655449810713e-06,
+      "loss": 0.8715,
+      "step": 4858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9655822211865355e-06,
+      "loss": 0.9604,
+      "step": 4859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9645989010901875e-06,
+      "loss": 0.8088,
+      "step": 4860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9636155847976853e-06,
+      "loss": 0.863,
+      "step": 4861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.962632272414684e-06,
+      "loss": 0.7965,
+      "step": 4862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.961648964046844e-06,
+      "loss": 0.964,
+      "step": 4863
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.96066565979982e-06,
+      "loss": 0.7843,
+      "step": 4864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9596823597792685e-06,
+      "loss": 0.8575,
+      "step": 4865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9586990640908427e-06,
+      "loss": 0.7679,
+      "step": 4866
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9577157728402013e-06,
+      "loss": 0.7675,
+      "step": 4867
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9567324861329968e-06,
+      "loss": 0.9986,
+      "step": 4868
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9557492040748833e-06,
+      "loss": 0.8811,
+      "step": 4869
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.954765926771514e-06,
+      "loss": 0.935,
+      "step": 4870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9537826543285437e-06,
+      "loss": 0.9139,
+      "step": 4871
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.952799386851624e-06,
+      "loss": 0.8914,
+      "step": 4872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9518161244464066e-06,
+      "loss": 0.943,
+      "step": 4873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.950832867218542e-06,
+      "loss": 1.014,
+      "step": 4874
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9498496152736833e-06,
+      "loss": 0.8065,
+      "step": 4875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9488663687174803e-06,
+      "loss": 0.7588,
+      "step": 4876
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.947883127655582e-06,
+      "loss": 1.0231,
+      "step": 4877
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9468998921936365e-06,
+      "loss": 1.163,
+      "step": 4878
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9459166624372953e-06,
+      "loss": 1.0346,
+      "step": 4879
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.944933438492205e-06,
+      "loss": 1.119,
+      "step": 4880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9439502204640126e-06,
+      "loss": 0.8924,
+      "step": 4881
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9429670084583636e-06,
+      "loss": 0.8066,
+      "step": 4882
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9419838025809073e-06,
+      "loss": 1.0171,
+      "step": 4883
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.941000602937287e-06,
+      "loss": 0.805,
+      "step": 4884
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.940017409633148e-06,
+      "loss": 1.0707,
+      "step": 4885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.939034222774132e-06,
+      "loss": 0.8886,
+      "step": 4886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9380510424658863e-06,
+      "loss": 0.7881,
+      "step": 4887
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9370678688140513e-06,
+      "loss": 1.1671,
+      "step": 4888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9360847019242697e-06,
+      "loss": 0.7981,
+      "step": 4889
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9351015419021808e-06,
+      "loss": 0.9717,
+      "step": 4890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.934118388853428e-06,
+      "loss": 1.052,
+      "step": 4891
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9331352428836494e-06,
+      "loss": 0.8204,
+      "step": 4892
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.932152104098483e-06,
+      "loss": 1.1149,
+      "step": 4893
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9311689726035695e-06,
+      "loss": 0.7495,
+      "step": 4894
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.930185848504545e-06,
+      "loss": 0.7219,
+      "step": 4895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9292027319070458e-06,
+      "loss": 1.0186,
+      "step": 4896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9282196229167073e-06,
+      "loss": 0.8162,
+      "step": 4897
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.927236521639166e-06,
+      "loss": 1.0066,
+      "step": 4898
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9262534281800554e-06,
+      "loss": 1.0199,
+      "step": 4899
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9252703426450088e-06,
+      "loss": 1.0815,
+      "step": 4900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9242872651396566e-06,
+      "loss": 0.7699,
+      "step": 4901
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9233041957696346e-06,
+      "loss": 1.0586,
+      "step": 4902
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.922321134640571e-06,
+      "loss": 0.8827,
+      "step": 4903
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.921338081858096e-06,
+      "loss": 0.8453,
+      "step": 4904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.920355037527837e-06,
+      "loss": 0.9465,
+      "step": 4905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9193720017554246e-06,
+      "loss": 0.8433,
+      "step": 4906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.918388974646485e-06,
+      "loss": 0.8208,
+      "step": 4907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9174059563066444e-06,
+      "loss": 0.8775,
+      "step": 4908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9164229468415265e-06,
+      "loss": 0.8866,
+      "step": 4909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9154399463567583e-06,
+      "loss": 0.845,
+      "step": 4910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.914456954957962e-06,
+      "loss": 1.0267,
+      "step": 4911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9134739727507598e-06,
+      "loss": 0.8045,
+      "step": 4912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9124909998407714e-06,
+      "loss": 0.8161,
+      "step": 4913
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9115080363336208e-06,
+      "loss": 0.9978,
+      "step": 4914
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9105250823349245e-06,
+      "loss": 0.9319,
+      "step": 4915
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9095421379503025e-06,
+      "loss": 0.7482,
+      "step": 4916
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9085592032853695e-06,
+      "loss": 0.8058,
+      "step": 4917
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.907576278445745e-06,
+      "loss": 0.855,
+      "step": 4918
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.906593363537043e-06,
+      "loss": 1.1925,
+      "step": 4919
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9056104586648775e-06,
+      "loss": 1.0621,
+      "step": 4920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.90462756393486e-06,
+      "loss": 1.0863,
+      "step": 4921
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.903644679452605e-06,
+      "loss": 0.9526,
+      "step": 4922
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9026618053237224e-06,
+      "loss": 0.7859,
+      "step": 4923
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.901678941653822e-06,
+      "loss": 0.851,
+      "step": 4924
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.900696088548511e-06,
+      "loss": 1.093,
+      "step": 4925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8997132461133994e-06,
+      "loss": 0.8763,
+      "step": 4926
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8987304144540925e-06,
+      "loss": 0.8493,
+      "step": 4927
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8977475936761953e-06,
+      "loss": 1.069,
+      "step": 4928
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8967647838853114e-06,
+      "loss": 0.6994,
+      "step": 4929
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.895781985187044e-06,
+      "loss": 0.6432,
+      "step": 4930
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.894799197686995e-06,
+      "loss": 1.2045,
+      "step": 4931
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8938164214907653e-06,
+      "loss": 0.905,
+      "step": 4932
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8928336567039527e-06,
+      "loss": 0.999,
+      "step": 4933
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.891850903432156e-06,
+      "loss": 0.9282,
+      "step": 4934
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8908681617809726e-06,
+      "loss": 0.9901,
+      "step": 4935
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8898854318559978e-06,
+      "loss": 1.0452,
+      "step": 4936
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8889027137628243e-06,
+      "loss": 0.8073,
+      "step": 4937
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8879200076070467e-06,
+      "loss": 0.976,
+      "step": 4938
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8869373134942564e-06,
+      "loss": 0.9426,
+      "step": 4939
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8859546315300443e-06,
+      "loss": 0.9028,
+      "step": 4940
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.884971961819998e-06,
+      "loss": 0.8477,
+      "step": 4941
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.883989304469706e-06,
+      "loss": 0.9899,
+      "step": 4942
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.883006659584756e-06,
+      "loss": 0.9921,
+      "step": 4943
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.882024027270732e-06,
+      "loss": 1.0557,
+      "step": 4944
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8810414076332173e-06,
+      "loss": 1.0046,
+      "step": 4945
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8800588007777946e-06,
+      "loss": 0.8501,
+      "step": 4946
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.879076206810046e-06,
+      "loss": 0.8767,
+      "step": 4947
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.87809362583555e-06,
+      "loss": 0.8047,
+      "step": 4948
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8771110579598846e-06,
+      "loss": 0.8906,
+      "step": 4949
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8761285032886278e-06,
+      "loss": 0.8758,
+      "step": 4950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8751459619273545e-06,
+      "loss": 0.9637,
+      "step": 4951
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8741634339816394e-06,
+      "loss": 0.8348,
+      "step": 4952
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.873180919557053e-06,
+      "loss": 0.8219,
+      "step": 4953
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8721984187591683e-06,
+      "loss": 0.9589,
+      "step": 4954
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.871215931693555e-06,
+      "loss": 0.8255,
+      "step": 4955
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.87023345846578e-06,
+      "loss": 0.9057,
+      "step": 4956
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8692509991814103e-06,
+      "loss": 0.8457,
+      "step": 4957
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8682685539460116e-06,
+      "loss": 0.9313,
+      "step": 4958
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.867286122865148e-06,
+      "loss": 0.8886,
+      "step": 4959
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8663037060443804e-06,
+      "loss": 0.815,
+      "step": 4960
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8653213035892704e-06,
+      "loss": 0.7729,
+      "step": 4961
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8643389156053762e-06,
+      "loss": 0.9942,
+      "step": 4962
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8633565421982566e-06,
+      "loss": 0.7948,
+      "step": 4963
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.862374183473466e-06,
+      "loss": 0.7796,
+      "step": 4964
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.86139183953656e-06,
+      "loss": 0.997,
+      "step": 4965
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8604095104930906e-06,
+      "loss": 0.731,
+      "step": 4966
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8594271964486103e-06,
+      "loss": 0.788,
+      "step": 4967
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8584448975086667e-06,
+      "loss": 0.9927,
+      "step": 4968
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8574626137788095e-06,
+      "loss": 1.0731,
+      "step": 4969
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8564803453645836e-06,
+      "loss": 0.9894,
+      "step": 4970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.855498092371535e-06,
+      "loss": 0.8738,
+      "step": 4971
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8545158549052055e-06,
+      "loss": 0.8113,
+      "step": 4972
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8535336330711374e-06,
+      "loss": 0.8419,
+      "step": 4973
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8525514269748695e-06,
+      "loss": 1.1289,
+      "step": 4974
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8515692367219407e-06,
+      "loss": 0.8231,
+      "step": 4975
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8505870624178864e-06,
+      "loss": 0.9375,
+      "step": 4976
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8496049041682426e-06,
+      "loss": 1.0668,
+      "step": 4977
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8486227620785398e-06,
+      "loss": 1.0179,
+      "step": 4978
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8476406362543116e-06,
+      "loss": 0.7766,
+      "step": 4979
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8466585268010847e-06,
+      "loss": 0.8783,
+      "step": 4980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8456764338243893e-06,
+      "loss": 1.2013,
+      "step": 4981
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.84469435742975e-06,
+      "loss": 0.8621,
+      "step": 4982
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8437122977226906e-06,
+      "loss": 0.9092,
+      "step": 4983
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8427302548087337e-06,
+      "loss": 1.038,
+      "step": 4984
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8417482287934e-06,
+      "loss": 0.8483,
+      "step": 4985
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8407662197822073e-06,
+      "loss": 0.86,
+      "step": 4986
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8397842278806737e-06,
+      "loss": 0.938,
+      "step": 4987
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.838802253194312e-06,
+      "loss": 1.0393,
+      "step": 4988
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.837820295828638e-06,
+      "loss": 0.9221,
+      "step": 4989
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8368383558891614e-06,
+      "loss": 0.9856,
+      "step": 4990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8358564334813913e-06,
+      "loss": 0.819,
+      "step": 4991
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.834874528710836e-06,
+      "loss": 0.9211,
+      "step": 4992
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.833892641683001e-06,
+      "loss": 0.9553,
+      "step": 4993
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.83291077250339e-06,
+      "loss": 0.7451,
+      "step": 4994
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8319289212775042e-06,
+      "loss": 0.7096,
+      "step": 4995
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.830947088110843e-06,
+      "loss": 0.7362,
+      "step": 4996
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8299652731089066e-06,
+      "loss": 0.9312,
+      "step": 4997
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8289834763771894e-06,
+      "loss": 1.1367,
+      "step": 4998
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.828001698021184e-06,
+      "loss": 0.8969,
+      "step": 4999
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8270199381463846e-06,
+      "loss": 0.7711,
+      "step": 5000
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0679,
+      "eval_samples_per_second": 351.41,
+      "eval_steps_per_second": 43.935,
+      "step": 5000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.826038196858281e-06,
+      "loss": 0.7925,
+      "step": 5001
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.825056474262361e-06,
+      "loss": 0.8544,
+      "step": 5002
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.824074770464109e-06,
+      "loss": 0.8577,
+      "step": 5003
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.823093085569011e-06,
+      "loss": 1.1027,
+      "step": 5004
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8221114196825485e-06,
+      "loss": 0.868,
+      "step": 5005
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8211297729102017e-06,
+      "loss": 0.7795,
+      "step": 5006
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8201481453574474e-06,
+      "loss": 0.9592,
+      "step": 5007
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8191665371297617e-06,
+      "loss": 1.1029,
+      "step": 5008
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8181849483326197e-06,
+      "loss": 0.7897,
+      "step": 5009
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8172033790714917e-06,
+      "loss": 0.7872,
+      "step": 5010
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.816221829451847e-06,
+      "loss": 0.8785,
+      "step": 5011
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8152402995791536e-06,
+      "loss": 0.9566,
+      "step": 5012
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8142587895588776e-06,
+      "loss": 1.1451,
+      "step": 5013
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8132772994964813e-06,
+      "loss": 0.8867,
+      "step": 5014
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8122958294974254e-06,
+      "loss": 0.9566,
+      "step": 5015
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8113143796671687e-06,
+      "loss": 0.8351,
+      "step": 5016
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8103329501111696e-06,
+      "loss": 1.1455,
+      "step": 5017
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.809351540934881e-06,
+      "loss": 0.8174,
+      "step": 5018
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8083701522437556e-06,
+      "loss": 0.7852,
+      "step": 5019
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.807388784143244e-06,
+      "loss": 0.8544,
+      "step": 5020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8064074367387934e-06,
+      "loss": 0.9393,
+      "step": 5021
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8054261101358504e-06,
+      "loss": 1.1143,
+      "step": 5022
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8044448044398575e-06,
+      "loss": 0.9109,
+      "step": 5023
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8034635197562563e-06,
+      "loss": 0.9565,
+      "step": 5024
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.802482256190486e-06,
+      "loss": 1.0175,
+      "step": 5025
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.801501013847984e-06,
+      "loss": 1.0275,
+      "step": 5026
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8005197928341825e-06,
+      "loss": 1.1145,
+      "step": 5027
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.799538593254516e-06,
+      "loss": 0.843,
+      "step": 5028
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7985574152144132e-06,
+      "loss": 0.6934,
+      "step": 5029
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.797576258819302e-06,
+      "loss": 0.924,
+      "step": 5030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7965951241746062e-06,
+      "loss": 1.0669,
+      "step": 5031
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7956140113857507e-06,
+      "loss": 0.6598,
+      "step": 5032
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7946329205581556e-06,
+      "loss": 0.7969,
+      "step": 5033
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.793651851797238e-06,
+      "loss": 1.1766,
+      "step": 5034
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.792670805208413e-06,
+      "loss": 0.9353,
+      "step": 5035
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.791689780897097e-06,
+      "loss": 0.8661,
+      "step": 5036
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7907087789686987e-06,
+      "loss": 0.8104,
+      "step": 5037
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7897277995286272e-06,
+      "loss": 0.9052,
+      "step": 5038
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7887468426822875e-06,
+      "loss": 0.991,
+      "step": 5039
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7877659085350858e-06,
+      "loss": 0.9302,
+      "step": 5040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.786784997192422e-06,
+      "loss": 1.0034,
+      "step": 5041
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7858041087596947e-06,
+      "loss": 0.7973,
+      "step": 5042
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7848232433423e-06,
+      "loss": 0.9125,
+      "step": 5043
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.783842401045633e-06,
+      "loss": 0.9479,
+      "step": 5044
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.782861581975085e-06,
+      "loss": 0.728,
+      "step": 5045
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7818807862360448e-06,
+      "loss": 0.9541,
+      "step": 5046
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.780900013933897e-06,
+      "loss": 0.8156,
+      "step": 5047
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7799192651740286e-06,
+      "loss": 0.8869,
+      "step": 5048
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.778938540061819e-06,
+      "loss": 0.9201,
+      "step": 5049
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7779578387026474e-06,
+      "loss": 0.8351,
+      "step": 5050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7769771612018894e-06,
+      "loss": 1.0088,
+      "step": 5051
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7759965076649203e-06,
+      "loss": 0.8211,
+      "step": 5052
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.775015878197111e-06,
+      "loss": 0.7549,
+      "step": 5053
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.774035272903829e-06,
+      "loss": 0.7762,
+      "step": 5054
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.77305469189044e-06,
+      "loss": 0.9624,
+      "step": 5055
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.772074135262309e-06,
+      "loss": 0.9319,
+      "step": 5056
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7710936031247966e-06,
+      "loss": 0.9544,
+      "step": 5057
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7701130955832594e-06,
+      "loss": 1.2526,
+      "step": 5058
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.769132612743053e-06,
+      "loss": 1.1122,
+      "step": 5059
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.768152154709532e-06,
+      "loss": 0.7325,
+      "step": 5060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.767171721588045e-06,
+      "loss": 0.9479,
+      "step": 5061
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7661913134839404e-06,
+      "loss": 0.9745,
+      "step": 5062
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.765210930502561e-06,
+      "loss": 0.9945,
+      "step": 5063
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7642305727492516e-06,
+      "loss": 1.0837,
+      "step": 5064
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7632502403293503e-06,
+      "loss": 0.8235,
+      "step": 5065
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.762269933348194e-06,
+      "loss": 0.7867,
+      "step": 5066
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7612896519111148e-06,
+      "loss": 0.9705,
+      "step": 5067
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7603093961234473e-06,
+      "loss": 0.8995,
+      "step": 5068
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7593291660905175e-06,
+      "loss": 0.9968,
+      "step": 5069
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.758348961917652e-06,
+      "loss": 0.9433,
+      "step": 5070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.757368783710172e-06,
+      "loss": 1.2072,
+      "step": 5071
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7563886315734e-06,
+      "loss": 1.1102,
+      "step": 5072
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7554085056126523e-06,
+      "loss": 0.9284,
+      "step": 5073
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.754428405933243e-06,
+      "loss": 0.8762,
+      "step": 5074
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.753448332640483e-06,
+      "loss": 0.9954,
+      "step": 5075
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.752468285839683e-06,
+      "loss": 0.977,
+      "step": 5076
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7514882656361484e-06,
+      "loss": 0.9597,
+      "step": 5077
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7505082721351815e-06,
+      "loss": 1.026,
+      "step": 5078
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.749528305442082e-06,
+      "loss": 1.0157,
+      "step": 5079
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.748548365662149e-06,
+      "loss": 0.9687,
+      "step": 5080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.747568452900677e-06,
+      "loss": 0.863,
+      "step": 5081
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.746588567262956e-06,
+      "loss": 0.8272,
+      "step": 5082
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.745608708854274e-06,
+      "loss": 0.737,
+      "step": 5083
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.74462887777992e-06,
+      "loss": 0.9075,
+      "step": 5084
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7436490741451745e-06,
+      "loss": 0.913,
+      "step": 5085
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.742669298055318e-06,
+      "loss": 0.8773,
+      "step": 5086
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7416895496156253e-06,
+      "loss": 0.8703,
+      "step": 5087
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7407098289313737e-06,
+      "loss": 0.9843,
+      "step": 5088
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7397301361078322e-06,
+      "loss": 0.6712,
+      "step": 5089
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7387504712502694e-06,
+      "loss": 1.0755,
+      "step": 5090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.737770834463948e-06,
+      "loss": 0.7956,
+      "step": 5091
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7367912258541332e-06,
+      "loss": 0.9749,
+      "step": 5092
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7358116455260828e-06,
+      "loss": 0.847,
+      "step": 5093
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7348320935850506e-06,
+      "loss": 0.922,
+      "step": 5094
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.733852570136292e-06,
+      "loss": 1.0573,
+      "step": 5095
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.732873075285056e-06,
+      "loss": 0.9509,
+      "step": 5096
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7318936091365896e-06,
+      "loss": 0.9445,
+      "step": 5097
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7309141717961337e-06,
+      "loss": 1.0337,
+      "step": 5098
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7299347633689327e-06,
+      "loss": 0.9722,
+      "step": 5099
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7289553839602216e-06,
+      "loss": 0.983,
+      "step": 5100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7279760336752354e-06,
+      "loss": 0.8549,
+      "step": 5101
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7269967126192038e-06,
+      "loss": 0.9359,
+      "step": 5102
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7260174208973568e-06,
+      "loss": 0.9654,
+      "step": 5103
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7250381586149187e-06,
+      "loss": 0.9362,
+      "step": 5104
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.724058925877111e-06,
+      "loss": 1.0239,
+      "step": 5105
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7230797227891507e-06,
+      "loss": 0.9569,
+      "step": 5106
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.722100549456255e-06,
+      "loss": 0.9636,
+      "step": 5107
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.721121405983636e-06,
+      "loss": 0.7328,
+      "step": 5108
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7201422924765023e-06,
+      "loss": 1.0647,
+      "step": 5109
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.719163209040058e-06,
+      "loss": 0.7425,
+      "step": 5110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7181841557795074e-06,
+      "loss": 0.9469,
+      "step": 5111
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7172051328000497e-06,
+      "loss": 0.9559,
+      "step": 5112
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.71622614020688e-06,
+      "loss": 0.9677,
+      "step": 5113
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.71524717810519e-06,
+      "loss": 1.0308,
+      "step": 5114
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.714268246600172e-06,
+      "loss": 0.7548,
+      "step": 5115
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7132893457970107e-06,
+      "loss": 0.8815,
+      "step": 5116
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7123104758008885e-06,
+      "loss": 0.9685,
+      "step": 5117
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7113316367169835e-06,
+      "loss": 0.7969,
+      "step": 5118
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.710352828650475e-06,
+      "loss": 0.7693,
+      "step": 5119
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.709374051706535e-06,
+      "loss": 0.8285,
+      "step": 5120
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7083953059903314e-06,
+      "loss": 0.9948,
+      "step": 5121
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.707416591607031e-06,
+      "loss": 0.7371,
+      "step": 5122
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7064379086617978e-06,
+      "loss": 0.7764,
+      "step": 5123
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.70545925725979e-06,
+      "loss": 0.7368,
+      "step": 5124
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7044806375061645e-06,
+      "loss": 0.7218,
+      "step": 5125
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7035020495060723e-06,
+      "loss": 0.8972,
+      "step": 5126
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7025234933646646e-06,
+      "loss": 0.8921,
+      "step": 5127
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7015449691870866e-06,
+      "loss": 0.8893,
+      "step": 5128
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.70056647707848e-06,
+      "loss": 0.9701,
+      "step": 5129
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.699588017143983e-06,
+      "loss": 0.8893,
+      "step": 5130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.698609589488733e-06,
+      "loss": 1.071,
+      "step": 5131
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.697631194217862e-06,
+      "loss": 0.8426,
+      "step": 5132
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6966528314364973e-06,
+      "loss": 0.8375,
+      "step": 5133
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6956745012497625e-06,
+      "loss": 0.7207,
+      "step": 5134
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.694696203762782e-06,
+      "loss": 0.7397,
+      "step": 5135
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6937179390806732e-06,
+      "loss": 0.972,
+      "step": 5136
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6927397073085498e-06,
+      "loss": 0.8002,
+      "step": 5137
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6917615085515213e-06,
+      "loss": 1.143,
+      "step": 5138
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6907833429146982e-06,
+      "loss": 0.9981,
+      "step": 5139
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.689805210503183e-06,
+      "loss": 0.8833,
+      "step": 5140
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.688827111422076e-06,
+      "loss": 0.7883,
+      "step": 5141
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.687849045776472e-06,
+      "loss": 0.886,
+      "step": 5142
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.686871013671467e-06,
+      "loss": 0.9139,
+      "step": 5143
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6858930152121497e-06,
+      "loss": 1.0063,
+      "step": 5144
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.684915050503605e-06,
+      "loss": 0.8107,
+      "step": 5145
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6839371196509147e-06,
+      "loss": 0.8096,
+      "step": 5146
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.682959222759159e-06,
+      "loss": 0.9493,
+      "step": 5147
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6819813599334127e-06,
+      "loss": 0.7603,
+      "step": 5148
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6810035312787467e-06,
+      "loss": 0.8316,
+      "step": 5149
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.680025736900227e-06,
+      "loss": 0.7575,
+      "step": 5150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6790479769029207e-06,
+      "loss": 0.8597,
+      "step": 5151
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6780702513918857e-06,
+      "loss": 0.8572,
+      "step": 5152
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6770925604721796e-06,
+      "loss": 1.0721,
+      "step": 5153
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6761149042488536e-06,
+      "loss": 0.9597,
+      "step": 5154
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.675137282826959e-06,
+      "loss": 0.9718,
+      "step": 5155
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6741596963115405e-06,
+      "loss": 0.9384,
+      "step": 5156
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.673182144807638e-06,
+      "loss": 0.8398,
+      "step": 5157
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.672204628420292e-06,
+      "loss": 1.0086,
+      "step": 5158
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6712271472545354e-06,
+      "loss": 0.8579,
+      "step": 5159
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6702497014153986e-06,
+      "loss": 1.0966,
+      "step": 5160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6692722910079058e-06,
+      "loss": 0.8094,
+      "step": 5161
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6682949161370835e-06,
+      "loss": 0.8408,
+      "step": 5162
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6673175769079487e-06,
+      "loss": 0.7665,
+      "step": 5163
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.666340273425517e-06,
+      "loss": 0.8694,
+      "step": 5164
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6653630057947967e-06,
+      "loss": 0.8964,
+      "step": 5165
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6643857741207997e-06,
+      "loss": 0.8647,
+      "step": 5166
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.663408578508527e-06,
+      "loss": 1.179,
+      "step": 5167
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.662431419062979e-06,
+      "loss": 0.9626,
+      "step": 5168
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6614542958891493e-06,
+      "loss": 0.975,
+      "step": 5169
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6604772090920322e-06,
+      "loss": 0.9602,
+      "step": 5170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6595001587766153e-06,
+      "loss": 0.8227,
+      "step": 5171
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.658523145047882e-06,
+      "loss": 1.0961,
+      "step": 5172
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6575461680108113e-06,
+      "loss": 0.8232,
+      "step": 5173
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.656569227770381e-06,
+      "loss": 0.9709,
+      "step": 5174
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.655592324431563e-06,
+      "loss": 0.776,
+      "step": 5175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6546154580993255e-06,
+      "loss": 0.9439,
+      "step": 5176
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.653638628878631e-06,
+      "loss": 0.8315,
+      "step": 5177
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.652661836874442e-06,
+      "loss": 0.8479,
+      "step": 5178
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6516850821917143e-06,
+      "loss": 0.8588,
+      "step": 5179
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6507083649353994e-06,
+      "loss": 0.8504,
+      "step": 5180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6497316852104448e-06,
+      "loss": 0.7226,
+      "step": 5181
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6487550431217966e-06,
+      "loss": 0.8435,
+      "step": 5182
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.647778438774394e-06,
+      "loss": 1.0032,
+      "step": 5183
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.646801872273173e-06,
+      "loss": 1.0002,
+      "step": 5184
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6458253437230645e-06,
+      "loss": 0.9403,
+      "step": 5185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6448488532289985e-06,
+      "loss": 0.8692,
+      "step": 5186
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6438724008958983e-06,
+      "loss": 0.8035,
+      "step": 5187
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.642895986828683e-06,
+      "loss": 1.0486,
+      "step": 5188
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.641919611132267e-06,
+      "loss": 0.7863,
+      "step": 5189
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6409432739115643e-06,
+      "loss": 1.2217,
+      "step": 5190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.639966975271482e-06,
+      "loss": 0.8663,
+      "step": 5191
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.638990715316922e-06,
+      "loss": 1.0437,
+      "step": 5192
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6380144941527826e-06,
+      "loss": 0.8771,
+      "step": 5193
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6370383118839615e-06,
+      "loss": 0.827,
+      "step": 5194
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6360621686153475e-06,
+      "loss": 0.9614,
+      "step": 5195
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6350860644518275e-06,
+      "loss": 0.897,
+      "step": 5196
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6341099994982827e-06,
+      "loss": 0.9184,
+      "step": 5197
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.633133973859594e-06,
+      "loss": 0.7468,
+      "step": 5198
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.632157987640633e-06,
+      "loss": 0.8127,
+      "step": 5199
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6311820409462704e-06,
+      "loss": 0.8456,
+      "step": 5200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.63020613388137e-06,
+      "loss": 0.9151,
+      "step": 5201
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.629230266550795e-06,
+      "loss": 1.0365,
+      "step": 5202
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.628254439059402e-06,
+      "loss": 0.8518,
+      "step": 5203
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6272786515120424e-06,
+      "loss": 0.9852,
+      "step": 5204
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.626302904013564e-06,
+      "loss": 0.7815,
+      "step": 5205
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.625327196668813e-06,
+      "loss": 0.9495,
+      "step": 5206
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6243515295826283e-06,
+      "loss": 0.9349,
+      "step": 5207
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.623375902859845e-06,
+      "loss": 0.8513,
+      "step": 5208
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6224003166052922e-06,
+      "loss": 0.8394,
+      "step": 5209
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6214247709238007e-06,
+      "loss": 0.9631,
+      "step": 5210
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6204492659201897e-06,
+      "loss": 0.8531,
+      "step": 5211
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.619473801699278e-06,
+      "loss": 1.0087,
+      "step": 5212
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6184983783658783e-06,
+      "loss": 0.7728,
+      "step": 5213
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.617522996024802e-06,
+      "loss": 0.8124,
+      "step": 5214
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6165476547808523e-06,
+      "loss": 0.8797,
+      "step": 5215
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6155723547388294e-06,
+      "loss": 0.8628,
+      "step": 5216
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6145970960035287e-06,
+      "loss": 0.7396,
+      "step": 5217
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6136218786797437e-06,
+      "loss": 0.7556,
+      "step": 5218
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.61264670287226e-06,
+      "loss": 0.8182,
+      "step": 5219
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6116715686858605e-06,
+      "loss": 0.7876,
+      "step": 5220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6106964762253224e-06,
+      "loss": 0.9361,
+      "step": 5221
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.609721425595421e-06,
+      "loss": 0.936,
+      "step": 5222
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.608746416900925e-06,
+      "loss": 0.9006,
+      "step": 5223
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6077714502465974e-06,
+      "loss": 0.7237,
+      "step": 5224
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6067965257371996e-06,
+      "loss": 0.9579,
+      "step": 5225
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6058216434774877e-06,
+      "loss": 0.7425,
+      "step": 5226
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6048468035722123e-06,
+      "loss": 0.9062,
+      "step": 5227
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6038720061261186e-06,
+      "loss": 1.0381,
+      "step": 5228
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6028972512439505e-06,
+      "loss": 1.0303,
+      "step": 5229
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6019225390304438e-06,
+      "loss": 0.6919,
+      "step": 5230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.600947869590333e-06,
+      "loss": 0.923,
+      "step": 5231
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.599973243028344e-06,
+      "loss": 0.7918,
+      "step": 5232
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5989986594492014e-06,
+      "loss": 0.9626,
+      "step": 5233
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.598024118957625e-06,
+      "loss": 0.8959,
+      "step": 5234
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.597049621658328e-06,
+      "loss": 0.7392,
+      "step": 5235
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.59607516765602e-06,
+      "loss": 0.9978,
+      "step": 5236
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5951007570554066e-06,
+      "loss": 0.7804,
+      "step": 5237
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5941263899611886e-06,
+      "loss": 1.0637,
+      "step": 5238
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5931520664780607e-06,
+      "loss": 0.9909,
+      "step": 5239
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.592177786710714e-06,
+      "loss": 1.0615,
+      "step": 5240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5912035507638346e-06,
+      "loss": 0.8195,
+      "step": 5241
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5902293587421057e-06,
+      "loss": 0.8515,
+      "step": 5242
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5892552107502026e-06,
+      "loss": 0.8845,
+      "step": 5243
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.588281106892797e-06,
+      "loss": 0.9361,
+      "step": 5244
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5873070472745573e-06,
+      "loss": 0.8893,
+      "step": 5245
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5863330320001466e-06,
+      "loss": 0.9266,
+      "step": 5246
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.585359061174222e-06,
+      "loss": 0.8823,
+      "step": 5247
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.584385134901436e-06,
+      "loss": 0.8765,
+      "step": 5248
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5834112532864375e-06,
+      "loss": 0.95,
+      "step": 5249
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5824374164338713e-06,
+      "loss": 0.7734,
+      "step": 5250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5814636244483743e-06,
+      "loss": 1.1288,
+      "step": 5251
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5804898774345805e-06,
+      "loss": 0.8314,
+      "step": 5252
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5795161754971193e-06,
+      "loss": 0.9537,
+      "step": 5253
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.578542518740616e-06,
+      "loss": 1.01,
+      "step": 5254
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.577568907269689e-06,
+      "loss": 1.0062,
+      "step": 5255
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5765953411889523e-06,
+      "loss": 0.8384,
+      "step": 5256
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5756218206030154e-06,
+      "loss": 1.0225,
+      "step": 5257
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5746483456164844e-06,
+      "loss": 1.1161,
+      "step": 5258
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5736749163339585e-06,
+      "loss": 0.7405,
+      "step": 5259
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.572701532860032e-06,
+      "loss": 0.8706,
+      "step": 5260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5717281952992955e-06,
+      "loss": 0.8597,
+      "step": 5261
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5707549037563335e-06,
+      "loss": 0.9166,
+      "step": 5262
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5697816583357267e-06,
+      "loss": 0.9452,
+      "step": 5263
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.5688084591420497e-06,
+      "loss": 0.8585,
+      "step": 5264
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5678353062798734e-06,
+      "loss": 0.8866,
+      "step": 5265
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.566862199853762e-06,
+      "loss": 0.8755,
+      "step": 5266
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.565889139968277e-06,
+      "loss": 0.9105,
+      "step": 5267
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5649161267279728e-06,
+      "loss": 0.9641,
+      "step": 5268
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5639431602374e-06,
+      "loss": 0.8825,
+      "step": 5269
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5629702406011023e-06,
+      "loss": 1.1768,
+      "step": 5270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5619973679236226e-06,
+      "loss": 0.8002,
+      "step": 5271
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5610245423094936e-06,
+      "loss": 1.0945,
+      "step": 5272
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5600517638632463e-06,
+      "loss": 0.987,
+      "step": 5273
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5590790326894056e-06,
+      "loss": 1.0344,
+      "step": 5274
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5581063488924914e-06,
+      "loss": 1.0447,
+      "step": 5275
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5571337125770183e-06,
+      "loss": 0.9885,
+      "step": 5276
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5561611238474966e-06,
+      "loss": 0.9099,
+      "step": 5277
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5551885828084303e-06,
+      "loss": 0.9503,
+      "step": 5278
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.554216089564319e-06,
+      "loss": 0.7687,
+      "step": 5279
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.553243644219657e-06,
+      "loss": 0.873,
+      "step": 5280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5522712468789337e-06,
+      "loss": 0.7234,
+      "step": 5281
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5512988976466327e-06,
+      "loss": 0.9265,
+      "step": 5282
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5503265966272337e-06,
+      "loss": 0.9434,
+      "step": 5283
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5493543439252088e-06,
+      "loss": 0.8885,
+      "step": 5284
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5483821396450284e-06,
+      "loss": 0.8281,
+      "step": 5285
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5474099838911543e-06,
+      "loss": 0.824,
+      "step": 5286
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.546437876768046e-06,
+      "loss": 0.8565,
+      "step": 5287
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5454658183801542e-06,
+      "loss": 0.7567,
+      "step": 5288
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.544493808831929e-06,
+      "loss": 0.9076,
+      "step": 5289
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5435218482278107e-06,
+      "loss": 0.9721,
+      "step": 5290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.542549936672237e-06,
+      "loss": 1.014,
+      "step": 5291
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.541578074269641e-06,
+      "loss": 1.1479,
+      "step": 5292
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.540606261124448e-06,
+      "loss": 0.7412,
+      "step": 5293
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.539634497341079e-06,
+      "loss": 1.0028,
+      "step": 5294
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.53866278302395e-06,
+      "loss": 0.8662,
+      "step": 5295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5376911182774728e-06,
+      "loss": 0.8552,
+      "step": 5296
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.536719503206052e-06,
+      "loss": 1.0158,
+      "step": 5297
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5357479379140873e-06,
+      "loss": 0.8522,
+      "step": 5298
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.534776422505973e-06,
+      "loss": 0.7945,
+      "step": 5299
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5338049570861e-06,
+      "loss": 0.9268,
+      "step": 5300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5328335417588506e-06,
+      "loss": 0.9811,
+      "step": 5301
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5318621766286033e-06,
+      "loss": 0.8622,
+      "step": 5302
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.530890861799732e-06,
+      "loss": 0.9728,
+      "step": 5303
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5299195973766048e-06,
+      "loss": 0.9703,
+      "step": 5304
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5289483834635827e-06,
+      "loss": 0.993,
+      "step": 5305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.527977220165023e-06,
+      "loss": 0.8691,
+      "step": 5306
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5270061075852772e-06,
+      "loss": 0.9669,
+      "step": 5307
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5260350458286926e-06,
+      "loss": 1.0025,
+      "step": 5308
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.525064034999608e-06,
+      "loss": 0.8415,
+      "step": 5309
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.524093075202358e-06,
+      "loss": 0.7974,
+      "step": 5310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5231221665412733e-06,
+      "loss": 1.1382,
+      "step": 5311
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.5221513091206785e-06,
+      "loss": 0.8827,
+      "step": 5312
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5211805030448917e-06,
+      "loss": 0.9453,
+      "step": 5313
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5202097484182244e-06,
+      "loss": 0.8106,
+      "step": 5314
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5192390453449864e-06,
+      "loss": 0.7253,
+      "step": 5315
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5182683939294786e-06,
+      "loss": 0.7936,
+      "step": 5316
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5172977942759975e-06,
+      "loss": 1.0965,
+      "step": 5317
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.516327246488833e-06,
+      "loss": 0.876,
+      "step": 5318
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.515356750672272e-06,
+      "loss": 0.7897,
+      "step": 5319
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5143863069305937e-06,
+      "loss": 1.0369,
+      "step": 5320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.513415915368073e-06,
+      "loss": 1.0007,
+      "step": 5321
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5124455760889766e-06,
+      "loss": 1.0471,
+      "step": 5322
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.511475289197567e-06,
+      "loss": 0.8659,
+      "step": 5323
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.510505054798104e-06,
+      "loss": 1.0309,
+      "step": 5324
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5095348729948387e-06,
+      "loss": 0.8687,
+      "step": 5325
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5085647438920156e-06,
+      "loss": 0.9464,
+      "step": 5326
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5075946675938745e-06,
+      "loss": 0.8846,
+      "step": 5327
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5066246442046524e-06,
+      "loss": 0.952,
+      "step": 5328
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.505654673828577e-06,
+      "loss": 0.973,
+      "step": 5329
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.504684756569872e-06,
+      "loss": 1.1483,
+      "step": 5330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.503714892532753e-06,
+      "loss": 0.792,
+      "step": 5331
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.502745081821435e-06,
+      "loss": 0.9955,
+      "step": 5332
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.501775324540122e-06,
+      "loss": 0.8566,
+      "step": 5333
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5008056207930154e-06,
+      "loss": 0.9623,
+      "step": 5334
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4998359706843078e-06,
+      "loss": 0.9417,
+      "step": 5335
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.49886637431819e-06,
+      "loss": 1.1414,
+      "step": 5336
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.497896831798845e-06,
+      "loss": 1.1319,
+      "step": 5337
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.49692734323045e-06,
+      "loss": 1.1232,
+      "step": 5338
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4959579087171744e-06,
+      "loss": 0.7627,
+      "step": 5339
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.494988528363187e-06,
+      "loss": 0.7209,
+      "step": 5340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.494019202272646e-06,
+      "loss": 1.159,
+      "step": 5341
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.493049930549706e-06,
+      "loss": 0.8077,
+      "step": 5342
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.492080713298513e-06,
+      "loss": 0.7464,
+      "step": 5343
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.491111550623213e-06,
+      "loss": 0.9151,
+      "step": 5344
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.49014244262794e-06,
+      "loss": 0.7375,
+      "step": 5345
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4891733894168255e-06,
+      "loss": 0.9447,
+      "step": 5346
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4882043910939924e-06,
+      "loss": 0.7676,
+      "step": 5347
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4872354477635617e-06,
+      "loss": 1.0621,
+      "step": 5348
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.486266559529646e-06,
+      "loss": 0.8594,
+      "step": 5349
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.485297726496352e-06,
+      "loss": 0.9788,
+      "step": 5350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.484328948767779e-06,
+      "loss": 0.8993,
+      "step": 5351
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4833602264480255e-06,
+      "loss": 0.8525,
+      "step": 5352
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4823915596411787e-06,
+      "loss": 1.002,
+      "step": 5353
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4814229484513215e-06,
+      "loss": 0.9537,
+      "step": 5354
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4804543929825302e-06,
+      "loss": 0.8692,
+      "step": 5355
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4794858933388785e-06,
+      "loss": 0.8332,
+      "step": 5356
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4785174496244307e-06,
+      "loss": 0.8221,
+      "step": 5357
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4775490619432444e-06,
+      "loss": 1.0476,
+      "step": 5358
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.476580730399375e-06,
+      "loss": 0.7571,
+      "step": 5359
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4756124550968693e-06,
+      "loss": 0.8967,
+      "step": 5360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.474644236139768e-06,
+      "loss": 0.9004,
+      "step": 5361
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4736760736321043e-06,
+      "loss": 0.8777,
+      "step": 5362
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4727079676779104e-06,
+      "loss": 0.8797,
+      "step": 5363
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4717399183812073e-06,
+      "loss": 0.8421,
+      "step": 5364
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.470771925846013e-06,
+      "loss": 0.6207,
+      "step": 5365
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4698039901763357e-06,
+      "loss": 0.7403,
+      "step": 5366
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4688361114761835e-06,
+      "loss": 1.0208,
+      "step": 5367
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.467868289849553e-06,
+      "loss": 0.713,
+      "step": 5368
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.466900525400437e-06,
+      "loss": 0.9579,
+      "step": 5369
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.46593281823282e-06,
+      "loss": 0.7959,
+      "step": 5370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.464965168450685e-06,
+      "loss": 0.9666,
+      "step": 5371
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.463997576158004e-06,
+      "loss": 0.7785,
+      "step": 5372
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.463030041458746e-06,
+      "loss": 0.9794,
+      "step": 5373
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4620625644568695e-06,
+      "loss": 0.9188,
+      "step": 5374
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.461095145256334e-06,
+      "loss": 0.8623,
+      "step": 5375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.460127783961087e-06,
+      "loss": 0.974,
+      "step": 5376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4591604806750703e-06,
+      "loss": 0.9384,
+      "step": 5377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4581932355022204e-06,
+      "loss": 0.8196,
+      "step": 5378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4572260485464697e-06,
+      "loss": 1.0069,
+      "step": 5379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4562589199117414e-06,
+      "loss": 0.8614,
+      "step": 5380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.455291849701954e-06,
+      "loss": 0.9308,
+      "step": 5381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4543248380210166e-06,
+      "loss": 0.6554,
+      "step": 5382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.453357884972838e-06,
+      "loss": 0.8718,
+      "step": 5383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.452390990661315e-06,
+      "loss": 0.8975,
+      "step": 5384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.451424155190342e-06,
+      "loss": 0.7643,
+      "step": 5385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.450457378663802e-06,
+      "loss": 0.8519,
+      "step": 5386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4494906611855797e-06,
+      "loss": 0.9264,
+      "step": 5387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.448524002859546e-06,
+      "loss": 0.9676,
+      "step": 5388
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.447557403789569e-06,
+      "loss": 0.6134,
+      "step": 5389
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4465908640795088e-06,
+      "loss": 0.8566,
+      "step": 5390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.445624383833221e-06,
+      "loss": 0.8363,
+      "step": 5391
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4446579631545544e-06,
+      "loss": 0.7613,
+      "step": 5392
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4436916021473497e-06,
+      "loss": 0.7831,
+      "step": 5393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4427253009154417e-06,
+      "loss": 0.9016,
+      "step": 5394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4417590595626614e-06,
+      "loss": 0.8641,
+      "step": 5395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4407928781928302e-06,
+      "loss": 0.9774,
+      "step": 5396
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.439826756909764e-06,
+      "loss": 0.8459,
+      "step": 5397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.438860695817272e-06,
+      "loss": 0.998,
+      "step": 5398
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.43789469501916e-06,
+      "loss": 0.8946,
+      "step": 5399
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.436928754619222e-06,
+      "loss": 0.8568,
+      "step": 5400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4359628747212492e-06,
+      "loss": 0.9063,
+      "step": 5401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4349970554290247e-06,
+      "loss": 0.9024,
+      "step": 5402
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.434031296846327e-06,
+      "loss": 0.8009,
+      "step": 5403
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.433065599076926e-06,
+      "loss": 0.939,
+      "step": 5404
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4320999622245858e-06,
+      "loss": 0.9569,
+      "step": 5405
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4311343863930634e-06,
+      "loss": 1.0814,
+      "step": 5406
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.430168871686112e-06,
+      "loss": 1.222,
+      "step": 5407
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4292034182074742e-06,
+      "loss": 0.7838,
+      "step": 5408
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.428238026060888e-06,
+      "loss": 0.8051,
+      "step": 5409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4272726953500844e-06,
+      "loss": 1.1764,
+      "step": 5410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.42630742617879e-06,
+      "loss": 0.8174,
+      "step": 5411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4253422186507216e-06,
+      "loss": 0.7176,
+      "step": 5412
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.424377072869591e-06,
+      "loss": 0.981,
+      "step": 5413
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4234119889391014e-06,
+      "loss": 0.927,
+      "step": 5414
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.422446966962954e-06,
+      "loss": 1.0797,
+      "step": 5415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.421482007044838e-06,
+      "loss": 0.7752,
+      "step": 5416
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4205171092884398e-06,
+      "loss": 0.7886,
+      "step": 5417
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.419552273797436e-06,
+      "loss": 0.9621,
+      "step": 5418
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4185875006754994e-06,
+      "loss": 0.7536,
+      "step": 5419
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4176227900262943e-06,
+      "loss": 0.9179,
+      "step": 5420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4166581419534793e-06,
+      "loss": 1.0436,
+      "step": 5421
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.415693556560704e-06,
+      "loss": 0.7963,
+      "step": 5422
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4147290339516154e-06,
+      "loss": 1.0848,
+      "step": 5423
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4137645742298505e-06,
+      "loss": 0.8711,
+      "step": 5424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.412800177499039e-06,
+      "loss": 0.7632,
+      "step": 5425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4118358438628077e-06,
+      "loss": 0.8809,
+      "step": 5426
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4108715734247732e-06,
+      "loss": 0.7748,
+      "step": 5427
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.409907366288546e-06,
+      "loss": 0.9742,
+      "step": 5428
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4089432225577286e-06,
+      "loss": 0.8633,
+      "step": 5429
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4079791423359214e-06,
+      "loss": 1.0183,
+      "step": 5430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4070151257267133e-06,
+      "loss": 0.8488,
+      "step": 5431
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4060511728336876e-06,
+      "loss": 0.6782,
+      "step": 5432
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.405087283760419e-06,
+      "loss": 0.9489,
+      "step": 5433
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4041234586104814e-06,
+      "loss": 0.9842,
+      "step": 5434
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.403159697487435e-06,
+      "loss": 0.7827,
+      "step": 5435
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.402196000494837e-06,
+      "loss": 0.7819,
+      "step": 5436
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4012323677362344e-06,
+      "loss": 0.9578,
+      "step": 5437
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4002687993151726e-06,
+      "loss": 0.8058,
+      "step": 5438
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3993052953351856e-06,
+      "loss": 0.8173,
+      "step": 5439
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3983418558998017e-06,
+      "loss": 0.9474,
+      "step": 5440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3973784811125413e-06,
+      "loss": 0.7881,
+      "step": 5441
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3964151710769213e-06,
+      "loss": 0.8114,
+      "step": 5442
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.395451925896448e-06,
+      "loss": 0.9245,
+      "step": 5443
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3944887456746226e-06,
+      "loss": 0.9082,
+      "step": 5444
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3935256305149364e-06,
+      "loss": 1.0817,
+      "step": 5445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3925625805208795e-06,
+      "loss": 0.6958,
+      "step": 5446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3915995957959298e-06,
+      "loss": 1.0326,
+      "step": 5447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.39063667644356e-06,
+      "loss": 0.7838,
+      "step": 5448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.389673822567235e-06,
+      "loss": 0.8171,
+      "step": 5449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.388711034270415e-06,
+      "loss": 0.7969,
+      "step": 5450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3877483116565514e-06,
+      "loss": 0.7174,
+      "step": 5451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3867856548290873e-06,
+      "loss": 0.9831,
+      "step": 5452
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.38582306389146e-06,
+      "loss": 1.0048,
+      "step": 5453
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.384860538947102e-06,
+      "loss": 0.7778,
+      "step": 5454
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.383898080099435e-06,
+      "loss": 0.8828,
+      "step": 5455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.382935687451875e-06,
+      "loss": 1.096,
+      "step": 5456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.3819733611078303e-06,
+      "loss": 0.6968,
+      "step": 5457
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.381011101170705e-06,
+      "loss": 0.9428,
+      "step": 5458
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3800489077438933e-06,
+      "loss": 0.9098,
+      "step": 5459
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3790867809307818e-06,
+      "loss": 0.8492,
+      "step": 5460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.37812472083475e-06,
+      "loss": 0.9224,
+      "step": 5461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.377162727559174e-06,
+      "loss": 1.1467,
+      "step": 5462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3762008012074185e-06,
+      "loss": 1.0859,
+      "step": 5463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.375238941882843e-06,
+      "loss": 0.8281,
+      "step": 5464
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3742771496887965e-06,
+      "loss": 0.88,
+      "step": 5465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.373315424728627e-06,
+      "loss": 0.8811,
+      "step": 5466
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3723537671056704e-06,
+      "loss": 0.7716,
+      "step": 5467
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.371392176923257e-06,
+      "loss": 0.8681,
+      "step": 5468
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.370430654284708e-06,
+      "loss": 0.987,
+      "step": 5469
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.369469199293341e-06,
+      "loss": 0.8321,
+      "step": 5470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.368507812052464e-06,
+      "loss": 0.9756,
+      "step": 5471
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3675464926653777e-06,
+      "loss": 0.6987,
+      "step": 5472
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.366585241235374e-06,
+      "loss": 0.769,
+      "step": 5473
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3656240578657424e-06,
+      "loss": 0.8341,
+      "step": 5474
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3646629426597608e-06,
+      "loss": 0.8595,
+      "step": 5475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3637018957207002e-06,
+      "loss": 0.8775,
+      "step": 5476
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3627409171518242e-06,
+      "loss": 1.1025,
+      "step": 5477
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.361780007056393e-06,
+      "loss": 0.8462,
+      "step": 5478
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.360819165537654e-06,
+      "loss": 0.9123,
+      "step": 5479
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3598583926988504e-06,
+      "loss": 0.8974,
+      "step": 5480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3588976886432153e-06,
+      "loss": 0.8425,
+      "step": 5481
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3579370534739793e-06,
+      "loss": 0.9674,
+      "step": 5482
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.356976487294361e-06,
+      "loss": 0.9959,
+      "step": 5483
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3560159902075732e-06,
+      "loss": 0.8991,
+      "step": 5484
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3550555623168196e-06,
+      "loss": 0.7641,
+      "step": 5485
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3540952037253017e-06,
+      "loss": 0.6567,
+      "step": 5486
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.353134914536208e-06,
+      "loss": 0.8492,
+      "step": 5487
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.352174694852721e-06,
+      "loss": 0.9502,
+      "step": 5488
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.351214544778016e-06,
+      "loss": 0.9703,
+      "step": 5489
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.350254464415263e-06,
+      "loss": 0.8644,
+      "step": 5490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3492944538676216e-06,
+      "loss": 0.862,
+      "step": 5491
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3483345132382434e-06,
+      "loss": 0.9087,
+      "step": 5492
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3473746426302767e-06,
+      "loss": 0.9434,
+      "step": 5493
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3464148421468577e-06,
+      "loss": 0.9579,
+      "step": 5494
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3454551118911178e-06,
+      "loss": 0.8264,
+      "step": 5495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3444954519661773e-06,
+      "loss": 0.8286,
+      "step": 5496
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.343535862475156e-06,
+      "loss": 0.8817,
+      "step": 5497
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3425763435211587e-06,
+      "loss": 0.9308,
+      "step": 5498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3416168952072865e-06,
+      "loss": 0.7876,
+      "step": 5499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3406575176366306e-06,
+      "loss": 0.9679,
+      "step": 5500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.339698210912279e-06,
+      "loss": 0.8899,
+      "step": 5501
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.338738975137307e-06,
+      "loss": 0.889,
+      "step": 5502
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3377798104147854e-06,
+      "loss": 0.8001,
+      "step": 5503
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3368207168477745e-06,
+      "loss": 0.8805,
+      "step": 5504
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.3358616945393314e-06,
+      "loss": 1.0487,
+      "step": 5505
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3349027435925023e-06,
+      "loss": 0.8726,
+      "step": 5506
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3339438641103253e-06,
+      "loss": 0.9239,
+      "step": 5507
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3329850561958314e-06,
+      "loss": 0.8958,
+      "step": 5508
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3320263199520474e-06,
+      "loss": 0.8072,
+      "step": 5509
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3310676554819873e-06,
+      "loss": 1.0489,
+      "step": 5510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.33010906288866e-06,
+      "loss": 0.8533,
+      "step": 5511
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3291505422750647e-06,
+      "loss": 0.7483,
+      "step": 5512
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3281920937441972e-06,
+      "loss": 0.8683,
+      "step": 5513
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.327233717399041e-06,
+      "loss": 0.8054,
+      "step": 5514
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.326275413342574e-06,
+      "loss": 0.893,
+      "step": 5515
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3253171816777644e-06,
+      "loss": 0.8805,
+      "step": 5516
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.324359022507577e-06,
+      "loss": 0.9211,
+      "step": 5517
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.323400935934964e-06,
+      "loss": 0.9211,
+      "step": 5518
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3224429220628728e-06,
+      "loss": 0.7844,
+      "step": 5519
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3214849809942395e-06,
+      "loss": 1.1045,
+      "step": 5520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.320527112831998e-06,
+      "loss": 1.0465,
+      "step": 5521
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.31956931767907e-06,
+      "loss": 0.8235,
+      "step": 5522
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3186115956383697e-06,
+      "loss": 0.8931,
+      "step": 5523
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3176539468128034e-06,
+      "loss": 1.0376,
+      "step": 5524
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3166963713052735e-06,
+      "loss": 0.9745,
+      "step": 5525
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.31573886921867e-06,
+      "loss": 1.1493,
+      "step": 5526
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.314781440655876e-06,
+      "loss": 0.8328,
+      "step": 5527
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.313824085719767e-06,
+      "loss": 1.0781,
+      "step": 5528
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3128668045132103e-06,
+      "loss": 0.7675,
+      "step": 5529
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3119095971390674e-06,
+      "loss": 0.9608,
+      "step": 5530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3109524637001892e-06,
+      "loss": 0.8963,
+      "step": 5531
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3099954042994187e-06,
+      "loss": 0.8578,
+      "step": 5532
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3090384190395932e-06,
+      "loss": 0.6581,
+      "step": 5533
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.308081508023541e-06,
+      "loss": 0.9181,
+      "step": 5534
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3071246713540807e-06,
+      "loss": 0.925,
+      "step": 5535
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.306167909134025e-06,
+      "loss": 0.994,
+      "step": 5536
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3052112214661776e-06,
+      "loss": 1.0201,
+      "step": 5537
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3042546084533356e-06,
+      "loss": 0.9187,
+      "step": 5538
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3032980701982866e-06,
+      "loss": 0.8483,
+      "step": 5539
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.302341606803809e-06,
+      "loss": 0.8908,
+      "step": 5540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.3013852183726768e-06,
+      "loss": 0.8856,
+      "step": 5541
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.300428905007653e-06,
+      "loss": 0.7518,
+      "step": 5542
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2994726668114935e-06,
+      "loss": 0.947,
+      "step": 5543
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2985165038869455e-06,
+      "loss": 0.9937,
+      "step": 5544
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2975604163367488e-06,
+      "loss": 0.885,
+      "step": 5545
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2966044042636365e-06,
+      "loss": 1.0408,
+      "step": 5546
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.29564846777033e-06,
+      "loss": 0.8382,
+      "step": 5547
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2946926069595455e-06,
+      "loss": 0.6981,
+      "step": 5548
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2937368219339896e-06,
+      "loss": 0.9272,
+      "step": 5549
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.292781112796362e-06,
+      "loss": 0.9389,
+      "step": 5550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2918254796493545e-06,
+      "loss": 0.8521,
+      "step": 5551
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.290869922595647e-06,
+      "loss": 0.8924,
+      "step": 5552
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.2899144417379165e-06,
+      "loss": 0.8914,
+      "step": 5553
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2889590371788285e-06,
+      "loss": 0.9827,
+      "step": 5554
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2880037090210413e-06,
+      "loss": 0.9992,
+      "step": 5555
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2870484573672047e-06,
+      "loss": 0.6508,
+      "step": 5556
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2860932823199603e-06,
+      "loss": 0.9602,
+      "step": 5557
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.285138183981942e-06,
+      "loss": 0.9286,
+      "step": 5558
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2841831624557747e-06,
+      "loss": 0.7895,
+      "step": 5559
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2832282178440757e-06,
+      "loss": 1.0617,
+      "step": 5560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.282273350249453e-06,
+      "loss": 0.962,
+      "step": 5561
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2813185597745085e-06,
+      "loss": 0.8417,
+      "step": 5562
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.280363846521833e-06,
+      "loss": 0.9419,
+      "step": 5563
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.279409210594011e-06,
+      "loss": 0.7282,
+      "step": 5564
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.278454652093617e-06,
+      "loss": 0.8804,
+      "step": 5565
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.27750017112322e-06,
+      "loss": 0.9513,
+      "step": 5566
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.276545767785378e-06,
+      "loss": 0.8251,
+      "step": 5567
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.275591442182642e-06,
+      "loss": 1.0187,
+      "step": 5568
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2746371944175528e-06,
+      "loss": 0.9546,
+      "step": 5569
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2736830245926463e-06,
+      "loss": 0.9331,
+      "step": 5570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2727289328104464e-06,
+      "loss": 0.8649,
+      "step": 5571
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.271774919173471e-06,
+      "loss": 0.8431,
+      "step": 5572
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2708209837842283e-06,
+      "loss": 1.0025,
+      "step": 5573
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2698671267452196e-06,
+      "loss": 0.9501,
+      "step": 5574
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.268913348158935e-06,
+      "loss": 0.9841,
+      "step": 5575
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.26795964812786e-06,
+      "loss": 1.0272,
+      "step": 5576
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2670060267544676e-06,
+      "loss": 1.0812,
+      "step": 5577
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.266052484141226e-06,
+      "loss": 1.0769,
+      "step": 5578
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2650990203905923e-06,
+      "loss": 1.0406,
+      "step": 5579
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2641456356050167e-06,
+      "loss": 1.0352,
+      "step": 5580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2631923298869394e-06,
+      "loss": 0.9705,
+      "step": 5581
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2622391033387943e-06,
+      "loss": 0.8707,
+      "step": 5582
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2612859560630046e-06,
+      "loss": 1.0313,
+      "step": 5583
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2603328881619866e-06,
+      "loss": 1.1498,
+      "step": 5584
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2593798997381464e-06,
+      "loss": 1.0202,
+      "step": 5585
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2584269908938837e-06,
+      "loss": 0.9039,
+      "step": 5586
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.257474161731587e-06,
+      "loss": 0.8543,
+      "step": 5587
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2565214123536397e-06,
+      "loss": 1.0249,
+      "step": 5588
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2555687428624138e-06,
+      "loss": 0.9083,
+      "step": 5589
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2546161533602724e-06,
+      "loss": 0.9349,
+      "step": 5590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2536636439495723e-06,
+      "loss": 1.0166,
+      "step": 5591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2527112147326605e-06,
+      "loss": 0.919,
+      "step": 5592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.251758865811876e-06,
+      "loss": 0.925,
+      "step": 5593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.250806597289547e-06,
+      "loss": 1.0044,
+      "step": 5594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2498544092679957e-06,
+      "loss": 1.0786,
+      "step": 5595
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.248902301849535e-06,
+      "loss": 1.0428,
+      "step": 5596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.247950275136469e-06,
+      "loss": 0.931,
+      "step": 5597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2469983292310914e-06,
+      "loss": 0.855,
+      "step": 5598
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2460464642356888e-06,
+      "loss": 0.982,
+      "step": 5599
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.2450946802525413e-06,
+      "loss": 0.9526,
+      "step": 5600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.244142977383916e-06,
+      "loss": 0.7631,
+      "step": 5601
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.243191355732074e-06,
+      "loss": 1.158,
+      "step": 5602
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2422398153992664e-06,
+      "loss": 1.1165,
+      "step": 5603
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2412883564877374e-06,
+      "loss": 0.765,
+      "step": 5604
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.24033697909972e-06,
+      "loss": 0.9883,
+      "step": 5605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2393856833374392e-06,
+      "loss": 0.7167,
+      "step": 5606
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.238434469303113e-06,
+      "loss": 0.9491,
+      "step": 5607
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2374833370989493e-06,
+      "loss": 1.008,
+      "step": 5608
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.236532286827146e-06,
+      "loss": 0.7814,
+      "step": 5609
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2355813185898937e-06,
+      "loss": 0.8405,
+      "step": 5610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.234630432489374e-06,
+      "loss": 0.7971,
+      "step": 5611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.233679628627761e-06,
+      "loss": 0.7374,
+      "step": 5612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2327289071072168e-06,
+      "loss": 0.7473,
+      "step": 5613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.231778268029896e-06,
+      "loss": 0.7713,
+      "step": 5614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.230827711497946e-06,
+      "loss": 0.9075,
+      "step": 5615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2298772376135034e-06,
+      "loss": 0.7237,
+      "step": 5616
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.228926846478697e-06,
+      "loss": 1.0265,
+      "step": 5617
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2279765381956455e-06,
+      "loss": 0.9112,
+      "step": 5618
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.22702631286646e-06,
+      "loss": 0.892,
+      "step": 5619
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2260761705932423e-06,
+      "loss": 1.0545,
+      "step": 5620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2251261114780845e-06,
+      "loss": 0.8189,
+      "step": 5621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.224176135623071e-06,
+      "loss": 1.0189,
+      "step": 5622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2232262431302755e-06,
+      "loss": 0.881,
+      "step": 5623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2222764341017657e-06,
+      "loss": 0.9946,
+      "step": 5624
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2213267086395977e-06,
+      "loss": 0.8107,
+      "step": 5625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.220377066845818e-06,
+      "loss": 0.7502,
+      "step": 5626
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.219427508822468e-06,
+      "loss": 0.8684,
+      "step": 5627
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2184780346715766e-06,
+      "loss": 0.8671,
+      "step": 5628
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2175286444951645e-06,
+      "loss": 0.7699,
+      "step": 5629
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.216579338395243e-06,
+      "loss": 0.8299,
+      "step": 5630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.215630116473816e-06,
+      "loss": 0.9556,
+      "step": 5631
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2146809788328773e-06,
+      "loss": 0.8744,
+      "step": 5632
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2137319255744113e-06,
+      "loss": 0.9223,
+      "step": 5633
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2127829568003934e-06,
+      "loss": 0.7686,
+      "step": 5634
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2118340726127907e-06,
+      "loss": 0.8483,
+      "step": 5635
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2108852731135614e-06,
+      "loss": 0.7659,
+      "step": 5636
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2099365584046527e-06,
+      "loss": 1.0256,
+      "step": 5637
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2089879285880037e-06,
+      "loss": 0.823,
+      "step": 5638
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.208039383765546e-06,
+      "loss": 0.8936,
+      "step": 5639
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2070909240392004e-06,
+      "loss": 0.8905,
+      "step": 5640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2061425495108786e-06,
+      "loss": 1.0608,
+      "step": 5641
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2051942602824817e-06,
+      "loss": 0.9651,
+      "step": 5642
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.204246056455906e-06,
+      "loss": 1.0775,
+      "step": 5643
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2032979381330347e-06,
+      "loss": 0.8742,
+      "step": 5644
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2023499054157434e-06,
+      "loss": 0.9758,
+      "step": 5645
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2014019584058964e-06,
+      "loss": 0.8902,
+      "step": 5646
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.2004540972053536e-06,
+      "loss": 0.9287,
+      "step": 5647
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.1995063219159605e-06,
+      "loss": 0.9106,
+      "step": 5648
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.198558632639556e-06,
+      "loss": 0.8146,
+      "step": 5649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.197611029477968e-06,
+      "loss": 0.8216,
+      "step": 5650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.196663512533019e-06,
+      "loss": 0.828,
+      "step": 5651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1957160819065186e-06,
+      "loss": 0.9108,
+      "step": 5652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1947687377002673e-06,
+      "loss": 0.8463,
+      "step": 5653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.193821480016057e-06,
+      "loss": 1.0868,
+      "step": 5654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.192874308955672e-06,
+      "loss": 0.853,
+      "step": 5655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1919272246208843e-06,
+      "loss": 0.7841,
+      "step": 5656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1909802271134595e-06,
+      "loss": 0.8883,
+      "step": 5657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.19003331653515e-06,
+      "loss": 0.8858,
+      "step": 5658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.189086492987704e-06,
+      "loss": 0.9463,
+      "step": 5659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1881397565728565e-06,
+      "loss": 1.035,
+      "step": 5660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1871931073923343e-06,
+      "loss": 0.9025,
+      "step": 5661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.186246545547854e-06,
+      "loss": 0.8028,
+      "step": 5662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.185300071141125e-06,
+      "loss": 0.8379,
+      "step": 5663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.184353684273845e-06,
+      "loss": 0.995,
+      "step": 5664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1834073850477038e-06,
+      "loss": 0.8797,
+      "step": 5665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1824611735643797e-06,
+      "loss": 0.8696,
+      "step": 5666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.181515049925545e-06,
+      "loss": 0.8451,
+      "step": 5667
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.18056901423286e-06,
+      "loss": 0.9118,
+      "step": 5668
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.179623066587976e-06,
+      "loss": 0.8301,
+      "step": 5669
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1786772070925333e-06,
+      "loss": 1.0058,
+      "step": 5670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.177731435848168e-06,
+      "loss": 0.7521,
+      "step": 5671
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1767857529565006e-06,
+      "loss": 0.7776,
+      "step": 5672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.175840158519146e-06,
+      "loss": 1.064,
+      "step": 5673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1748946526377055e-06,
+      "loss": 0.7572,
+      "step": 5674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.173949235413777e-06,
+      "loss": 1.0631,
+      "step": 5675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1730039069489447e-06,
+      "loss": 0.8719,
+      "step": 5676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1720586673447835e-06,
+      "loss": 0.9575,
+      "step": 5677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.171113516702858e-06,
+      "loss": 0.7841,
+      "step": 5678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1701684551247277e-06,
+      "loss": 0.8419,
+      "step": 5679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1692234827119373e-06,
+      "loss": 0.7977,
+      "step": 5680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1682785995660243e-06,
+      "loss": 0.8249,
+      "step": 5681
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.167333805788515e-06,
+      "loss": 0.9933,
+      "step": 5682
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1663891014809308e-06,
+      "loss": 0.9642,
+      "step": 5683
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1654444867447773e-06,
+      "loss": 0.8426,
+      "step": 5684
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.164499961681555e-06,
+      "loss": 0.8807,
+      "step": 5685
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1635555263927503e-06,
+      "loss": 1.1491,
+      "step": 5686
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.162611180979846e-06,
+      "loss": 0.8824,
+      "step": 5687
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1616669255443108e-06,
+      "loss": 0.9719,
+      "step": 5688
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1607227601876046e-06,
+      "loss": 0.9146,
+      "step": 5689
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1597786850111763e-06,
+      "loss": 0.6946,
+      "step": 5690
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.15883470011647e-06,
+      "loss": 0.9826,
+      "step": 5691
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.157890805604915e-06,
+      "loss": 0.9407,
+      "step": 5692
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.156947001577931e-06,
+      "loss": 0.895,
+      "step": 5693
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1560032881369337e-06,
+      "loss": 0.9068,
+      "step": 5694
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1550596653833227e-06,
+      "loss": 0.86,
+      "step": 5695
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.15411613341849e-06,
+      "loss": 0.9297,
+      "step": 5696
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.1531726923438175e-06,
+      "loss": 1.04,
+      "step": 5697
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.15222934226068e-06,
+      "loss": 0.9492,
+      "step": 5698
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.151286083270439e-06,
+      "loss": 0.9311,
+      "step": 5699
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.150342915474448e-06,
+      "loss": 1.0052,
+      "step": 5700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1493998389740487e-06,
+      "loss": 0.9864,
+      "step": 5701
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1484568538705766e-06,
+      "loss": 1.0153,
+      "step": 5702
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1475139602653555e-06,
+      "loss": 1.1474,
+      "step": 5703
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1465711582596987e-06,
+      "loss": 0.7756,
+      "step": 5704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1456284479549087e-06,
+      "loss": 0.9611,
+      "step": 5705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.144685829452282e-06,
+      "loss": 0.975,
+      "step": 5706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1437433028531016e-06,
+      "loss": 0.8028,
+      "step": 5707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.142800868258643e-06,
+      "loss": 0.8146,
+      "step": 5708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.141858525770168e-06,
+      "loss": 0.79,
+      "step": 5709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.140916275488935e-06,
+      "loss": 0.636,
+      "step": 5710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.139974117516187e-06,
+      "loss": 0.7092,
+      "step": 5711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.139032051953158e-06,
+      "loss": 1.006,
+      "step": 5712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1380900789010725e-06,
+      "loss": 0.8167,
+      "step": 5713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.137148198461148e-06,
+      "loss": 0.8613,
+      "step": 5714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1362064107345878e-06,
+      "loss": 0.964,
+      "step": 5715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1352647158225873e-06,
+      "loss": 0.9335,
+      "step": 5716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.13432311382633e-06,
+      "loss": 1.0682,
+      "step": 5717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.133381604846994e-06,
+      "loss": 0.8114,
+      "step": 5718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.132440188985742e-06,
+      "loss": 0.9259,
+      "step": 5719
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1314988663437304e-06,
+      "loss": 0.9711,
+      "step": 5720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.130557637022102e-06,
+      "loss": 1.1779,
+      "step": 5721
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1296165011219947e-06,
+      "loss": 0.8686,
+      "step": 5722
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.128675458744533e-06,
+      "loss": 0.8824,
+      "step": 5723
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.12773450999083e-06,
+      "loss": 0.8448,
+      "step": 5724
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.126793654961991e-06,
+      "loss": 0.995,
+      "step": 5725
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1258528937591126e-06,
+      "loss": 1.0905,
+      "step": 5726
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1249122264832783e-06,
+      "loss": 0.7187,
+      "step": 5727
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1239716532355627e-06,
+      "loss": 0.9567,
+      "step": 5728
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.12303117411703e-06,
+      "loss": 1.0126,
+      "step": 5729
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1220907892287357e-06,
+      "loss": 1.0955,
+      "step": 5730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1211504986717233e-06,
+      "loss": 1.1024,
+      "step": 5731
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1202103025470274e-06,
+      "loss": 0.7869,
+      "step": 5732
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1192702009556705e-06,
+      "loss": 0.8325,
+      "step": 5733
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1183301939986686e-06,
+      "loss": 0.6544,
+      "step": 5734
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1173902817770245e-06,
+      "loss": 0.8533,
+      "step": 5735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1164504643917314e-06,
+      "loss": 0.995,
+      "step": 5736
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.115510741943772e-06,
+      "loss": 1.0661,
+      "step": 5737
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1145711145341217e-06,
+      "loss": 0.8991,
+      "step": 5738
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1136315822637415e-06,
+      "loss": 0.9965,
+      "step": 5739
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.112692145233585e-06,
+      "loss": 0.8644,
+      "step": 5740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.111752803544593e-06,
+      "loss": 0.9062,
+      "step": 5741
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.110813557297699e-06,
+      "loss": 1.0016,
+      "step": 5742
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1098744065938256e-06,
+      "loss": 0.7329,
+      "step": 5743
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.1089353515338843e-06,
+      "loss": 0.9739,
+      "step": 5744
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.107996392218774e-06,
+      "loss": 0.8691,
+      "step": 5745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.107057528749389e-06,
+      "loss": 0.8004,
+      "step": 5746
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1061187612266087e-06,
+      "loss": 0.959,
+      "step": 5747
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.105180089751304e-06,
+      "loss": 0.9097,
+      "step": 5748
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.104241514424333e-06,
+      "loss": 1.0171,
+      "step": 5749
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1033030353465495e-06,
+      "loss": 0.7767,
+      "step": 5750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1023646526187903e-06,
+      "loss": 0.8448,
+      "step": 5751
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.101426366341885e-06,
+      "loss": 0.8401,
+      "step": 5752
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.1004881766166514e-06,
+      "loss": 0.951,
+      "step": 5753
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0995500835439003e-06,
+      "loss": 0.948,
+      "step": 5754
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0986120872244286e-06,
+      "loss": 0.9432,
+      "step": 5755
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0976741877590227e-06,
+      "loss": 1.0906,
+      "step": 5756
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.096736385248462e-06,
+      "loss": 0.8354,
+      "step": 5757
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0957986797935118e-06,
+      "loss": 0.6783,
+      "step": 5758
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.09486107149493e-06,
+      "loss": 1.0799,
+      "step": 5759
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.093923560453459e-06,
+      "loss": 0.9067,
+      "step": 5760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0929861467698387e-06,
+      "loss": 0.8264,
+      "step": 5761
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.092048830544792e-06,
+      "loss": 0.7049,
+      "step": 5762
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.091111611879034e-06,
+      "loss": 1.0155,
+      "step": 5763
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0901744908732667e-06,
+      "loss": 0.959,
+      "step": 5764
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.089237467628187e-06,
+      "loss": 0.9197,
+      "step": 5765
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0883005422444766e-06,
+      "loss": 0.8472,
+      "step": 5766
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0873637148228074e-06,
+      "loss": 0.9117,
+      "step": 5767
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0864269854638407e-06,
+      "loss": 0.975,
+      "step": 5768
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.085490354268231e-06,
+      "loss": 0.8263,
+      "step": 5769
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0845538213366173e-06,
+      "loss": 0.8815,
+      "step": 5770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.08361738676963e-06,
+      "loss": 0.64,
+      "step": 5771
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0826810506678877e-06,
+      "loss": 0.7922,
+      "step": 5772
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0817448131320026e-06,
+      "loss": 0.9503,
+      "step": 5773
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0808086742625717e-06,
+      "loss": 0.967,
+      "step": 5774
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.079872634160183e-06,
+      "loss": 0.9548,
+      "step": 5775
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.078936692925413e-06,
+      "loss": 0.8336,
+      "step": 5776
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.078000850658831e-06,
+      "loss": 1.0339,
+      "step": 5777
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0770651074609915e-06,
+      "loss": 0.8448,
+      "step": 5778
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0761294634324406e-06,
+      "loss": 0.888,
+      "step": 5779
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0751939186737116e-06,
+      "loss": 0.8832,
+      "step": 5780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0742584732853313e-06,
+      "loss": 0.9126,
+      "step": 5781
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0733231273678116e-06,
+      "loss": 0.9399,
+      "step": 5782
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.072387881021656e-06,
+      "loss": 0.8625,
+      "step": 5783
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0714527343473553e-06,
+      "loss": 0.8162,
+      "step": 5784
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0705176874453934e-06,
+      "loss": 0.9331,
+      "step": 5785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0695827404162397e-06,
+      "loss": 0.8288,
+      "step": 5786
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0686478933603536e-06,
+      "loss": 0.9352,
+      "step": 5787
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0677131463781843e-06,
+      "loss": 0.96,
+      "step": 5788
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0667784995701716e-06,
+      "loss": 0.764,
+      "step": 5789
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.065843953036743e-06,
+      "loss": 1.0334,
+      "step": 5790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.064909506878314e-06,
+      "loss": 0.7106,
+      "step": 5791
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.063975161195292e-06,
+      "loss": 0.6875,
+      "step": 5792
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.063040916088072e-06,
+      "loss": 0.7823,
+      "step": 5793
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.06210677165704e-06,
+      "loss": 1.0071,
+      "step": 5794
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.061172728002568e-06,
+      "loss": 0.8239,
+      "step": 5795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0602387852250174e-06,
+      "loss": 0.8723,
+      "step": 5796
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.059304943424744e-06,
+      "loss": 0.9923,
+      "step": 5797
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0583712027020874e-06,
+      "loss": 0.8474,
+      "step": 5798
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.057437563157378e-06,
+      "loss": 0.7737,
+      "step": 5799
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0565040248909337e-06,
+      "loss": 1.19,
+      "step": 5800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.055570588003066e-06,
+      "loss": 1.0343,
+      "step": 5801
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.054637252594071e-06,
+      "loss": 0.8008,
+      "step": 5802
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0537040187642358e-06,
+      "loss": 0.8104,
+      "step": 5803
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.052770886613835e-06,
+      "loss": 0.927,
+      "step": 5804
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0518378562431367e-06,
+      "loss": 0.9942,
+      "step": 5805
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.050904927752392e-06,
+      "loss": 1.0359,
+      "step": 5806
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.049972101241846e-06,
+      "loss": 0.9274,
+      "step": 5807
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.049039376811728e-06,
+      "loss": 0.9198,
+      "step": 5808
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.048106754562263e-06,
+      "loss": 0.8861,
+      "step": 5809
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.047174234593659e-06,
+      "loss": 0.9958,
+      "step": 5810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0462418170061155e-06,
+      "loss": 0.7964,
+      "step": 5811
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0453095018998194e-06,
+      "loss": 0.8221,
+      "step": 5812
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0443772893749505e-06,
+      "loss": 0.891,
+      "step": 5813
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0434451795316736e-06,
+      "loss": 0.9647,
+      "step": 5814
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0425131724701442e-06,
+      "loss": 0.941,
+      "step": 5815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0415812682905047e-06,
+      "loss": 1.1199,
+      "step": 5816
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0406494670928903e-06,
+      "loss": 0.7921,
+      "step": 5817
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.039717768977423e-06,
+      "loss": 0.7692,
+      "step": 5818
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0387861740442125e-06,
+      "loss": 0.7491,
+      "step": 5819
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0378546823933576e-06,
+      "loss": 0.9522,
+      "step": 5820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.03692329412495e-06,
+      "loss": 0.7655,
+      "step": 5821
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0359920093390654e-06,
+      "loss": 0.889,
+      "step": 5822
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.03506082813577e-06,
+      "loss": 0.9084,
+      "step": 5823
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.03412975061512e-06,
+      "loss": 0.9185,
+      "step": 5824
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0331987768771594e-06,
+      "loss": 0.7785,
+      "step": 5825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.032267907021922e-06,
+      "loss": 0.8567,
+      "step": 5826
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0313371411494283e-06,
+      "loss": 1.1906,
+      "step": 5827
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.030406479359689e-06,
+      "loss": 0.787,
+      "step": 5828
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0294759217527053e-06,
+      "loss": 0.9943,
+      "step": 5829
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0285454684284647e-06,
+      "loss": 0.9325,
+      "step": 5830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0276151194869433e-06,
+      "loss": 1.0559,
+      "step": 5831
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0266848750281082e-06,
+      "loss": 0.9285,
+      "step": 5832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0257547351519145e-06,
+      "loss": 0.8734,
+      "step": 5833
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0248246999583047e-06,
+      "loss": 0.7703,
+      "step": 5834
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0238947695472106e-06,
+      "loss": 0.9302,
+      "step": 5835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.022964944018554e-06,
+      "loss": 0.8589,
+      "step": 5836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0220352234722453e-06,
+      "loss": 0.8343,
+      "step": 5837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.021105608008182e-06,
+      "loss": 0.8353,
+      "step": 5838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.02017609772625e-06,
+      "loss": 1.0552,
+      "step": 5839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.019246692726327e-06,
+      "loss": 0.9043,
+      "step": 5840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.018317393108277e-06,
+      "loss": 0.7668,
+      "step": 5841
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0173881989719535e-06,
+      "loss": 0.8291,
+      "step": 5842
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0164591104171974e-06,
+      "loss": 0.9041,
+      "step": 5843
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0155301275438398e-06,
+      "loss": 1.0043,
+      "step": 5844
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0146012504517e-06,
+      "loss": 0.9959,
+      "step": 5845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0136724792405854e-06,
+      "loss": 0.8879,
+      "step": 5846
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.012743814010292e-06,
+      "loss": 0.7254,
+      "step": 5847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.011815254860606e-06,
+      "loss": 0.6722,
+      "step": 5848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0108868018913e-06,
+      "loss": 0.7352,
+      "step": 5849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.009958455202137e-06,
+      "loss": 1.004,
+      "step": 5850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.009030214892867e-06,
+      "loss": 0.7231,
+      "step": 5851
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.008102081063229e-06,
+      "loss": 0.7372,
+      "step": 5852
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0071740538129524e-06,
+      "loss": 0.9621,
+      "step": 5853
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0062461332417537e-06,
+      "loss": 0.8341,
+      "step": 5854
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0053183194493355e-06,
+      "loss": 1.1004,
+      "step": 5855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0043906125353935e-06,
+      "loss": 0.9457,
+      "step": 5856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0034630125996096e-06,
+      "loss": 0.9368,
+      "step": 5857
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0025355197416532e-06,
+      "loss": 0.8397,
+      "step": 5858
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0016081340611843e-06,
+      "loss": 0.9884,
+      "step": 5859
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.0006808556578505e-06,
+      "loss": 0.6653,
+      "step": 5860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.999753684631287e-06,
+      "loss": 0.9909,
+      "step": 5861
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998826621081119e-06,
+      "loss": 0.9588,
+      "step": 5862
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9978996651069586e-06,
+      "loss": 0.9239,
+      "step": 5863
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9969728168084084e-06,
+      "loss": 0.7223,
+      "step": 5864
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.996046076285057e-06,
+      "loss": 1.0695,
+      "step": 5865
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.995119443636483e-06,
+      "loss": 1.0172,
+      "step": 5866
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9941929189622526e-06,
+      "loss": 0.923,
+      "step": 5867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9932665023619222e-06,
+      "loss": 0.8049,
+      "step": 5868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.992340193935034e-06,
+      "loss": 0.9733,
+      "step": 5869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9914139937811206e-06,
+      "loss": 0.7627,
+      "step": 5870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9904879019997004e-06,
+      "loss": 1.1354,
+      "step": 5871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9895619186902837e-06,
+      "loss": 0.8811,
+      "step": 5872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.988636043952367e-06,
+      "loss": 0.9765,
+      "step": 5873
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.987710277885435e-06,
+      "loss": 0.8455,
+      "step": 5874
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.986784620588961e-06,
+      "loss": 0.8408,
+      "step": 5875
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.985859072162408e-06,
+      "loss": 0.9446,
+      "step": 5876
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9849336327052247e-06,
+      "loss": 0.9313,
+      "step": 5877
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9840083023168507e-06,
+      "loss": 0.9063,
+      "step": 5878
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9830830810967113e-06,
+      "loss": 0.9788,
+      "step": 5879
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.982157969144223e-06,
+      "loss": 0.7999,
+      "step": 5880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9812329665587878e-06,
+      "loss": 0.9618,
+      "step": 5881
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.980308073439798e-06,
+      "loss": 0.9173,
+      "step": 5882
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9793832898866325e-06,
+      "loss": 0.8825,
+      "step": 5883
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.97845861599866e-06,
+      "loss": 0.7667,
+      "step": 5884
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9775340518752363e-06,
+      "loss": 0.7889,
+      "step": 5885
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.976609597615706e-06,
+      "loss": 0.9735,
+      "step": 5886
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.975685253319401e-06,
+      "loss": 0.8752,
+      "step": 5887
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9747610190856425e-06,
+      "loss": 0.9743,
+      "step": 5888
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9738368950137395e-06,
+      "loss": 0.8717,
+      "step": 5889
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9729128812029887e-06,
+      "loss": 1.1434,
+      "step": 5890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9719889777526757e-06,
+      "loss": 1.2185,
+      "step": 5891
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9710651847620743e-06,
+      "loss": 0.8891,
+      "step": 5892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.970141502330444e-06,
+      "loss": 0.9726,
+      "step": 5893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9692179305570366e-06,
+      "loss": 0.7808,
+      "step": 5894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.968294469541089e-06,
+      "loss": 0.8929,
+      "step": 5895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.967371119381827e-06,
+      "loss": 1.0235,
+      "step": 5896
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9664478801784647e-06,
+      "loss": 0.9623,
+      "step": 5897
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.965524752030203e-06,
+      "loss": 0.6864,
+      "step": 5898
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9646017350362342e-06,
+      "loss": 0.8736,
+      "step": 5899
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9636788292957345e-06,
+      "loss": 0.835,
+      "step": 5900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9627560349078703e-06,
+      "loss": 0.8484,
+      "step": 5901
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.961833351971796e-06,
+      "loss": 0.8363,
+      "step": 5902
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.960910780586655e-06,
+      "loss": 0.7197,
+      "step": 5903
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9599883208515763e-06,
+      "loss": 0.8618,
+      "step": 5904
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590659728656772e-06,
+      "loss": 0.7077,
+      "step": 5905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9581437367280657e-06,
+      "loss": 0.9388,
+      "step": 5906
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9572216125378356e-06,
+      "loss": 0.8907,
+      "step": 5907
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9562996003940686e-06,
+      "loss": 0.8687,
+      "step": 5908
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.955377700395835e-06,
+      "loss": 0.7465,
+      "step": 5909
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9544559126421927e-06,
+      "loss": 0.8004,
+      "step": 5910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9535342372321887e-06,
+      "loss": 0.8085,
+      "step": 5911
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.952612674264856e-06,
+      "loss": 0.9726,
+      "step": 5912
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.951691223839216e-06,
+      "loss": 0.9722,
+      "step": 5913
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.95076988605428e-06,
+      "loss": 0.9029,
+      "step": 5914
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.949848661009045e-06,
+      "loss": 1.0242,
+      "step": 5915
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9489275488024956e-06,
+      "loss": 0.9046,
+      "step": 5916
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9480065495336064e-06,
+      "loss": 0.7905,
+      "step": 5917
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.947085663301338e-06,
+      "loss": 0.9168,
+      "step": 5918
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9461648902046406e-06,
+      "loss": 1.0392,
+      "step": 5919
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.945244230342451e-06,
+      "loss": 1.1302,
+      "step": 5920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.944323683813693e-06,
+      "loss": 0.6724,
+      "step": 5921
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9434032507172794e-06,
+      "loss": 0.8838,
+      "step": 5922
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9424829311521115e-06,
+      "loss": 0.9155,
+      "step": 5923
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.941562725217078e-06,
+      "loss": 1.0448,
+      "step": 5924
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9406426330110534e-06,
+      "loss": 0.9446,
+      "step": 5925
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9397226546329014e-06,
+      "loss": 0.9246,
+      "step": 5926
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9388027901814757e-06,
+      "loss": 0.937,
+      "step": 5927
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.937883039755614e-06,
+      "loss": 0.8551,
+      "step": 5928
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9369634034541445e-06,
+      "loss": 0.8775,
+      "step": 5929
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9360438813758798e-06,
+      "loss": 0.9527,
+      "step": 5930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.935124473619626e-06,
+      "loss": 0.753,
+      "step": 5931
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.934205180284171e-06,
+      "loss": 0.765,
+      "step": 5932
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9332860014682933e-06,
+      "loss": 1.1864,
+      "step": 5933
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.932366937270757e-06,
+      "loss": 0.9378,
+      "step": 5934
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9314479877903194e-06,
+      "loss": 0.7995,
+      "step": 5935
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9305291531257185e-06,
+      "loss": 0.8935,
+      "step": 5936
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9296104333756843e-06,
+      "loss": 0.8831,
+      "step": 5937
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.928691828638931e-06,
+      "loss": 0.8184,
+      "step": 5938
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9277733390141653e-06,
+      "loss": 1.0694,
+      "step": 5939
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.926854964600078e-06,
+      "loss": 0.8065,
+      "step": 5940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9259367054953484e-06,
+      "loss": 0.9645,
+      "step": 5941
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.925018561798642e-06,
+      "loss": 1.0232,
+      "step": 5942
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9241005336086153e-06,
+      "loss": 0.9543,
+      "step": 5943
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9231826210239096e-06,
+      "loss": 0.8099,
+      "step": 5944
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.922264824143155e-06,
+      "loss": 1.0304,
+      "step": 5945
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.921347143064966e-06,
+      "loss": 1.2458,
+      "step": 5946
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9204295778879513e-06,
+      "loss": 0.9721,
+      "step": 5947
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.919512128710701e-06,
+      "loss": 0.7167,
+      "step": 5948
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9185947956317955e-06,
+      "loss": 1.1369,
+      "step": 5949
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9176775787498012e-06,
+      "loss": 0.9562,
+      "step": 5950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.916760478163275e-06,
+      "loss": 0.8168,
+      "step": 5951
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.915843493970758e-06,
+      "loss": 0.8894,
+      "step": 5952
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.91492662627078e-06,
+      "loss": 0.9239,
+      "step": 5953
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.914009875161857e-06,
+      "loss": 0.828,
+      "step": 5954
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9130932407424976e-06,
+      "loss": 1.1202,
+      "step": 5955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9121767231111913e-06,
+      "loss": 0.7607,
+      "step": 5956
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9112603223664172e-06,
+      "loss": 0.7591,
+      "step": 5957
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9103440386066454e-06,
+      "loss": 1.1535,
+      "step": 5958
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9094278719303288e-06,
+      "loss": 0.9375,
+      "step": 5959
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9085118224359093e-06,
+      "loss": 0.7997,
+      "step": 5960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9075958902218154e-06,
+      "loss": 0.9328,
+      "step": 5961
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9066800753864665e-06,
+      "loss": 0.8812,
+      "step": 5962
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9057643780282657e-06,
+      "loss": 0.7584,
+      "step": 5963
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9048487982456037e-06,
+      "loss": 0.9116,
+      "step": 5964
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9039333361368595e-06,
+      "loss": 0.754,
+      "step": 5965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9030179918004012e-06,
+      "loss": 0.843,
+      "step": 5966
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9021027653345818e-06,
+      "loss": 0.8722,
+      "step": 5967
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9011876568377416e-06,
+      "loss": 1.0289,
+      "step": 5968
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.900272666408208e-06,
+      "loss": 1.0252,
+      "step": 5969
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.899357794144299e-06,
+      "loss": 0.9011,
+      "step": 5970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8984430401443166e-06,
+      "loss": 0.9614,
+      "step": 5971
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.897528404506551e-06,
+      "loss": 0.8813,
+      "step": 5972
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8966138873292782e-06,
+      "loss": 1.0581,
+      "step": 5973
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.895699488710766e-06,
+      "loss": 0.8795,
+      "step": 5974
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.894785208749265e-06,
+      "loss": 0.8181,
+      "step": 5975
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.893871047543014e-06,
+      "loss": 0.8204,
+      "step": 5976
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8929570051902389e-06,
+      "loss": 0.8824,
+      "step": 5977
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8920430817891564e-06,
+      "loss": 0.9394,
+      "step": 5978
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.891129277437965e-06,
+      "loss": 0.8035,
+      "step": 5979
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.890215592234854e-06,
+      "loss": 0.7297,
+      "step": 5980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8893020262779973e-06,
+      "loss": 0.8863,
+      "step": 5981
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8883885796655595e-06,
+      "loss": 0.7729,
+      "step": 5982
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8874752524956896e-06,
+      "loss": 0.9723,
+      "step": 5983
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8865620448665247e-06,
+      "loss": 0.936,
+      "step": 5984
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8856489568761869e-06,
+      "loss": 0.9153,
+      "step": 5985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8847359886227907e-06,
+      "loss": 0.6754,
+      "step": 5986
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.883823140204433e-06,
+      "loss": 0.9466,
+      "step": 5987
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.882910411719199e-06,
+      "loss": 1.031,
+      "step": 5988
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8819978032651606e-06,
+      "loss": 1.0642,
+      "step": 5989
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8810853149403793e-06,
+      "loss": 0.758,
+      "step": 5990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8801729468429017e-06,
+      "loss": 1.1194,
+      "step": 5991
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8792606990707605e-06,
+      "loss": 0.9489,
+      "step": 5992
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8783485717219761e-06,
+      "loss": 1.109,
+      "step": 5993
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8774365648945586e-06,
+      "loss": 0.972,
+      "step": 5994
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8765246786865025e-06,
+      "loss": 0.9471,
+      "step": 5995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8756129131957894e-06,
+      "loss": 1.0747,
+      "step": 5996
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8747012685203876e-06,
+      "loss": 1.0877,
+      "step": 5997
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.873789744758255e-06,
+      "loss": 0.8866,
+      "step": 5998
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8728783420073343e-06,
+      "loss": 0.766,
+      "step": 5999
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.871967060365555e-06,
+      "loss": 0.9616,
+      "step": 6000
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0476,
+      "eval_samples_per_second": 351.882,
+      "eval_steps_per_second": 43.994,
+      "step": 6000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8710558999308333e-06,
+      "loss": 0.7334,
+      "step": 6001
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.870144860801076e-06,
+      "loss": 1.087,
+      "step": 6002
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8692339430741728e-06,
+      "loss": 1.171,
+      "step": 6003
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8683231468480018e-06,
+      "loss": 0.8325,
+      "step": 6004
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8674124722204266e-06,
+      "loss": 0.8344,
+      "step": 6005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866501919289302e-06,
+      "loss": 1.1737,
+      "step": 6006
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8655914881524643e-06,
+      "loss": 0.813,
+      "step": 6007
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.864681178907741e-06,
+      "loss": 0.794,
+      "step": 6008
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.863770991652943e-06,
+      "loss": 1.0518,
+      "step": 6009
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8628609264858716e-06,
+      "loss": 1.0793,
+      "step": 6010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8619509835043125e-06,
+      "loss": 1.0489,
+      "step": 6011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8610411628060394e-06,
+      "loss": 0.7703,
+      "step": 6012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8601314644888108e-06,
+      "loss": 0.9001,
+      "step": 6013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.859221888650376e-06,
+      "loss": 0.9587,
+      "step": 6014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8583124353884675e-06,
+      "loss": 0.8702,
+      "step": 6015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.857403104800807e-06,
+      "loss": 1.1964,
+      "step": 6016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8564938969851e-06,
+      "loss": 0.9816,
+      "step": 6017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8555848120390434e-06,
+      "loss": 0.6898,
+      "step": 6018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8546758500603169e-06,
+      "loss": 0.9403,
+      "step": 6019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8537670111465888e-06,
+      "loss": 0.8586,
+      "step": 6020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.852858295395512e-06,
+      "loss": 0.859,
+      "step": 6021
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8519497029047308e-06,
+      "loss": 0.9582,
+      "step": 6022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8510412337718722e-06,
+      "loss": 0.9122,
+      "step": 6023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8501328880945498e-06,
+      "loss": 0.7779,
+      "step": 6024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8492246659703682e-06,
+      "loss": 0.8311,
+      "step": 6025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8483165674969134e-06,
+      "loss": 0.818,
+      "step": 6026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.847408592771762e-06,
+      "loss": 1.0137,
+      "step": 6027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8465007418924734e-06,
+      "loss": 0.8189,
+      "step": 6028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.845593014956599e-06,
+      "loss": 1.1313,
+      "step": 6029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8446854120616724e-06,
+      "loss": 0.8797,
+      "step": 6030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.843777933305216e-06,
+      "loss": 0.8788,
+      "step": 6031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8428705787847374e-06,
+      "loss": 1.2206,
+      "step": 6032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8419633485977335e-06,
+      "loss": 0.9204,
+      "step": 6033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8410562428416851e-06,
+      "loss": 1.0369,
+      "step": 6034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8401492616140608e-06,
+      "loss": 0.7957,
+      "step": 6035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8392424050123143e-06,
+      "loss": 0.9675,
+      "step": 6036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8383356731338896e-06,
+      "loss": 0.7549,
+      "step": 6037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.837429066076214e-06,
+      "loss": 1.0352,
+      "step": 6038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8365225839367026e-06,
+      "loss": 0.7648,
+      "step": 6039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8356162268127553e-06,
+      "loss": 0.818,
+      "step": 6040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.834709994801762e-06,
+      "loss": 1.0697,
+      "step": 6041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.833803888001097e-06,
+      "loss": 0.823,
+      "step": 6042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8328979065081212e-06,
+      "loss": 1.0941,
+      "step": 6043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8319920504201806e-06,
+      "loss": 0.8946,
+      "step": 6044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.831086319834612e-06,
+      "loss": 0.8262,
+      "step": 6045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.830180714848735e-06,
+      "loss": 0.9893,
+      "step": 6046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8292752355598569e-06,
+      "loss": 1.1304,
+      "step": 6047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8283698820652698e-06,
+      "loss": 0.9088,
+      "step": 6048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8274646544622566e-06,
+      "loss": 0.9102,
+      "step": 6049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8265595528480822e-06,
+      "loss": 1.0948,
+      "step": 6050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8256545773200006e-06,
+      "loss": 0.8512,
+      "step": 6051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8247497279752491e-06,
+      "loss": 1.0994,
+      "step": 6052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8238450049110576e-06,
+      "loss": 0.7229,
+      "step": 6053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8229404082246355e-06,
+      "loss": 0.9562,
+      "step": 6054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8220359380131828e-06,
+      "loss": 0.904,
+      "step": 6055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8211315943738834e-06,
+      "loss": 1.0957,
+      "step": 6056
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8202273774039113e-06,
+      "loss": 0.8709,
+      "step": 6057
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8193232872004234e-06,
+      "loss": 1.009,
+      "step": 6058
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.818419323860564e-06,
+      "loss": 1.1701,
+      "step": 6059
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8175154874814623e-06,
+      "loss": 0.8199,
+      "step": 6060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.816611778160239e-06,
+      "loss": 0.8028,
+      "step": 6061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.815708195993995e-06,
+      "loss": 0.8552,
+      "step": 6062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8148047410798213e-06,
+      "loss": 0.7969,
+      "step": 6063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.813901413514792e-06,
+      "loss": 0.7576,
+      "step": 6064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8129982133959729e-06,
+      "loss": 0.9265,
+      "step": 6065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8120951408204106e-06,
+      "loss": 1.1469,
+      "step": 6066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8111921958851408e-06,
+      "loss": 0.7657,
+      "step": 6067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8102893786871835e-06,
+      "loss": 0.7063,
+      "step": 6068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8093866893235489e-06,
+      "loss": 0.9192,
+      "step": 6069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8084841278912295e-06,
+      "loss": 1.0966,
+      "step": 6070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8075816944872055e-06,
+      "loss": 0.8378,
+      "step": 6071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8066793892084422e-06,
+      "loss": 0.8692,
+      "step": 6072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8057772121518945e-06,
+      "loss": 0.9061,
+      "step": 6073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8048751634145e-06,
+      "loss": 0.8315,
+      "step": 6074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8039732430931842e-06,
+      "loss": 0.7282,
+      "step": 6075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8030714512848568e-06,
+      "loss": 0.8962,
+      "step": 6076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8021697880864179e-06,
+      "loss": 0.8811,
+      "step": 6077
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8012682535947496e-06,
+      "loss": 1.0578,
+      "step": 6078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8003668479067217e-06,
+      "loss": 1.1296,
+      "step": 6079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7994655711191894e-06,
+      "loss": 0.867,
+      "step": 6080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7985644233289974e-06,
+      "loss": 0.8472,
+      "step": 6081
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7976634046329718e-06,
+      "loss": 1.0592,
+      "step": 6082
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7967625151279276e-06,
+      "loss": 0.9746,
+      "step": 6083
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7958617549106646e-06,
+      "loss": 0.8077,
+      "step": 6084
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7949611240779706e-06,
+      "loss": 0.7412,
+      "step": 6085
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7940606227266183e-06,
+      "loss": 0.9674,
+      "step": 6086
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7931602509533661e-06,
+      "loss": 0.7623,
+      "step": 6087
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7922600088549577e-06,
+      "loss": 0.8917,
+      "step": 6088
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.791359896528126e-06,
+      "loss": 1.0034,
+      "step": 6089
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.790459914069587e-06,
+      "loss": 0.9192,
+      "step": 6090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7895600615760432e-06,
+      "loss": 0.8832,
+      "step": 6091
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7886603391441853e-06,
+      "loss": 0.9643,
+      "step": 6092
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7877607468706878e-06,
+      "loss": 0.7251,
+      "step": 6093
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7868612848522114e-06,
+      "loss": 0.898,
+      "step": 6094
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7859619531854023e-06,
+      "loss": 1.0841,
+      "step": 6095
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7850627519668955e-06,
+      "loss": 1.232,
+      "step": 6096
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7841636812933096e-06,
+      "loss": 0.8786,
+      "step": 6097
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7832647412612496e-06,
+      "loss": 1.0893,
+      "step": 6098
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7823659319673047e-06,
+      "loss": 0.8073,
+      "step": 6099
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.781467253508055e-06,
+      "loss": 0.7899,
+      "step": 6100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7805687059800616e-06,
+      "loss": 0.7355,
+      "step": 6101
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7796702894798744e-06,
+      "loss": 0.992,
+      "step": 6102
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.778772004104026e-06,
+      "loss": 0.7668,
+      "step": 6103
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7778738499490395e-06,
+      "loss": 1.0202,
+      "step": 6104
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7769758271114208e-06,
+      "loss": 1.2435,
+      "step": 6105
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7760779356876624e-06,
+      "loss": 0.8656,
+      "step": 6106
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7751801757742408e-06,
+      "loss": 0.9971,
+      "step": 6107
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7742825474676236e-06,
+      "loss": 1.0055,
+      "step": 6108
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7733850508642589e-06,
+      "loss": 0.7929,
+      "step": 6109
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7724876860605832e-06,
+      "loss": 1.1683,
+      "step": 6110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7715904531530164e-06,
+      "loss": 0.9933,
+      "step": 6111
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7706933522379693e-06,
+      "loss": 0.9211,
+      "step": 6112
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.769796383411834e-06,
+      "loss": 0.9322,
+      "step": 6113
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7688995467709895e-06,
+      "loss": 1.0455,
+      "step": 6114
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.768002842411799e-06,
+      "loss": 0.9758,
+      "step": 6115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.767106270430617e-06,
+      "loss": 0.8386,
+      "step": 6116
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7662098309237782e-06,
+      "loss": 0.8856,
+      "step": 6117
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7653135239876045e-06,
+      "loss": 0.9046,
+      "step": 6118
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7644173497184042e-06,
+      "loss": 0.8257,
+      "step": 6119
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7635213082124722e-06,
+      "loss": 0.9017,
+      "step": 6120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7626253995660871e-06,
+      "loss": 0.821,
+      "step": 6121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7617296238755153e-06,
+      "loss": 1.0106,
+      "step": 6122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.760833981237005e-06,
+      "loss": 0.991,
+      "step": 6123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7599384717467967e-06,
+      "loss": 0.8135,
+      "step": 6124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7590430955011112e-06,
+      "loss": 0.9882,
+      "step": 6125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.758147852596156e-06,
+      "loss": 0.8935,
+      "step": 6126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7572527431281253e-06,
+      "loss": 0.821,
+      "step": 6127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7563577671931987e-06,
+      "loss": 0.7991,
+      "step": 6128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7554629248875418e-06,
+      "loss": 0.6532,
+      "step": 6129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7545682163073046e-06,
+      "loss": 0.9341,
+      "step": 6130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7536736415486233e-06,
+      "loss": 0.9226,
+      "step": 6131
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7527792007076201e-06,
+      "loss": 0.8733,
+      "step": 6132
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7518848938804033e-06,
+      "loss": 0.9524,
+      "step": 6133
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7509907211630659e-06,
+      "loss": 0.9187,
+      "step": 6134
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7500966826516858e-06,
+      "loss": 0.8043,
+      "step": 6135
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7492027784423274e-06,
+      "loss": 1.039,
+      "step": 6136
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.748309008631043e-06,
+      "loss": 0.9259,
+      "step": 6137
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.747415373313865e-06,
+      "loss": 0.9261,
+      "step": 6138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7465218725868156e-06,
+      "loss": 1.0159,
+      "step": 6139
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7456285065459014e-06,
+      "loss": 0.7694,
+      "step": 6140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447352752871164e-06,
+      "loss": 0.9484,
+      "step": 6141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7438421789064348e-06,
+      "loss": 0.7444,
+      "step": 6142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7429492174998216e-06,
+      "loss": 1.0403,
+      "step": 6143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7420563911632253e-06,
+      "loss": 0.9108,
+      "step": 6144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7411636999925815e-06,
+      "loss": 0.8635,
+      "step": 6145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7402711440838063e-06,
+      "loss": 0.8406,
+      "step": 6146
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7393787235328077e-06,
+      "loss": 0.8295,
+      "step": 6147
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7384864384354748e-06,
+      "loss": 1.1513,
+      "step": 6148
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7375942888876854e-06,
+      "loss": 0.9351,
+      "step": 6149
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7367022749852982e-06,
+      "loss": 1.0673,
+      "step": 6150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7358103968241616e-06,
+      "loss": 0.8407,
+      "step": 6151
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7349186545001075e-06,
+      "loss": 0.7166,
+      "step": 6152
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7340270481089548e-06,
+      "loss": 0.8415,
+      "step": 6153
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7331355777465045e-06,
+      "loss": 1.139,
+      "step": 6154
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7322442435085459e-06,
+      "loss": 1.1235,
+      "step": 6155
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7313530454908531e-06,
+      "loss": 0.9599,
+      "step": 6156
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.730461983789186e-06,
+      "loss": 0.6763,
+      "step": 6157
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7295710584992857e-06,
+      "loss": 0.7838,
+      "step": 6158
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.728680269716887e-06,
+      "loss": 1.0594,
+      "step": 6159
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7277896175377013e-06,
+      "loss": 0.9367,
+      "step": 6160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7268991020574316e-06,
+      "loss": 0.8286,
+      "step": 6161
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.72600872337176e-06,
+      "loss": 0.7283,
+      "step": 6162
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7251184815763632e-06,
+      "loss": 0.736,
+      "step": 6163
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.724228376766893e-06,
+      "loss": 1.093,
+      "step": 6164
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7233384090389933e-06,
+      "loss": 0.8445,
+      "step": 6165
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.722448578488289e-06,
+      "loss": 1.0559,
+      "step": 6166
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7215588852103958e-06,
+      "loss": 0.829,
+      "step": 6167
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.720669329300908e-06,
+      "loss": 0.9469,
+      "step": 6168
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7197799108554107e-06,
+      "loss": 0.9053,
+      "step": 6169
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7188906299694677e-06,
+      "loss": 0.8337,
+      "step": 6170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7180014867386381e-06,
+      "loss": 1.031,
+      "step": 6171
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7171124812584555e-06,
+      "loss": 0.9751,
+      "step": 6172
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7162236136244473e-06,
+      "loss": 0.8649,
+      "step": 6173
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7153348839321173e-06,
+      "loss": 0.8392,
+      "step": 6174
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7144462922769648e-06,
+      "loss": 0.8508,
+      "step": 6175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.713557838754465e-06,
+      "loss": 1.0268,
+      "step": 6176
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7126695234600848e-06,
+      "loss": 0.9356,
+      "step": 6177
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7117813464892707e-06,
+      "loss": 1.0111,
+      "step": 6178
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7108933079374607e-06,
+      "loss": 0.9264,
+      "step": 6179
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.710005407900072e-06,
+      "loss": 0.8498,
+      "step": 6180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7091176464725115e-06,
+      "loss": 0.9505,
+      "step": 6181
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7082300237501652e-06,
+      "loss": 0.9764,
+      "step": 6182
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7073425398284126e-06,
+      "loss": 0.7815,
+      "step": 6183
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7064551948026111e-06,
+      "loss": 0.6727,
+      "step": 6184
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7055679887681078e-06,
+      "loss": 0.8685,
+      "step": 6185
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7046809218202295e-06,
+      "loss": 1.008,
+      "step": 6186
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7037939940542956e-06,
+      "loss": 1.0032,
+      "step": 6187
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.702907205565604e-06,
+      "loss": 0.7433,
+      "step": 6188
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7020205564494412e-06,
+      "loss": 1.1251,
+      "step": 6189
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.701134046801075e-06,
+      "loss": 0.9687,
+      "step": 6190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.700247676715765e-06,
+      "loss": 0.8505,
+      "step": 6191
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.699361446288748e-06,
+      "loss": 0.9139,
+      "step": 6192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6984753556152523e-06,
+      "loss": 0.8837,
+      "step": 6193
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6975894047904846e-06,
+      "loss": 0.9306,
+      "step": 6194
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6967035939096441e-06,
+      "loss": 0.9663,
+      "step": 6195
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6958179230679086e-06,
+      "loss": 0.9366,
+      "step": 6196
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6949323923604456e-06,
+      "loss": 0.8347,
+      "step": 6197
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6940470018824014e-06,
+      "loss": 1.0598,
+      "step": 6198
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6931617517289157e-06,
+      "loss": 0.78,
+      "step": 6199
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.692276641995106e-06,
+      "loss": 0.9287,
+      "step": 6200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6913916727760786e-06,
+      "loss": 0.9251,
+      "step": 6201
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.69050684416692e-06,
+      "loss": 0.8396,
+      "step": 6202
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6896221562627102e-06,
+      "loss": 0.7182,
+      "step": 6203
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6887376091585048e-06,
+      "loss": 0.7879,
+      "step": 6204
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6878532029493512e-06,
+      "loss": 0.8719,
+      "step": 6205
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6869689377302747e-06,
+      "loss": 0.8647,
+      "step": 6206
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6860848135962946e-06,
+      "loss": 0.9297,
+      "step": 6207
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6852008306424065e-06,
+      "loss": 1.0546,
+      "step": 6208
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6843169889635962e-06,
+      "loss": 0.8382,
+      "step": 6209
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6834332886548292e-06,
+      "loss": 0.94,
+      "step": 6210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.682549729811064e-06,
+      "loss": 0.8753,
+      "step": 6211
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6816663125272354e-06,
+      "loss": 0.9137,
+      "step": 6212
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6807830368982682e-06,
+      "loss": 0.9677,
+      "step": 6213
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6798999030190669e-06,
+      "loss": 0.8722,
+      "step": 6214
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6790169109845298e-06,
+      "loss": 0.8263,
+      "step": 6215
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6781340608895298e-06,
+      "loss": 1.0007,
+      "step": 6216
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6772513528289318e-06,
+      "loss": 0.7785,
+      "step": 6217
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6763687868975792e-06,
+      "loss": 1.1013,
+      "step": 6218
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6754863631903085e-06,
+      "loss": 1.066,
+      "step": 6219
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6746040818019326e-06,
+      "loss": 0.8698,
+      "step": 6220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.673721942827254e-06,
+      "loss": 0.8136,
+      "step": 6221
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6728399463610566e-06,
+      "loss": 0.7275,
+      "step": 6222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6719580924981137e-06,
+      "loss": 1.0515,
+      "step": 6223
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6710763813331785e-06,
+      "loss": 0.8105,
+      "step": 6224
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.670194812960991e-06,
+      "loss": 1.0229,
+      "step": 6225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6693133874762758e-06,
+      "loss": 0.6953,
+      "step": 6226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6684321049737438e-06,
+      "loss": 0.8671,
+      "step": 6227
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6675509655480854e-06,
+      "loss": 0.9071,
+      "step": 6228
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.666669969293981e-06,
+      "loss": 0.9225,
+      "step": 6229
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.665789116306093e-06,
+      "loss": 0.7641,
+      "step": 6230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6649084066790707e-06,
+      "loss": 0.91,
+      "step": 6231
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6640278405075437e-06,
+      "loss": 0.8002,
+      "step": 6232
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6631474178861296e-06,
+      "loss": 1.0139,
+      "step": 6233
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6622671389094302e-06,
+      "loss": 1.0736,
+      "step": 6234
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6613870036720324e-06,
+      "loss": 0.9638,
+      "step": 6235
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6605070122685046e-06,
+      "loss": 0.7511,
+      "step": 6236
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6596271647934024e-06,
+      "loss": 0.9495,
+      "step": 6237
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.658747461341266e-06,
+      "loss": 0.7576,
+      "step": 6238
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.65786790200662e-06,
+      "loss": 0.9337,
+      "step": 6239
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6569884868839716e-06,
+      "loss": 0.8762,
+      "step": 6240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6561092160678138e-06,
+      "loss": 0.8497,
+      "step": 6241
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6552300896526252e-06,
+      "loss": 0.9074,
+      "step": 6242
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6543511077328681e-06,
+      "loss": 0.8182,
+      "step": 6243
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.653472270402988e-06,
+      "loss": 0.7974,
+      "step": 6244
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6525935777574162e-06,
+      "loss": 0.8408,
+      "step": 6245
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.651715029890568e-06,
+      "loss": 0.7185,
+      "step": 6246
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6508366268968449e-06,
+      "loss": 0.8391,
+      "step": 6247
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.649958368870629e-06,
+      "loss": 1.1535,
+      "step": 6248
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6490802559062895e-06,
+      "loss": 0.8201,
+      "step": 6249
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.648202288098181e-06,
+      "loss": 1.0687,
+      "step": 6250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6473244655406408e-06,
+      "loss": 1.02,
+      "step": 6251
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6464467883279893e-06,
+      "loss": 0.8659,
+      "step": 6252
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6455692565545332e-06,
+      "loss": 0.7524,
+      "step": 6253
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6446918703145646e-06,
+      "loss": 0.9569,
+      "step": 6254
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6438146297023587e-06,
+      "loss": 0.8073,
+      "step": 6255
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6429375348121726e-06,
+      "loss": 0.9109,
+      "step": 6256
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.642060585738252e-06,
+      "loss": 0.8413,
+      "step": 6257
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6411837825748244e-06,
+      "loss": 0.8455,
+      "step": 6258
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6403071254161038e-06,
+      "loss": 0.9941,
+      "step": 6259
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.639430614356284e-06,
+      "loss": 1.1455,
+      "step": 6260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6385542494895476e-06,
+      "loss": 0.8294,
+      "step": 6261
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6376780309100604e-06,
+      "loss": 0.9929,
+      "step": 6262
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6368019587119723e-06,
+      "loss": 0.8904,
+      "step": 6263
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6359260329894152e-06,
+      "loss": 0.9965,
+      "step": 6264
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6350502538365087e-06,
+      "loss": 0.7961,
+      "step": 6265
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6341746213473552e-06,
+      "loss": 1.1837,
+      "step": 6266
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6332991356160422e-06,
+      "loss": 1.0118,
+      "step": 6267
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6324237967366383e-06,
+      "loss": 0.8274,
+      "step": 6268
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6315486048031997e-06,
+      "loss": 0.9059,
+      "step": 6269
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6306735599097663e-06,
+      "loss": 0.7645,
+      "step": 6270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.629798662150362e-06,
+      "loss": 1.0571,
+      "step": 6271
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6289239116189926e-06,
+      "loss": 1.0293,
+      "step": 6272
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6280493084096512e-06,
+      "loss": 0.9126,
+      "step": 6273
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.627174852616314e-06,
+      "loss": 0.8203,
+      "step": 6274
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6263005443329416e-06,
+      "loss": 0.8327,
+      "step": 6275
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6254263836534764e-06,
+      "loss": 0.7758,
+      "step": 6276
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6245523706718484e-06,
+      "loss": 1.1886,
+      "step": 6277
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6236785054819703e-06,
+      "loss": 0.7934,
+      "step": 6278
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6228047881777394e-06,
+      "loss": 0.7606,
+      "step": 6279
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6219312188530346e-06,
+      "loss": 0.9329,
+      "step": 6280
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.621057797601722e-06,
+      "loss": 0.9313,
+      "step": 6281
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6201845245176506e-06,
+      "loss": 0.9242,
+      "step": 6282
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6193113996946547e-06,
+      "loss": 0.839,
+      "step": 6283
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6184384232265495e-06,
+      "loss": 0.7903,
+      "step": 6284
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.617565595207137e-06,
+      "loss": 0.9539,
+      "step": 6285
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6166929157302025e-06,
+      "loss": 0.8768,
+      "step": 6286
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.615820384889517e-06,
+      "loss": 1.0711,
+      "step": 6287
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.614948002778831e-06,
+      "loss": 0.8161,
+      "step": 6288
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6140757694918833e-06,
+      "loss": 0.7887,
+      "step": 6289
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6132036851223952e-06,
+      "loss": 0.8481,
+      "step": 6290
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6123317497640736e-06,
+      "loss": 0.7175,
+      "step": 6291
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6114599635106047e-06,
+      "loss": 0.7986,
+      "step": 6292
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.610588326455664e-06,
+      "loss": 0.8596,
+      "step": 6293
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6097168386929086e-06,
+      "loss": 0.883,
+      "step": 6294
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6088455003159806e-06,
+      "loss": 1.02,
+      "step": 6295
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6079743114185035e-06,
+      "loss": 0.8691,
+      "step": 6296
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.607103272094087e-06,
+      "loss": 0.7803,
+      "step": 6297
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6062323824363248e-06,
+      "loss": 0.7346,
+      "step": 6298
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6053616425387944e-06,
+      "loss": 1.0815,
+      "step": 6299
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.604491052495055e-06,
+      "loss": 0.8306,
+      "step": 6300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6036206123986525e-06,
+      "loss": 0.9098,
+      "step": 6301
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6027503223431152e-06,
+      "loss": 0.9758,
+      "step": 6302
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.601880182421958e-06,
+      "loss": 0.9194,
+      "step": 6303
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6010101927286736e-06,
+      "loss": 0.8249,
+      "step": 6304
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6001403533567445e-06,
+      "loss": 1.0197,
+      "step": 6305
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5992706643996341e-06,
+      "loss": 0.972,
+      "step": 6306
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5984011259507925e-06,
+      "loss": 0.8045,
+      "step": 6307
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5975317381036486e-06,
+      "loss": 1.1157,
+      "step": 6308
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5966625009516195e-06,
+      "loss": 1.0806,
+      "step": 6309
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5957934145881044e-06,
+      "loss": 0.7344,
+      "step": 6310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5949244791064879e-06,
+      "loss": 1.0892,
+      "step": 6311
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5940556946001344e-06,
+      "loss": 1.0712,
+      "step": 6312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5931870611623964e-06,
+      "loss": 0.8763,
+      "step": 6313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5923185788866083e-06,
+      "loss": 0.9586,
+      "step": 6314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5914502478660899e-06,
+      "loss": 0.8999,
+      "step": 6315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5905820681941402e-06,
+      "loss": 0.9955,
+      "step": 6316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.589714039964047e-06,
+      "loss": 1.2504,
+      "step": 6317
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5888461632690797e-06,
+      "loss": 0.9251,
+      "step": 6318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5879784382024921e-06,
+      "loss": 0.8448,
+      "step": 6319
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5871108648575197e-06,
+      "loss": 1.0182,
+      "step": 6320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.586243443327384e-06,
+      "loss": 0.9388,
+      "step": 6321
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5853761737052894e-06,
+      "loss": 0.8286,
+      "step": 6322
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5845090560844253e-06,
+      "loss": 0.6769,
+      "step": 6323
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5836420905579611e-06,
+      "loss": 0.8509,
+      "step": 6324
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5827752772190528e-06,
+      "loss": 0.9496,
+      "step": 6325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5819086161608399e-06,
+      "loss": 1.0023,
+      "step": 6326
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5810421074764464e-06,
+      "loss": 1.004,
+      "step": 6327
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.580175751258976e-06,
+      "loss": 0.9957,
+      "step": 6328
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5793095476015197e-06,
+      "loss": 1.1213,
+      "step": 6329
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.578443496597151e-06,
+      "loss": 0.7954,
+      "step": 6330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5775775983389286e-06,
+      "loss": 0.8017,
+      "step": 6331
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5767118529198906e-06,
+      "loss": 0.7391,
+      "step": 6332
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5758462604330622e-06,
+      "loss": 0.7956,
+      "step": 6333
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5749808209714518e-06,
+      "loss": 0.952,
+      "step": 6334
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5741155346280511e-06,
+      "loss": 1.0083,
+      "step": 6335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5732504014958336e-06,
+      "loss": 0.7832,
+      "step": 6336
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5723854216677588e-06,
+      "loss": 0.9005,
+      "step": 6337
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5715205952367685e-06,
+      "loss": 0.9964,
+      "step": 6338
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5706559222957897e-06,
+      "loss": 0.9652,
+      "step": 6339
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5697914029377283e-06,
+      "loss": 0.7807,
+      "step": 6340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.568927037255479e-06,
+      "loss": 0.9763,
+      "step": 6341
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5680628253419173e-06,
+      "loss": 0.8453,
+      "step": 6342
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5671987672899039e-06,
+      "loss": 0.7279,
+      "step": 6343
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5663348631922797e-06,
+      "loss": 0.7467,
+      "step": 6344
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5654711131418725e-06,
+      "loss": 1.1367,
+      "step": 6345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5646075172314916e-06,
+      "loss": 0.98,
+      "step": 6346
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5637440755539317e-06,
+      "loss": 0.8268,
+      "step": 6347
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5628807882019666e-06,
+      "loss": 0.8519,
+      "step": 6348
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5620176552683591e-06,
+      "loss": 0.8129,
+      "step": 6349
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5611546768458516e-06,
+      "loss": 0.8017,
+      "step": 6350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5602918530271723e-06,
+      "loss": 1.0613,
+      "step": 6351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.55942918390503e-06,
+      "loss": 0.8492,
+      "step": 6352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5585666695721186e-06,
+      "loss": 0.9083,
+      "step": 6353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5577043101211171e-06,
+      "loss": 1.0572,
+      "step": 6354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5568421056446823e-06,
+      "loss": 0.9786,
+      "step": 6355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5559800562354628e-06,
+      "loss": 0.8268,
+      "step": 6356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5551181619860817e-06,
+      "loss": 0.8307,
+      "step": 6357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5542564229891524e-06,
+      "loss": 1.0004,
+      "step": 6358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5533948393372644e-06,
+      "loss": 0.8556,
+      "step": 6359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5525334111230005e-06,
+      "loss": 0.8365,
+      "step": 6360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.551672138438917e-06,
+      "loss": 0.9705,
+      "step": 6361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5508110213775604e-06,
+      "loss": 0.8012,
+      "step": 6362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5499500600314534e-06,
+      "loss": 0.898,
+      "step": 6363
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.549089254493111e-06,
+      "loss": 0.7168,
+      "step": 6364
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5482286048550241e-06,
+      "loss": 0.9309,
+      "step": 6365
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5473681112096712e-06,
+      "loss": 0.8917,
+      "step": 6366
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5465077736495083e-06,
+      "loss": 1.0472,
+      "step": 6367
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.545647592266984e-06,
+      "loss": 0.763,
+      "step": 6368
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.544787567154521e-06,
+      "loss": 1.0612,
+      "step": 6369
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5439276984045311e-06,
+      "loss": 0.8555,
+      "step": 6370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5430679861094038e-06,
+      "loss": 0.7581,
+      "step": 6371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.54220843036152e-06,
+      "loss": 0.8629,
+      "step": 6372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.541349031253235e-06,
+      "loss": 0.6856,
+      "step": 6373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5404897888768935e-06,
+      "loss": 1.0474,
+      "step": 6374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5396307033248176e-06,
+      "loss": 0.8698,
+      "step": 6375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.538771774689321e-06,
+      "loss": 0.8298,
+      "step": 6376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5379130030626918e-06,
+      "loss": 0.7206,
+      "step": 6377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5370543885372074e-06,
+      "loss": 1.03,
+      "step": 6378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5361959312051221e-06,
+      "loss": 0.7159,
+      "step": 6379
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5353376311586816e-06,
+      "loss": 1.1282,
+      "step": 6380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5344794884901072e-06,
+      "loss": 0.8819,
+      "step": 6381
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5336215032916082e-06,
+      "loss": 0.9351,
+      "step": 6382
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5327636756553723e-06,
+      "loss": 0.8408,
+      "step": 6383
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5319060056735762e-06,
+      "loss": 0.952,
+      "step": 6384
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5310484934383741e-06,
+      "loss": 0.9616,
+      "step": 6385
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5301911390419081e-06,
+      "loss": 1.099,
+      "step": 6386
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5293339425762969e-06,
+      "loss": 0.9369,
+      "step": 6387
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.528476904133651e-06,
+      "loss": 0.8772,
+      "step": 6388
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5276200238060553e-06,
+      "loss": 1.0025,
+      "step": 6389
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5267633016855841e-06,
+      "loss": 0.8877,
+      "step": 6390
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.525906737864289e-06,
+      "loss": 0.6582,
+      "step": 6391
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5250503324342118e-06,
+      "loss": 0.889,
+      "step": 6392
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.52419408548737e-06,
+      "loss": 1.0023,
+      "step": 6393
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5233379971157697e-06,
+      "loss": 0.7995,
+      "step": 6394
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5224820674113938e-06,
+      "loss": 1.1307,
+      "step": 6395
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.521626296466217e-06,
+      "loss": 0.8896,
+      "step": 6396
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5207706843721872e-06,
+      "loss": 0.9437,
+      "step": 6397
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5199152312212434e-06,
+      "loss": 0.9501,
+      "step": 6398
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5190599371053002e-06,
+      "loss": 0.7327,
+      "step": 6399
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5182048021162633e-06,
+      "loss": 0.9795,
+      "step": 6400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5173498263460136e-06,
+      "loss": 1.0659,
+      "step": 6401
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5164950098864206e-06,
+      "loss": 1.0153,
+      "step": 6402
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5156403528293304e-06,
+      "loss": 1.0047,
+      "step": 6403
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5147858552665805e-06,
+      "loss": 0.919,
+      "step": 6404
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5139315172899833e-06,
+      "loss": 0.851,
+      "step": 6405
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5130773389913398e-06,
+      "loss": 0.9996,
+      "step": 6406
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5122233204624279e-06,
+      "loss": 0.9605,
+      "step": 6407
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5113694617950164e-06,
+      "loss": 0.8986,
+      "step": 6408
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5105157630808488e-06,
+      "loss": 0.7347,
+      "step": 6409
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5096622244116568e-06,
+      "loss": 0.8675,
+      "step": 6410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5088088458791508e-06,
+      "loss": 0.7405,
+      "step": 6411
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5079556275750302e-06,
+      "loss": 0.7895,
+      "step": 6412
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5071025695909696e-06,
+      "loss": 0.9907,
+      "step": 6413
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5062496720186322e-06,
+      "loss": 0.6518,
+      "step": 6414
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5053969349496592e-06,
+      "loss": 0.9494,
+      "step": 6415
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.504544358475681e-06,
+      "loss": 0.7102,
+      "step": 6416
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5036919426883037e-06,
+      "loss": 1.0554,
+      "step": 6417
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5028396876791215e-06,
+      "loss": 0.9413,
+      "step": 6418
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5019875935397054e-06,
+      "loss": 0.8175,
+      "step": 6419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5011356603616182e-06,
+      "loss": 0.963,
+      "step": 6420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5002838882363962e-06,
+      "loss": 0.9991,
+      "step": 6421
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4994322772555628e-06,
+      "loss": 0.8045,
+      "step": 6422
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4985808275106247e-06,
+      "loss": 0.688,
+      "step": 6423
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4977295390930705e-06,
+      "loss": 1.0202,
+      "step": 6424
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4968784120943688e-06,
+      "loss": 0.7427,
+      "step": 6425
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4960274466059748e-06,
+      "loss": 0.6579,
+      "step": 6426
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4951766427193244e-06,
+      "loss": 0.9227,
+      "step": 6427
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4943260005258371e-06,
+      "loss": 0.7788,
+      "step": 6428
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4934755201169125e-06,
+      "loss": 0.7915,
+      "step": 6429
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.492625201583936e-06,
+      "loss": 1.121,
+      "step": 6430
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.491775045018274e-06,
+      "loss": 0.8462,
+      "step": 6431
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.490925050511277e-06,
+      "loss": 0.9979,
+      "step": 6432
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4900752181542744e-06,
+      "loss": 0.7653,
+      "step": 6433
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4892255480385815e-06,
+      "loss": 0.9326,
+      "step": 6434
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4883760402554962e-06,
+      "loss": 1.0958,
+      "step": 6435
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4875266948962985e-06,
+      "loss": 0.7471,
+      "step": 6436
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4866775120522482e-06,
+      "loss": 0.8245,
+      "step": 6437
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4858284918145913e-06,
+      "loss": 0.8288,
+      "step": 6438
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.484979634274555e-06,
+      "loss": 0.8852,
+      "step": 6439
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4841309395233501e-06,
+      "loss": 0.7825,
+      "step": 6440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4832824076521661e-06,
+      "loss": 0.8879,
+      "step": 6441
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4824340387521794e-06,
+      "loss": 0.8528,
+      "step": 6442
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.481585832914547e-06,
+      "loss": 1.297,
+      "step": 6443
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4807377902304094e-06,
+      "loss": 0.7444,
+      "step": 6444
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.479889910790887e-06,
+      "loss": 1.0579,
+      "step": 6445
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.479042194687085e-06,
+      "loss": 1.0806,
+      "step": 6446
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4781946420100906e-06,
+      "loss": 0.77,
+      "step": 6447
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.477347252850975e-06,
+      "loss": 1.0815,
+      "step": 6448
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4765000273007867e-06,
+      "loss": 0.8441,
+      "step": 6449
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4756529654505619e-06,
+      "loss": 0.7747,
+      "step": 6450
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.474806067391317e-06,
+      "loss": 0.8592,
+      "step": 6451
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4739593332140528e-06,
+      "loss": 1.1064,
+      "step": 6452
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.473112763009748e-06,
+      "loss": 1.0341,
+      "step": 6453
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4722663568693682e-06,
+      "loss": 0.8132,
+      "step": 6454
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.471420114883859e-06,
+      "loss": 0.829,
+      "step": 6455
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4705740371441507e-06,
+      "loss": 1.0679,
+      "step": 6456
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4697281237411516e-06,
+      "loss": 0.8693,
+      "step": 6457
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4688823747657568e-06,
+      "loss": 0.9122,
+      "step": 6458
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.468036790308841e-06,
+      "loss": 1.0314,
+      "step": 6459
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4671913704612644e-06,
+      "loss": 0.9191,
+      "step": 6460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4663461153138644e-06,
+      "loss": 0.9692,
+      "step": 6461
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4655010249574648e-06,
+      "loss": 0.8001,
+      "step": 6462
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4646560994828704e-06,
+      "loss": 0.9268,
+      "step": 6463
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.46381133898087e-06,
+      "loss": 0.922,
+      "step": 6464
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4629667435422302e-06,
+      "loss": 0.8238,
+      "step": 6465
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4621223132577044e-06,
+      "loss": 0.7994,
+      "step": 6466
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4612780482180262e-06,
+      "loss": 0.8768,
+      "step": 6467
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4604339485139128e-06,
+      "loss": 0.9263,
+      "step": 6468
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.459590014236061e-06,
+      "loss": 1.0007,
+      "step": 6469
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4587462454751519e-06,
+      "loss": 1.1394,
+      "step": 6470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.457902642321849e-06,
+      "loss": 0.8521,
+      "step": 6471
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4570592048667983e-06,
+      "loss": 0.9581,
+      "step": 6472
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4562159332006246e-06,
+      "loss": 0.9183,
+      "step": 6473
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4553728274139392e-06,
+      "loss": 0.8707,
+      "step": 6474
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4545298875973328e-06,
+      "loss": 1.0591,
+      "step": 6475
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.453687113841381e-06,
+      "loss": 1.0025,
+      "step": 6476
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.452844506236637e-06,
+      "loss": 0.7434,
+      "step": 6477
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4520020648736407e-06,
+      "loss": 0.812,
+      "step": 6478
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.451159789842912e-06,
+      "loss": 0.8022,
+      "step": 6479
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4503176812349546e-06,
+      "loss": 0.9982,
+      "step": 6480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4494757391402507e-06,
+      "loss": 0.8646,
+      "step": 6481
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4486339636492677e-06,
+      "loss": 1.0913,
+      "step": 6482
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4477923548524547e-06,
+      "loss": 0.8986,
+      "step": 6483
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4469509128402437e-06,
+      "loss": 0.8862,
+      "step": 6484
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4461096377030453e-06,
+      "loss": 0.9357,
+      "step": 6485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.445268529531255e-06,
+      "loss": 0.7819,
+      "step": 6486
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4444275884152505e-06,
+      "loss": 0.7505,
+      "step": 6487
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4435868144453918e-06,
+      "loss": 0.97,
+      "step": 6488
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4427462077120163e-06,
+      "loss": 0.8285,
+      "step": 6489
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4419057683054524e-06,
+      "loss": 0.8555,
+      "step": 6490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.441065496316001e-06,
+      "loss": 0.7912,
+      "step": 6491
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.440225391833952e-06,
+      "loss": 0.7341,
+      "step": 6492
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.439385454949571e-06,
+      "loss": 1.0783,
+      "step": 6493
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4385456857531138e-06,
+      "loss": 0.8421,
+      "step": 6494
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4377060843348102e-06,
+      "loss": 0.7591,
+      "step": 6495
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4368666507848776e-06,
+      "loss": 0.9004,
+      "step": 6496
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4360273851935094e-06,
+      "loss": 0.9518,
+      "step": 6497
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4351882876508893e-06,
+      "loss": 0.7135,
+      "step": 6498
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4343493582471747e-06,
+      "loss": 1.0702,
+      "step": 6499
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4335105970725112e-06,
+      "loss": 0.9213,
+      "step": 6500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.43267200421702e-06,
+      "loss": 1.2515,
+      "step": 6501
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4318335797708126e-06,
+      "loss": 0.6504,
+      "step": 6502
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4309953238239736e-06,
+      "loss": 0.9165,
+      "step": 6503
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4301572364665764e-06,
+      "loss": 0.9404,
+      "step": 6504
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.42931931778867e-06,
+      "loss": 0.7413,
+      "step": 6505
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4284815678802927e-06,
+      "loss": 0.8186,
+      "step": 6506
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.427643986831458e-06,
+      "loss": 0.8692,
+      "step": 6507
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4268065747321657e-06,
+      "loss": 0.8415,
+      "step": 6508
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4259693316723928e-06,
+      "loss": 0.7132,
+      "step": 6509
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4251322577421048e-06,
+      "loss": 0.8842,
+      "step": 6510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.424295353031242e-06,
+      "loss": 0.8095,
+      "step": 6511
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4234586176297327e-06,
+      "loss": 0.8069,
+      "step": 6512
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4226220516274797e-06,
+      "loss": 1.0709,
+      "step": 6513
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4217856551143772e-06,
+      "loss": 0.919,
+      "step": 6514
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4209494281802922e-06,
+      "loss": 0.8702,
+      "step": 6515
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4201133709150797e-06,
+      "loss": 0.8902,
+      "step": 6516
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4192774834085706e-06,
+      "loss": 0.9974,
+      "step": 6517
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.418441765750585e-06,
+      "loss": 0.888,
+      "step": 6518
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4176062180309179e-06,
+      "loss": 1.0993,
+      "step": 6519
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.416770840339351e-06,
+      "loss": 0.9159,
+      "step": 6520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4159356327656421e-06,
+      "loss": 0.9144,
+      "step": 6521
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4151005953995384e-06,
+      "loss": 0.8711,
+      "step": 6522
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4142657283307616e-06,
+      "loss": 0.9718,
+      "step": 6523
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4134310316490203e-06,
+      "loss": 1.1161,
+      "step": 6524
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4125965054439994e-06,
+      "loss": 0.817,
+      "step": 6525
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4117621498053725e-06,
+      "loss": 0.8939,
+      "step": 6526
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4109279648227887e-06,
+      "loss": 0.8062,
+      "step": 6527
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4100939505858822e-06,
+      "loss": 0.9737,
+      "step": 6528
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4092601071842655e-06,
+      "loss": 1.1303,
+      "step": 6529
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4084264347075388e-06,
+      "loss": 0.71,
+      "step": 6530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4075929332452772e-06,
+      "loss": 0.7517,
+      "step": 6531
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4067596028870421e-06,
+      "loss": 1.0714,
+      "step": 6532
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.405926443722372e-06,
+      "loss": 0.9655,
+      "step": 6533
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4050934558407939e-06,
+      "loss": 1.0585,
+      "step": 6534
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4042606393318087e-06,
+      "loss": 0.9255,
+      "step": 6535
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4034279942849056e-06,
+      "loss": 0.9949,
+      "step": 6536
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4025955207895482e-06,
+      "loss": 0.9094,
+      "step": 6537
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4017632189351903e-06,
+      "loss": 0.9864,
+      "step": 6538
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4009310888112593e-06,
+      "loss": 0.7722,
+      "step": 6539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.40009913050717e-06,
+      "loss": 1.1161,
+      "step": 6540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3992673441123126e-06,
+      "loss": 0.8896,
+      "step": 6541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.398435729716067e-06,
+      "loss": 1.0284,
+      "step": 6542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3976042874077868e-06,
+      "loss": 0.9105,
+      "step": 6543
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3967730172768129e-06,
+      "loss": 1.0152,
+      "step": 6544
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3959419194124614e-06,
+      "loss": 0.9502,
+      "step": 6545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3951109939040386e-06,
+      "loss": 0.9993,
+      "step": 6546
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3942802408408238e-06,
+      "loss": 0.6866,
+      "step": 6547
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.393449660312084e-06,
+      "loss": 0.9295,
+      "step": 6548
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3926192524070607e-06,
+      "loss": 0.8397,
+      "step": 6549
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.391789017214987e-06,
+      "loss": 1.0773,
+      "step": 6550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.390958954825067e-06,
+      "loss": 0.992,
+      "step": 6551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3901290653264938e-06,
+      "loss": 0.7448,
+      "step": 6552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3892993488084353e-06,
+      "loss": 0.9002,
+      "step": 6553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.388469805360049e-06,
+      "loss": 0.8622,
+      "step": 6554
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.387640435070466e-06,
+      "loss": 0.8931,
+      "step": 6555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3868112380288035e-06,
+      "loss": 0.6334,
+      "step": 6556
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3859822143241582e-06,
+      "loss": 0.948,
+      "step": 6557
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3851533640456097e-06,
+      "loss": 0.8655,
+      "step": 6558
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.384324687282216e-06,
+      "loss": 1.0403,
+      "step": 6559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3834961841230196e-06,
+      "loss": 0.8378,
+      "step": 6560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3826678546570426e-06,
+      "loss": 0.9306,
+      "step": 6561
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3818396989732905e-06,
+      "loss": 0.9944,
+      "step": 6562
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3810117171607461e-06,
+      "loss": 0.9046,
+      "step": 6563
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.380183909308377e-06,
+      "loss": 0.9955,
+      "step": 6564
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3793562755051315e-06,
+      "loss": 0.87,
+      "step": 6565
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3785288158399397e-06,
+      "loss": 0.9818,
+      "step": 6566
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3777015304017096e-06,
+      "loss": 1.1803,
+      "step": 6567
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3768744192793343e-06,
+      "loss": 1.0755,
+      "step": 6568
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.376047482561687e-06,
+      "loss": 0.9743,
+      "step": 6569
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3752207203376229e-06,
+      "loss": 0.8801,
+      "step": 6570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.374394132695975e-06,
+      "loss": 0.7723,
+      "step": 6571
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3735677197255618e-06,
+      "loss": 0.7913,
+      "step": 6572
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3727414815151811e-06,
+      "loss": 0.9684,
+      "step": 6573
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3719154181536134e-06,
+      "loss": 0.6843,
+      "step": 6574
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3710895297296163e-06,
+      "loss": 0.6859,
+      "step": 6575
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3702638163319332e-06,
+      "loss": 0.9306,
+      "step": 6576
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3694382780492866e-06,
+      "loss": 0.8895,
+      "step": 6577
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.368612914970382e-06,
+      "loss": 0.9537,
+      "step": 6578
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3677877271839022e-06,
+      "loss": 0.8478,
+      "step": 6579
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3669627147785144e-06,
+      "loss": 0.8423,
+      "step": 6580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3661378778428664e-06,
+      "loss": 0.7934,
+      "step": 6581
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3653132164655881e-06,
+      "loss": 0.7367,
+      "step": 6582
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3644887307352869e-06,
+      "loss": 0.9978,
+      "step": 6583
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3636644207405548e-06,
+      "loss": 1.1592,
+      "step": 6584
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.362840286569964e-06,
+      "loss": 1.0023,
+      "step": 6585
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3620163283120687e-06,
+      "loss": 0.9426,
+      "step": 6586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3611925460554009e-06,
+      "loss": 1.0317,
+      "step": 6587
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.360368939888477e-06,
+      "loss": 0.7481,
+      "step": 6588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3595455098997934e-06,
+      "loss": 0.7705,
+      "step": 6589
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3587222561778293e-06,
+      "loss": 0.9572,
+      "step": 6590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3578991788110401e-06,
+      "loss": 0.8553,
+      "step": 6591
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.357076277887867e-06,
+      "loss": 1.1864,
+      "step": 6592
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3562535534967308e-06,
+      "loss": 0.8468,
+      "step": 6593
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.355431005726034e-06,
+      "loss": 0.6479,
+      "step": 6594
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3546086346641572e-06,
+      "loss": 0.8392,
+      "step": 6595
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3537864403994652e-06,
+      "loss": 0.8531,
+      "step": 6596
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3529644230203025e-06,
+      "loss": 0.8784,
+      "step": 6597
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3521425826149966e-06,
+      "loss": 0.8826,
+      "step": 6598
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3513209192718508e-06,
+      "loss": 1.1329,
+      "step": 6599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.350499433079155e-06,
+      "loss": 1.0777,
+      "step": 6600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3496781241251773e-06,
+      "loss": 0.8636,
+      "step": 6601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3488569924981688e-06,
+      "loss": 1.0466,
+      "step": 6602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3480360382863572e-06,
+      "loss": 0.8856,
+      "step": 6603
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3472152615779555e-06,
+      "loss": 0.9092,
+      "step": 6604
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3463946624611561e-06,
+      "loss": 0.9311,
+      "step": 6605
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3455742410241332e-06,
+      "loss": 0.7164,
+      "step": 6606
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.344753997355039e-06,
+      "loss": 0.8951,
+      "step": 6607
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3439339315420097e-06,
+      "loss": 0.9693,
+      "step": 6608
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3431140436731611e-06,
+      "loss": 0.9444,
+      "step": 6609
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3422943338365917e-06,
+      "loss": 0.7414,
+      "step": 6610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3414748021203764e-06,
+      "loss": 1.0658,
+      "step": 6611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3406554486125754e-06,
+      "loss": 1.1119,
+      "step": 6612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3398362734012281e-06,
+      "loss": 0.9073,
+      "step": 6613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3390172765743559e-06,
+      "loss": 0.7765,
+      "step": 6614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3381984582199576e-06,
+      "loss": 0.994,
+      "step": 6615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3373798184260162e-06,
+      "loss": 0.9415,
+      "step": 6616
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3365613572804947e-06,
+      "loss": 0.7317,
+      "step": 6617
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3357430748713383e-06,
+      "loss": 0.9485,
+      "step": 6618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3349249712864682e-06,
+      "loss": 0.7774,
+      "step": 6619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3341070466137912e-06,
+      "loss": 0.8216,
+      "step": 6620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3332893009411942e-06,
+      "loss": 1.0826,
+      "step": 6621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3324717343565417e-06,
+      "loss": 0.9731,
+      "step": 6622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3316543469476824e-06,
+      "loss": 0.8743,
+      "step": 6623
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3308371388024448e-06,
+      "loss": 0.8553,
+      "step": 6624
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3300201100086381e-06,
+      "loss": 0.8233,
+      "step": 6625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3292032606540506e-06,
+      "loss": 0.8455,
+      "step": 6626
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3283865908264538e-06,
+      "loss": 0.7132,
+      "step": 6627
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.327570100613598e-06,
+      "loss": 0.7533,
+      "step": 6628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3267537901032171e-06,
+      "loss": 0.7121,
+      "step": 6629
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3259376593830207e-06,
+      "loss": 0.854,
+      "step": 6630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3251217085407033e-06,
+      "loss": 1.0314,
+      "step": 6631
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.324305937663939e-06,
+      "loss": 0.8581,
+      "step": 6632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.323490346840383e-06,
+      "loss": 0.8971,
+      "step": 6633
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3226749361576686e-06,
+      "loss": 0.841,
+      "step": 6634
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3218597057034126e-06,
+      "loss": 0.7428,
+      "step": 6635
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3210446555652113e-06,
+      "loss": 0.9345,
+      "step": 6636
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3202297858306435e-06,
+      "loss": 1.1541,
+      "step": 6637
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.319415096587264e-06,
+      "loss": 0.9749,
+      "step": 6638
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3186005879226123e-06,
+      "loss": 0.8167,
+      "step": 6639
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3177862599242077e-06,
+      "loss": 0.8852,
+      "step": 6640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3169721126795503e-06,
+      "loss": 0.7853,
+      "step": 6641
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3161581462761183e-06,
+      "loss": 0.9683,
+      "step": 6642
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3153443608013733e-06,
+      "loss": 0.9565,
+      "step": 6643
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3145307563427568e-06,
+      "loss": 0.9808,
+      "step": 6644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.313717332987691e-06,
+      "loss": 1.0113,
+      "step": 6645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3129040908235765e-06,
+      "loss": 0.8932,
+      "step": 6646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.312091029937797e-06,
+      "loss": 0.72,
+      "step": 6647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3112781504177158e-06,
+      "loss": 0.7778,
+      "step": 6648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.310465452350678e-06,
+      "loss": 0.8635,
+      "step": 6649
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.309652935824006e-06,
+      "loss": 0.8762,
+      "step": 6650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3088406009250048e-06,
+      "loss": 1.0237,
+      "step": 6651
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3080284477409607e-06,
+      "loss": 0.9349,
+      "step": 6652
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3072164763591403e-06,
+      "loss": 1.0618,
+      "step": 6653
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3064046868667876e-06,
+      "loss": 0.9973,
+      "step": 6654
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3055930793511306e-06,
+      "loss": 0.8545,
+      "step": 6655
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3047816538993758e-06,
+      "loss": 0.8367,
+      "step": 6656
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.303970410598713e-06,
+      "loss": 0.9513,
+      "step": 6657
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3031593495363073e-06,
+      "loss": 1.0338,
+      "step": 6658
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.302348470799308e-06,
+      "loss": 0.9612,
+      "step": 6659
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3015377744748447e-06,
+      "loss": 0.7491,
+      "step": 6660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3007272606500274e-06,
+      "loss": 0.8788,
+      "step": 6661
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2999169294119438e-06,
+      "loss": 0.8427,
+      "step": 6662
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2991067808476646e-06,
+      "loss": 0.9055,
+      "step": 6663
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2982968150442402e-06,
+      "loss": 0.8682,
+      "step": 6664
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2974870320887028e-06,
+      "loss": 0.7889,
+      "step": 6665
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2966774320680615e-06,
+      "loss": 0.8446,
+      "step": 6666
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2958680150693085e-06,
+      "loss": 1.0341,
+      "step": 6667
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2950587811794155e-06,
+      "loss": 0.8566,
+      "step": 6668
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2942497304853361e-06,
+      "loss": 0.8221,
+      "step": 6669
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2934408630740003e-06,
+      "loss": 1.1999,
+      "step": 6670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.292632179032322e-06,
+      "loss": 0.9019,
+      "step": 6671
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.291823678447194e-06,
+      "loss": 0.911,
+      "step": 6672
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2910153614054911e-06,
+      "loss": 0.8821,
+      "step": 6673
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2902072279940645e-06,
+      "loss": 0.9117,
+      "step": 6674
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2893992782997493e-06,
+      "loss": 1.0332,
+      "step": 6675
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2885915124093598e-06,
+      "loss": 0.9276,
+      "step": 6676
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.287783930409691e-06,
+      "loss": 0.9504,
+      "step": 6677
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2869765323875156e-06,
+      "loss": 0.8496,
+      "step": 6678
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2861693184295896e-06,
+      "loss": 0.857,
+      "step": 6679
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2853622886226482e-06,
+      "loss": 1.0342,
+      "step": 6680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2845554430534075e-06,
+      "loss": 0.8265,
+      "step": 6681
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2837487818085608e-06,
+      "loss": 0.9281,
+      "step": 6682
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2829423049747852e-06,
+      "loss": 1.0113,
+      "step": 6683
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2821360126387365e-06,
+      "loss": 1.0332,
+      "step": 6684
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2813299048870514e-06,
+      "loss": 0.9541,
+      "step": 6685
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2805239818063446e-06,
+      "loss": 0.9614,
+      "step": 6686
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.279718243483213e-06,
+      "loss": 0.9916,
+      "step": 6687
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2789126900042335e-06,
+      "loss": 0.95,
+      "step": 6688
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.278107321455964e-06,
+      "loss": 0.9629,
+      "step": 6689
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2773021379249374e-06,
+      "loss": 0.9307,
+      "step": 6690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2764971394976756e-06,
+      "loss": 0.87,
+      "step": 6691
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2756923262606719e-06,
+      "loss": 0.6714,
+      "step": 6692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2748876983004059e-06,
+      "loss": 0.9719,
+      "step": 6693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2740832557033314e-06,
+      "loss": 1.0875,
+      "step": 6694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.27327899855589e-06,
+      "loss": 0.7716,
+      "step": 6695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.272474926944496e-06,
+      "loss": 0.6669,
+      "step": 6696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2716710409555487e-06,
+      "loss": 0.8407,
+      "step": 6697
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2708673406754229e-06,
+      "loss": 1.0069,
+      "step": 6698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.27006382619048e-06,
+      "loss": 0.9753,
+      "step": 6699
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.269260497587054e-06,
+      "loss": 0.9722,
+      "step": 6700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2684573549514657e-06,
+      "loss": 0.7055,
+      "step": 6701
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2676543983700089e-06,
+      "loss": 0.9855,
+      "step": 6702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.266851627928965e-06,
+      "loss": 0.9214,
+      "step": 6703
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2660490437145893e-06,
+      "loss": 0.9817,
+      "step": 6704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2652466458131212e-06,
+      "loss": 0.8992,
+      "step": 6705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.264444434310775e-06,
+      "loss": 0.8613,
+      "step": 6706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2636424092937528e-06,
+      "loss": 0.9161,
+      "step": 6707
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2628405708482285e-06,
+      "loss": 0.7327,
+      "step": 6708
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.262038919060362e-06,
+      "loss": 0.5397,
+      "step": 6709
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2612374540162878e-06,
+      "loss": 1.0688,
+      "step": 6710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.260436175802127e-06,
+      "loss": 0.9075,
+      "step": 6711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.259635084503974e-06,
+      "loss": 0.9346,
+      "step": 6712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.258834180207908e-06,
+      "loss": 0.8958,
+      "step": 6713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2580334629999826e-06,
+      "loss": 0.8132,
+      "step": 6714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.25723293296624e-06,
+      "loss": 0.8626,
+      "step": 6715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.256432590192693e-06,
+      "loss": 0.7124,
+      "step": 6716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2556324347653409e-06,
+      "loss": 0.9198,
+      "step": 6717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2548324667701572e-06,
+      "loss": 0.8463,
+      "step": 6718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2540326862931024e-06,
+      "loss": 0.7562,
+      "step": 6719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2532330934201095e-06,
+      "loss": 0.7193,
+      "step": 6720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2524336882370975e-06,
+      "loss": 1.1189,
+      "step": 6721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2516344708299588e-06,
+      "loss": 0.7992,
+      "step": 6722
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2508354412845735e-06,
+      "loss": 0.9314,
+      "step": 6723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2500365996867944e-06,
+      "loss": 0.8013,
+      "step": 6724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.249237946122459e-06,
+      "loss": 1.1225,
+      "step": 6725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2484394806773789e-06,
+      "loss": 1.1243,
+      "step": 6726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2476412034373536e-06,
+      "loss": 0.8424,
+      "step": 6727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2468431144881548e-06,
+      "loss": 0.8411,
+      "step": 6728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.24604521391554e-06,
+      "loss": 0.7648,
+      "step": 6729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.245247501805239e-06,
+      "loss": 0.8139,
+      "step": 6730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2444499782429711e-06,
+      "loss": 0.9499,
+      "step": 6731
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2436526433144266e-06,
+      "loss": 0.9052,
+      "step": 6732
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2428554971052818e-06,
+      "loss": 1.0495,
+      "step": 6733
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2420585397011857e-06,
+      "loss": 0.8181,
+      "step": 6734
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2412617711877764e-06,
+      "loss": 0.8947,
+      "step": 6735
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.240465191650663e-06,
+      "loss": 0.8688,
+      "step": 6736
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2396688011754406e-06,
+      "loss": 1.0235,
+      "step": 6737
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2388725998476777e-06,
+      "loss": 0.808,
+      "step": 6738
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2380765877529305e-06,
+      "loss": 0.9042,
+      "step": 6739
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.237280764976727e-06,
+      "loss": 0.7171,
+      "step": 6740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2364851316045805e-06,
+      "loss": 0.8332,
+      "step": 6741
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2356896877219787e-06,
+      "loss": 0.8975,
+      "step": 6742
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2348944334143965e-06,
+      "loss": 0.8839,
+      "step": 6743
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.23409936876728e-06,
+      "loss": 0.715,
+      "step": 6744
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2333044938660613e-06,
+      "loss": 0.8983,
+      "step": 6745
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2325098087961469e-06,
+      "loss": 0.9733,
+      "step": 6746
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2317153136429293e-06,
+      "loss": 0.9586,
+      "step": 6747
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2309210084917743e-06,
+      "loss": 0.9427,
+      "step": 6748
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2301268934280317e-06,
+      "loss": 0.8111,
+      "step": 6749
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2293329685370256e-06,
+      "loss": 1.0365,
+      "step": 6750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2285392339040682e-06,
+      "loss": 0.9636,
+      "step": 6751
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2277456896144425e-06,
+      "loss": 0.925,
+      "step": 6752
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2269523357534168e-06,
+      "loss": 0.9947,
+      "step": 6753
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2261591724062344e-06,
+      "loss": 0.9895,
+      "step": 6754
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2253661996581243e-06,
+      "loss": 0.6711,
+      "step": 6755
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2245734175942884e-06,
+      "loss": 1.0295,
+      "step": 6756
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2237808262999119e-06,
+      "loss": 0.9985,
+      "step": 6757
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.222988425860159e-06,
+      "loss": 0.7907,
+      "step": 6758
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2221962163601742e-06,
+      "loss": 0.7911,
+      "step": 6759
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2214041978850776e-06,
+      "loss": 1.0396,
+      "step": 6760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2206123705199732e-06,
+      "loss": 0.8043,
+      "step": 6761
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.219820734349943e-06,
+      "loss": 0.8628,
+      "step": 6762
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2190292894600483e-06,
+      "loss": 0.9498,
+      "step": 6763
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2182380359353288e-06,
+      "loss": 0.9521,
+      "step": 6764
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2174469738608049e-06,
+      "loss": 0.7847,
+      "step": 6765
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2166561033214766e-06,
+      "loss": 0.8683,
+      "step": 6766
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2158654244023237e-06,
+      "loss": 0.9259,
+      "step": 6767
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2150749371883025e-06,
+      "loss": 0.9482,
+      "step": 6768
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.214284641764352e-06,
+      "loss": 1.0697,
+      "step": 6769
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2134945382153895e-06,
+      "loss": 1.0602,
+      "step": 6770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2127046266263121e-06,
+      "loss": 0.9454,
+      "step": 6771
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2119149070819942e-06,
+      "loss": 0.9329,
+      "step": 6772
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.211125379667292e-06,
+      "loss": 1.0559,
+      "step": 6773
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.21033604446704e-06,
+      "loss": 1.047,
+      "step": 6774
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2095469015660533e-06,
+      "loss": 0.84,
+      "step": 6775
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2087579510491235e-06,
+      "loss": 0.9485,
+      "step": 6776
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2079691930010237e-06,
+      "loss": 0.9264,
+      "step": 6777
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2071806275065062e-06,
+      "loss": 0.8456,
+      "step": 6778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2063922546503037e-06,
+      "loss": 1.0153,
+      "step": 6779
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2056040745171245e-06,
+      "loss": 0.798,
+      "step": 6780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.204816087191659e-06,
+      "loss": 0.9401,
+      "step": 6781
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.204028292758577e-06,
+      "loss": 0.8624,
+      "step": 6782
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2032406913025279e-06,
+      "loss": 0.8841,
+      "step": 6783
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2024532829081371e-06,
+      "loss": 0.6948,
+      "step": 6784
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.201666067660013e-06,
+      "loss": 0.8691,
+      "step": 6785
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2008790456427416e-06,
+      "loss": 0.9622,
+      "step": 6786
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2000922169408896e-06,
+      "loss": 0.7134,
+      "step": 6787
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1993055816389988e-06,
+      "loss": 0.7631,
+      "step": 6788
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1985191398215952e-06,
+      "loss": 0.9268,
+      "step": 6789
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1977328915731814e-06,
+      "loss": 0.7994,
+      "step": 6790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1969468369782409e-06,
+      "loss": 0.9481,
+      "step": 6791
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1961609761212328e-06,
+      "loss": 0.6481,
+      "step": 6792
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1953753090865995e-06,
+      "loss": 0.8646,
+      "step": 6793
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.19458983595876e-06,
+      "loss": 0.8032,
+      "step": 6794
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.193804556822115e-06,
+      "loss": 0.6443,
+      "step": 6795
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1930194717610407e-06,
+      "loss": 0.7034,
+      "step": 6796
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1922345808598948e-06,
+      "loss": 1.1723,
+      "step": 6797
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1914498842030148e-06,
+      "loss": 0.8904,
+      "step": 6798
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.190665381874716e-06,
+      "loss": 1.0699,
+      "step": 6799
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1898810739592922e-06,
+      "loss": 1.1285,
+      "step": 6800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.189096960541018e-06,
+      "loss": 0.9191,
+      "step": 6801
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1883130417041456e-06,
+      "loss": 0.8099,
+      "step": 6802
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1875293175329089e-06,
+      "loss": 0.8462,
+      "step": 6803
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1867457881115168e-06,
+      "loss": 0.9509,
+      "step": 6804
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1859624535241603e-06,
+      "loss": 1.0174,
+      "step": 6805
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1851793138550087e-06,
+      "loss": 0.8237,
+      "step": 6806
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1843963691882113e-06,
+      "loss": 0.8453,
+      "step": 6807
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1836136196078935e-06,
+      "loss": 0.9093,
+      "step": 6808
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1828310651981626e-06,
+      "loss": 0.9509,
+      "step": 6809
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1820487060431042e-06,
+      "loss": 0.8912,
+      "step": 6810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1812665422267835e-06,
+      "loss": 0.799,
+      "step": 6811
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.180484573833242e-06,
+      "loss": 0.8043,
+      "step": 6812
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1797028009465033e-06,
+      "loss": 1.039,
+      "step": 6813
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1789212236505688e-06,
+      "loss": 0.8663,
+      "step": 6814
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1781398420294199e-06,
+      "loss": 0.9469,
+      "step": 6815
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1773586561670138e-06,
+      "loss": 0.8427,
+      "step": 6816
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.17657766614729e-06,
+      "loss": 0.8202,
+      "step": 6817
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1757968720541662e-06,
+      "loss": 0.8397,
+      "step": 6818
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.175016273971539e-06,
+      "loss": 0.7284,
+      "step": 6819
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1742358719832821e-06,
+      "loss": 1.0247,
+      "step": 6820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1734556661732502e-06,
+      "loss": 0.9489,
+      "step": 6821
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1726756566252772e-06,
+      "loss": 0.9965,
+      "step": 6822
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1718958434231747e-06,
+      "loss": 0.6961,
+      "step": 6823
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.171116226650732e-06,
+      "loss": 0.7918,
+      "step": 6824
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1703368063917222e-06,
+      "loss": 0.9467,
+      "step": 6825
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1695575827298907e-06,
+      "loss": 0.9079,
+      "step": 6826
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1687785557489672e-06,
+      "loss": 0.8293,
+      "step": 6827
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1679997255326559e-06,
+      "loss": 0.851,
+      "step": 6828
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1672210921646449e-06,
+      "loss": 0.8853,
+      "step": 6829
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1664426557285957e-06,
+      "loss": 0.9385,
+      "step": 6830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1656644163081536e-06,
+      "loss": 0.8966,
+      "step": 6831
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1648863739869364e-06,
+      "loss": 0.9809,
+      "step": 6832
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1641085288485502e-06,
+      "loss": 0.6787,
+      "step": 6833
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.16333088097657e-06,
+      "loss": 0.8369,
+      "step": 6834
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.162553430454557e-06,
+      "loss": 0.9212,
+      "step": 6835
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1617761773660446e-06,
+      "loss": 1.0998,
+      "step": 6836
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1609991217945523e-06,
+      "loss": 1.0333,
+      "step": 6837
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1602222638235728e-06,
+      "loss": 0.8484,
+      "step": 6838
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1594456035365806e-06,
+      "loss": 0.8359,
+      "step": 6839
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1586691410170247e-06,
+      "loss": 0.9133,
+      "step": 6840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1578928763483406e-06,
+      "loss": 1.1417,
+      "step": 6841
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1571168096139343e-06,
+      "loss": 0.892,
+      "step": 6842
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1563409408971963e-06,
+      "loss": 0.9959,
+      "step": 6843
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1555652702814907e-06,
+      "loss": 1.1017,
+      "step": 6844
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1547897978501668e-06,
+      "loss": 0.8715,
+      "step": 6845
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1540145236865468e-06,
+      "loss": 0.9625,
+      "step": 6846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1532394478739356e-06,
+      "loss": 0.9196,
+      "step": 6847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1524645704956117e-06,
+      "loss": 1.011,
+      "step": 6848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1516898916348402e-06,
+      "loss": 0.9545,
+      "step": 6849
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1509154113748565e-06,
+      "loss": 0.9104,
+      "step": 6850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1501411297988813e-06,
+      "loss": 1.0047,
+      "step": 6851
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1493670469901075e-06,
+      "loss": 1.1566,
+      "step": 6852
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1485931630317146e-06,
+      "loss": 0.7519,
+      "step": 6853
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.147819478006853e-06,
+      "loss": 0.8371,
+      "step": 6854
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1470459919986574e-06,
+      "loss": 0.9923,
+      "step": 6855
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1462727050902356e-06,
+      "loss": 0.7445,
+      "step": 6856
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1454996173646816e-06,
+      "loss": 0.9625,
+      "step": 6857
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1447267289050602e-06,
+      "loss": 0.9502,
+      "step": 6858
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1439540397944206e-06,
+      "loss": 0.8065,
+      "step": 6859
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1431815501157846e-06,
+      "loss": 0.6808,
+      "step": 6860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1424092599521613e-06,
+      "loss": 1.0784,
+      "step": 6861
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1416371693865285e-06,
+      "loss": 1.032,
+      "step": 6862
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.140865278501851e-06,
+      "loss": 0.7843,
+      "step": 6863
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1400935873810643e-06,
+      "loss": 0.9387,
+      "step": 6864
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1393220961070907e-06,
+      "loss": 0.9387,
+      "step": 6865
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1385508047628246e-06,
+      "loss": 0.9782,
+      "step": 6866
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1377797134311424e-06,
+      "loss": 1.192,
+      "step": 6867
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1370088221948954e-06,
+      "loss": 0.9467,
+      "step": 6868
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.136238131136919e-06,
+      "loss": 0.8818,
+      "step": 6869
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.135467640340022e-06,
+      "loss": 0.876,
+      "step": 6870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1346973498869946e-06,
+      "loss": 0.7543,
+      "step": 6871
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.133927259860602e-06,
+      "loss": 0.9252,
+      "step": 6872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.133157370343594e-06,
+      "loss": 0.9598,
+      "step": 6873
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1323876814186925e-06,
+      "loss": 0.8297,
+      "step": 6874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1316181931686023e-06,
+      "loss": 0.8426,
+      "step": 6875
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.130848905676002e-06,
+      "loss": 1.1608,
+      "step": 6876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.130079819023555e-06,
+      "loss": 0.9376,
+      "step": 6877
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1293109332938979e-06,
+      "loss": 0.8482,
+      "step": 6878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1285422485696477e-06,
+      "loss": 0.8321,
+      "step": 6879
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277737649333976e-06,
+      "loss": 0.9871,
+      "step": 6880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127005482467725e-06,
+      "loss": 0.9057,
+      "step": 6881
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1262374012551786e-06,
+      "loss": 0.8588,
+      "step": 6882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1254695213782906e-06,
+      "loss": 0.9413,
+      "step": 6883
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124701842919567e-06,
+      "loss": 0.8906,
+      "step": 6884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1239343659614984e-06,
+      "loss": 0.8612,
+      "step": 6885
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1231670905865472e-06,
+      "loss": 1.0163,
+      "step": 6886
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224000168771593e-06,
+      "loss": 0.9879,
+      "step": 6887
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216331449157535e-06,
+      "loss": 0.9622,
+      "step": 6888
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208664747847346e-06,
+      "loss": 0.8187,
+      "step": 6889
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1201000065664775e-06,
+      "loss": 0.9826,
+      "step": 6890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193337403433406e-06,
+      "loss": 0.9072,
+      "step": 6891
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1185676761976603e-06,
+      "loss": 0.8565,
+      "step": 6892
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1178018142117476e-06,
+      "loss": 0.9846,
+      "step": 6893
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.117036154467896e-06,
+      "loss": 0.755,
+      "step": 6894
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1162706970483752e-06,
+      "loss": 0.6764,
+      "step": 6895
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1155054420354344e-06,
+      "loss": 0.9133,
+      "step": 6896
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1147403895112987e-06,
+      "loss": 0.8393,
+      "step": 6897
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1139755395581733e-06,
+      "loss": 1.0558,
+      "step": 6898
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1132108922582416e-06,
+      "loss": 0.6037,
+      "step": 6899
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112446447693666e-06,
+      "loss": 0.8971,
+      "step": 6900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1116822059465837e-06,
+      "loss": 0.7729,
+      "step": 6901
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110918167099114e-06,
+      "loss": 0.6806,
+      "step": 6902
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101543312333519e-06,
+      "loss": 0.922,
+      "step": 6903
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1093906984313738e-06,
+      "loss": 0.8411,
+      "step": 6904
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086272687752288e-06,
+      "loss": 0.6663,
+      "step": 6905
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078640423469485e-06,
+      "loss": 0.7835,
+      "step": 6906
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1071010192285421e-06,
+      "loss": 1.0212,
+      "step": 6907
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1063381995019973e-06,
+      "loss": 0.9635,
+      "step": 6908
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1055755832492768e-06,
+      "loss": 1.0293,
+      "step": 6909
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1048131705523247e-06,
+      "loss": 0.8336,
+      "step": 6910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.104050961493062e-06,
+      "loss": 0.8488,
+      "step": 6911
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1032889561533892e-06,
+      "loss": 0.9155,
+      "step": 6912
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1025271546151819e-06,
+      "loss": 0.954,
+      "step": 6913
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1017655569602963e-06,
+      "loss": 0.8606,
+      "step": 6914
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1010041632705663e-06,
+      "loss": 1.1361,
+      "step": 6915
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1002429736278043e-06,
+      "loss": 0.882,
+      "step": 6916
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0994819881137984e-06,
+      "loss": 0.7701,
+      "step": 6917
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.098721206810317e-06,
+      "loss": 1.1854,
+      "step": 6918
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0979606297991063e-06,
+      "loss": 0.8696,
+      "step": 6919
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0972002571618911e-06,
+      "loss": 0.8657,
+      "step": 6920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0964400889803718e-06,
+      "loss": 1.0656,
+      "step": 6921
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0956801253362288e-06,
+      "loss": 0.7732,
+      "step": 6922
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.094920366311121e-06,
+      "loss": 1.0196,
+      "step": 6923
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0941608119866844e-06,
+      "loss": 1.2692,
+      "step": 6924
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0934014624445316e-06,
+      "loss": 1.0464,
+      "step": 6925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0926423177662559e-06,
+      "loss": 0.8957,
+      "step": 6926
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0918833780334266e-06,
+      "loss": 0.8658,
+      "step": 6927
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0911246433275933e-06,
+      "loss": 1.0239,
+      "step": 6928
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.09036611373028e-06,
+      "loss": 0.8621,
+      "step": 6929
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0896077893229916e-06,
+      "loss": 0.9479,
+      "step": 6930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0888496701872096e-06,
+      "loss": 0.7468,
+      "step": 6931
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0880917564043953e-06,
+      "loss": 0.8467,
+      "step": 6932
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0873340480559844e-06,
+      "loss": 0.9431,
+      "step": 6933
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0865765452233938e-06,
+      "loss": 1.0123,
+      "step": 6934
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0858192479880164e-06,
+      "loss": 0.9207,
+      "step": 6935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0850621564312255e-06,
+      "loss": 1.0186,
+      "step": 6936
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0843052706343682e-06,
+      "loss": 1.1213,
+      "step": 6937
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.083548590678773e-06,
+      "loss": 0.8484,
+      "step": 6938
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0827921166457447e-06,
+      "loss": 0.8504,
+      "step": 6939
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0820358486165678e-06,
+      "loss": 0.8368,
+      "step": 6940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.081279786672501e-06,
+      "loss": 0.8847,
+      "step": 6941
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.080523930894784e-06,
+      "loss": 0.9696,
+      "step": 6942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0797682813646342e-06,
+      "loss": 0.8344,
+      "step": 6943
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0790128381632463e-06,
+      "loss": 1.098,
+      "step": 6944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0782576013717909e-06,
+      "loss": 0.8419,
+      "step": 6945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0775025710714186e-06,
+      "loss": 1.0648,
+      "step": 6946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0767477473432585e-06,
+      "loss": 0.8576,
+      "step": 6947
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0759931302684165e-06,
+      "loss": 1.1398,
+      "step": 6948
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0752387199279744e-06,
+      "loss": 1.0138,
+      "step": 6949
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0744845164029943e-06,
+      "loss": 0.91,
+      "step": 6950
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0737305197745155e-06,
+      "loss": 0.9482,
+      "step": 6951
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0729767301235563e-06,
+      "loss": 1.0248,
+      "step": 6952
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0722231475311088e-06,
+      "loss": 0.8496,
+      "step": 6953
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0714697720781463e-06,
+      "loss": 0.9918,
+      "step": 6954
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0707166038456194e-06,
+      "loss": 1.0219,
+      "step": 6955
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0699636429144565e-06,
+      "loss": 1.1175,
+      "step": 6956
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0692108893655617e-06,
+      "loss": 0.8574,
+      "step": 6957
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.068458343279819e-06,
+      "loss": 0.9708,
+      "step": 6958
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0677060047380894e-06,
+      "loss": 1.0254,
+      "step": 6959
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0669538738212128e-06,
+      "loss": 0.8846,
+      "step": 6960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0662019506100034e-06,
+      "loss": 0.8671,
+      "step": 6961
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0654502351852565e-06,
+      "loss": 0.8968,
+      "step": 6962
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0646987276277438e-06,
+      "loss": 1.1522,
+      "step": 6963
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0639474280182156e-06,
+      "loss": 0.7665,
+      "step": 6964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0631963364373973e-06,
+      "loss": 0.6807,
+      "step": 6965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0624454529659942e-06,
+      "loss": 1.0461,
+      "step": 6966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.061694777684689e-06,
+      "loss": 0.8123,
+      "step": 6967
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0609443106741426e-06,
+      "loss": 0.7487,
+      "step": 6968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.060194052014991e-06,
+      "loss": 0.9206,
+      "step": 6969
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0594440017878495e-06,
+      "loss": 0.7423,
+      "step": 6970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.058694160073312e-06,
+      "loss": 0.9886,
+      "step": 6971
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0579445269519492e-06,
+      "loss": 0.7991,
+      "step": 6972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0571951025043076e-06,
+      "loss": 1.005,
+      "step": 6973
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0564458868109135e-06,
+      "loss": 0.9329,
+      "step": 6974
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0556968799522701e-06,
+      "loss": 0.8669,
+      "step": 6975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0549480820088595e-06,
+      "loss": 1.0039,
+      "step": 6976
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0541994930611373e-06,
+      "loss": 0.9639,
+      "step": 6977
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0534511131895412e-06,
+      "loss": 0.9464,
+      "step": 6978
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0527029424744837e-06,
+      "loss": 0.8774,
+      "step": 6979
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.051954980996357e-06,
+      "loss": 0.9106,
+      "step": 6980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0512072288355276e-06,
+      "loss": 0.9515,
+      "step": 6981
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0504596860723428e-06,
+      "loss": 1.0843,
+      "step": 6982
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0497123527871253e-06,
+      "loss": 0.962,
+      "step": 6983
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0489652290601773e-06,
+      "loss": 0.8703,
+      "step": 6984
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0482183149717754e-06,
+      "loss": 0.9278,
+      "step": 6985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.047471610602176e-06,
+      "loss": 0.8027,
+      "step": 6986
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.046725116031613e-06,
+      "loss": 0.759,
+      "step": 6987
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.045978831340298e-06,
+      "loss": 1.1695,
+      "step": 6988
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.045232756608417e-06,
+      "loss": 0.7467,
+      "step": 6989
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0444868919161365e-06,
+      "loss": 0.9544,
+      "step": 6990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0437412373436002e-06,
+      "loss": 0.7605,
+      "step": 6991
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0429957929709293e-06,
+      "loss": 0.8536,
+      "step": 6992
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0422505588782198e-06,
+      "loss": 0.9364,
+      "step": 6993
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0415055351455482e-06,
+      "loss": 1.0724,
+      "step": 6994
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0407607218529668e-06,
+      "loss": 1.0327,
+      "step": 6995
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.040016119080507e-06,
+      "loss": 0.8101,
+      "step": 6996
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0392717269081748e-06,
+      "loss": 1.0086,
+      "step": 6997
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.038527545415955e-06,
+      "loss": 0.8576,
+      "step": 6998
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0377835746838111e-06,
+      "loss": 1.0748,
+      "step": 6999
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0370398147916827e-06,
+      "loss": 1.1731,
+      "step": 7000
+    },
+    {
+      "epoch": 1.46,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0705,
+      "eval_samples_per_second": 351.348,
+      "eval_steps_per_second": 43.927,
+      "step": 7000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0362962658194854e-06,
+      "loss": 0.8256,
+      "step": 7001
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.035552927847114e-06,
+      "loss": 0.8147,
+      "step": 7002
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0348098009544403e-06,
+      "loss": 0.8934,
+      "step": 7003
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0340668852213143e-06,
+      "loss": 1.0364,
+      "step": 7004
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0333241807275604e-06,
+      "loss": 0.802,
+      "step": 7005
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0325816875529827e-06,
+      "loss": 0.7792,
+      "step": 7006
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.031839405777362e-06,
+      "loss": 0.9916,
+      "step": 7007
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0310973354804584e-06,
+      "loss": 0.9737,
+      "step": 7008
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.030355476742004e-06,
+      "loss": 0.9474,
+      "step": 7009
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0296138296417134e-06,
+      "loss": 0.843,
+      "step": 7010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0288723942592762e-06,
+      "loss": 0.7808,
+      "step": 7011
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0281311706743608e-06,
+      "loss": 0.8981,
+      "step": 7012
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.027390158966609e-06,
+      "loss": 0.8691,
+      "step": 7013
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.026649359215644e-06,
+      "loss": 0.8532,
+      "step": 7014
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0259087715010644e-06,
+      "loss": 0.9837,
+      "step": 7015
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0251683959024473e-06,
+      "loss": 0.8622,
+      "step": 7016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0244282324993446e-06,
+      "loss": 0.8015,
+      "step": 7017
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0236882813712871e-06,
+      "loss": 1.0249,
+      "step": 7018
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0229485425977828e-06,
+      "loss": 0.7243,
+      "step": 7019
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0222090162583173e-06,
+      "loss": 0.7028,
+      "step": 7020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.02146970243235e-06,
+      "loss": 0.6859,
+      "step": 7021
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.020730601199324e-06,
+      "loss": 0.9632,
+      "step": 7022
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0199917126386527e-06,
+      "loss": 0.9226,
+      "step": 7023
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0192530368297317e-06,
+      "loss": 0.838,
+      "step": 7024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0185145738519283e-06,
+      "loss": 0.8639,
+      "step": 7025
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.017776323784595e-06,
+      "loss": 0.9798,
+      "step": 7026
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.017038286707053e-06,
+      "loss": 0.877,
+      "step": 7027
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0163004626986064e-06,
+      "loss": 0.9733,
+      "step": 7028
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0155628518385318e-06,
+      "loss": 1.1905,
+      "step": 7029
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0148254542060886e-06,
+      "loss": 0.7651,
+      "step": 7030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0140882698805086e-06,
+      "loss": 0.7491,
+      "step": 7031
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0133512989410028e-06,
+      "loss": 0.8603,
+      "step": 7032
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0126145414667562e-06,
+      "loss": 0.9814,
+      "step": 7033
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.011877997536937e-06,
+      "loss": 0.799,
+      "step": 7034
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0111416672306843e-06,
+      "loss": 0.7596,
+      "step": 7035
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0104055506271185e-06,
+      "loss": 0.9788,
+      "step": 7036
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0096696478053328e-06,
+      "loss": 1.0703,
+      "step": 7037
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.008933958844403e-06,
+      "loss": 0.8649,
+      "step": 7038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0081984838233764e-06,
+      "loss": 0.9511,
+      "step": 7039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0074632228212813e-06,
+      "loss": 1.0418,
+      "step": 7040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0067281759171188e-06,
+      "loss": 1.0438,
+      "step": 7041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0059933431898737e-06,
+      "loss": 0.8478,
+      "step": 7042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0052587247185005e-06,
+      "loss": 0.792,
+      "step": 7043
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.004524320581936e-06,
+      "loss": 0.9582,
+      "step": 7044
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0037901308590888e-06,
+      "loss": 0.8292,
+      "step": 7045
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0030561556288517e-06,
+      "loss": 0.7458,
+      "step": 7046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0023223949700872e-06,
+      "loss": 1.0316,
+      "step": 7047
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0015888489616395e-06,
+      "loss": 1.008,
+      "step": 7048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0008555176823258e-06,
+      "loss": 0.7981,
+      "step": 7049
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.000122401210946e-06,
+      "loss": 1.0081,
+      "step": 7050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.993894996262704e-07,
+      "loss": 0.9156,
+      "step": 7051
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.986568130070511e-07,
+      "loss": 0.8844,
+      "step": 7052
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.979243414320127e-07,
+      "loss": 0.8409,
+      "step": 7053
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.971920849798628e-07,
+      "loss": 0.7789,
+      "step": 7054
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.964600437292798e-07,
+      "loss": 0.7946,
+      "step": 7055
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.957282177589227e-07,
+      "loss": 0.8788,
+      "step": 7056
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.949966071474238e-07,
+      "loss": 0.9163,
+      "step": 7057
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.942652119733985e-07,
+      "loss": 0.8832,
+      "step": 7058
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.935340323154316e-07,
+      "loss": 0.7219,
+      "step": 7059
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.92803068252091e-07,
+      "loss": 0.7828,
+      "step": 7060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.920723198619158e-07,
+      "loss": 1.0312,
+      "step": 7061
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.91341787223428e-07,
+      "loss": 1.0333,
+      "step": 7062
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.906114704151213e-07,
+      "loss": 0.75,
+      "step": 7063
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.898813695154692e-07,
+      "loss": 0.9922,
+      "step": 7064
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.891514846029189e-07,
+      "loss": 0.7877,
+      "step": 7065
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.884218157558998e-07,
+      "loss": 0.8212,
+      "step": 7066
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.876923630528125e-07,
+      "loss": 1.0042,
+      "step": 7067
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.869631265720379e-07,
+      "loss": 0.8574,
+      "step": 7068
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.862341063919296e-07,
+      "loss": 0.7124,
+      "step": 7069
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.85505302590825e-07,
+      "loss": 0.8332,
+      "step": 7070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.84776715247031e-07,
+      "loss": 1.0196,
+      "step": 7071
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.84048344438836e-07,
+      "loss": 0.8777,
+      "step": 7072
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.833201902445007e-07,
+      "loss": 0.9115,
+      "step": 7073
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.825922527422687e-07,
+      "loss": 0.9632,
+      "step": 7074
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.818645320103545e-07,
+      "loss": 0.8554,
+      "step": 7075
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.811370281269532e-07,
+      "loss": 0.8906,
+      "step": 7076
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.804097411702325e-07,
+      "loss": 0.9603,
+      "step": 7077
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.796826712183425e-07,
+      "loss": 0.8953,
+      "step": 7078
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.789558183494044e-07,
+      "loss": 0.7952,
+      "step": 7079
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.782291826415205e-07,
+      "loss": 0.8352,
+      "step": 7080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.775027641727647e-07,
+      "loss": 0.7592,
+      "step": 7081
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.767765630211946e-07,
+      "loss": 0.9707,
+      "step": 7082
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.760505792648372e-07,
+      "loss": 0.8329,
+      "step": 7083
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.753248129817019e-07,
+      "loss": 0.9825,
+      "step": 7084
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.745992642497689e-07,
+      "loss": 0.9627,
+      "step": 7085
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.738739331470022e-07,
+      "loss": 0.9038,
+      "step": 7086
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.731488197513362e-07,
+      "loss": 0.8155,
+      "step": 7087
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.724239241406847e-07,
+      "loss": 0.8229,
+      "step": 7088
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.71699246392938e-07,
+      "loss": 0.8636,
+      "step": 7089
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.709747865859631e-07,
+      "loss": 0.7606,
+      "step": 7090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.70250544797602e-07,
+      "loss": 0.8526,
+      "step": 7091
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.695265211056747e-07,
+      "loss": 0.9806,
+      "step": 7092
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.68802715587978e-07,
+      "loss": 1.112,
+      "step": 7093
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.680791283222859e-07,
+      "loss": 0.8522,
+      "step": 7094
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.673557593863448e-07,
+      "loss": 0.7918,
+      "step": 7095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.666326088578827e-07,
+      "loss": 0.8545,
+      "step": 7096
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.659096768146018e-07,
+      "loss": 0.9736,
+      "step": 7097
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.651869633341817e-07,
+      "loss": 0.8617,
+      "step": 7098
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.644644684942764e-07,
+      "loss": 0.6984,
+      "step": 7099
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.637421923725187e-07,
+      "loss": 0.9869,
+      "step": 7100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.63020135046517e-07,
+      "loss": 0.7819,
+      "step": 7101
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.622982965938574e-07,
+      "loss": 1.0986,
+      "step": 7102
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.615766770920994e-07,
+      "loss": 0.9586,
+      "step": 7103
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.60855276618782e-07,
+      "loss": 0.8248,
+      "step": 7104
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.601340952514197e-07,
+      "loss": 0.9166,
+      "step": 7105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.594131330675045e-07,
+      "loss": 0.8351,
+      "step": 7106
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.586923901445012e-07,
+      "loss": 1.0555,
+      "step": 7107
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.579718665598553e-07,
+      "loss": 0.9224,
+      "step": 7108
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.572515623909861e-07,
+      "loss": 1.0633,
+      "step": 7109
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.565314777152922e-07,
+      "loss": 0.845,
+      "step": 7110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.558116126101439e-07,
+      "loss": 0.8362,
+      "step": 7111
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.550919671528924e-07,
+      "loss": 0.7566,
+      "step": 7112
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.543725414208624e-07,
+      "loss": 0.7619,
+      "step": 7113
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.536533354913584e-07,
+      "loss": 0.7315,
+      "step": 7114
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.529343494416563e-07,
+      "loss": 0.8001,
+      "step": 7115
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.522155833490123e-07,
+      "loss": 0.9307,
+      "step": 7116
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.514970372906576e-07,
+      "loss": 0.9465,
+      "step": 7117
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.507787113438014e-07,
+      "loss": 0.7748,
+      "step": 7118
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.500606055856254e-07,
+      "loss": 1.1619,
+      "step": 7119
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.493427200932906e-07,
+      "loss": 0.8031,
+      "step": 7120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.486250549439347e-07,
+      "loss": 0.8828,
+      "step": 7121
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.47907610214671e-07,
+      "loss": 0.6354,
+      "step": 7122
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.471903859825871e-07,
+      "loss": 0.8046,
+      "step": 7123
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.464733823247497e-07,
+      "loss": 0.8536,
+      "step": 7124
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.457565993182011e-07,
+      "loss": 0.6967,
+      "step": 7125
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.450400370399602e-07,
+      "loss": 0.9006,
+      "step": 7126
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.443236955670198e-07,
+      "loss": 0.6944,
+      "step": 7127
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.436075749763515e-07,
+      "loss": 1.1203,
+      "step": 7128
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.428916753449025e-07,
+      "loss": 0.9868,
+      "step": 7129
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.421759967495972e-07,
+      "loss": 0.9606,
+      "step": 7130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.414605392673331e-07,
+      "loss": 0.7938,
+      "step": 7131
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.407453029749874e-07,
+      "loss": 0.8637,
+      "step": 7132
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.400302879494119e-07,
+      "loss": 0.8242,
+      "step": 7133
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.393154942674359e-07,
+      "loss": 0.8339,
+      "step": 7134
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.386009220058618e-07,
+      "loss": 0.8894,
+      "step": 7135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.378865712414718e-07,
+      "loss": 0.9755,
+      "step": 7136
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.371724420510226e-07,
+      "loss": 0.7256,
+      "step": 7137
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.364585345112482e-07,
+      "loss": 0.9301,
+      "step": 7138
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.357448486988559e-07,
+      "loss": 0.9208,
+      "step": 7139
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.350313846905324e-07,
+      "loss": 0.8832,
+      "step": 7140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.343181425629393e-07,
+      "loss": 1.1291,
+      "step": 7141
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.336051223927155e-07,
+      "loss": 0.9667,
+      "step": 7142
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.328923242564725e-07,
+      "loss": 1.1314,
+      "step": 7143
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.32179748230802e-07,
+      "loss": 0.7579,
+      "step": 7144
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.314673943922699e-07,
+      "loss": 0.8001,
+      "step": 7145
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.307552628174196e-07,
+      "loss": 0.8538,
+      "step": 7146
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.300433535827679e-07,
+      "loss": 1.0277,
+      "step": 7147
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.2933166676481e-07,
+      "loss": 0.7127,
+      "step": 7148
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.286202024400169e-07,
+      "loss": 0.8119,
+      "step": 7149
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.279089606848359e-07,
+      "loss": 1.1709,
+      "step": 7150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.271979415756885e-07,
+      "loss": 0.8502,
+      "step": 7151
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.264871451889742e-07,
+      "loss": 1.0141,
+      "step": 7152
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.257765716010685e-07,
+      "loss": 0.9561,
+      "step": 7153
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.250662208883226e-07,
+      "loss": 0.8967,
+      "step": 7154
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.243560931270617e-07,
+      "loss": 0.8681,
+      "step": 7155
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.236461883935924e-07,
+      "loss": 0.7292,
+      "step": 7156
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.22936506764191e-07,
+      "loss": 0.9458,
+      "step": 7157
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.222270483151144e-07,
+      "loss": 1.0712,
+      "step": 7158
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.215178131225925e-07,
+      "loss": 0.8696,
+      "step": 7159
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.20808801262833e-07,
+      "loss": 1.1702,
+      "step": 7160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.201000128120197e-07,
+      "loss": 0.9397,
+      "step": 7161
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.193914478463128e-07,
+      "loss": 1.0369,
+      "step": 7162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.18683106441845e-07,
+      "loss": 0.8191,
+      "step": 7163
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.179749886747289e-07,
+      "loss": 0.906,
+      "step": 7164
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.172670946210519e-07,
+      "loss": 0.7482,
+      "step": 7165
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.165594243568779e-07,
+      "loss": 0.8159,
+      "step": 7166
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.158519779582439e-07,
+      "loss": 0.8754,
+      "step": 7167
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.151447555011663e-07,
+      "loss": 0.8885,
+      "step": 7168
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.144377570616358e-07,
+      "loss": 0.8048,
+      "step": 7169
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.137309827156207e-07,
+      "loss": 1.0388,
+      "step": 7170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.130244325390616e-07,
+      "loss": 0.8239,
+      "step": 7171
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.123181066078781e-07,
+      "loss": 0.7858,
+      "step": 7172
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.116120049979652e-07,
+      "loss": 0.9194,
+      "step": 7173
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.109061277851946e-07,
+      "loss": 0.9224,
+      "step": 7174
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.102004750454101e-07,
+      "loss": 0.8489,
+      "step": 7175
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.094950468544358e-07,
+      "loss": 1.1737,
+      "step": 7176
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.08789843288069e-07,
+      "loss": 1.0209,
+      "step": 7177
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.080848644220856e-07,
+      "loss": 0.88,
+      "step": 7178
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.073801103322336e-07,
+      "loss": 0.9117,
+      "step": 7179
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.06675581094239e-07,
+      "loss": 0.7673,
+      "step": 7180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.05971276783804e-07,
+      "loss": 0.9266,
+      "step": 7181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.052671974766066e-07,
+      "loss": 0.9331,
+      "step": 7182
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.045633432482986e-07,
+      "loss": 0.7808,
+      "step": 7183
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.038597141745099e-07,
+      "loss": 1.0405,
+      "step": 7184
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.031563103308454e-07,
+      "loss": 0.8641,
+      "step": 7185
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.024531317928862e-07,
+      "loss": 0.8183,
+      "step": 7186
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.017501786361878e-07,
+      "loss": 0.8151,
+      "step": 7187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.010474509362829e-07,
+      "loss": 1.1236,
+      "step": 7188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.003449487686795e-07,
+      "loss": 0.8543,
+      "step": 7189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.996426722088627e-07,
+      "loss": 0.9915,
+      "step": 7190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.989406213322898e-07,
+      "loss": 1.0492,
+      "step": 7191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.982387962143971e-07,
+      "loss": 0.9803,
+      "step": 7192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.975371969305962e-07,
+      "loss": 0.9396,
+      "step": 7193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.968358235562739e-07,
+      "loss": 0.87,
+      "step": 7194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.961346761667916e-07,
+      "loss": 0.7566,
+      "step": 7195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.954337548374881e-07,
+      "loss": 0.873,
+      "step": 7196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.947330596436773e-07,
+      "loss": 0.9511,
+      "step": 7197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.940325906606503e-07,
+      "loss": 0.8341,
+      "step": 7198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.933323479636702e-07,
+      "loss": 0.8252,
+      "step": 7199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.926323316279789e-07,
+      "loss": 0.8038,
+      "step": 7200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.91932541728793e-07,
+      "loss": 0.9062,
+      "step": 7201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.912329783413062e-07,
+      "loss": 1.1006,
+      "step": 7202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.905336415406844e-07,
+      "loss": 1.0372,
+      "step": 7203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.898345314020724e-07,
+      "loss": 0.994,
+      "step": 7204
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.891356480005891e-07,
+      "loss": 0.8434,
+      "step": 7205
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.884369914113311e-07,
+      "loss": 0.8015,
+      "step": 7206
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.877385617093665e-07,
+      "loss": 0.9276,
+      "step": 7207
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.870403589697431e-07,
+      "loss": 1.0067,
+      "step": 7208
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.86342383267482e-07,
+      "loss": 0.7242,
+      "step": 7209
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.856446346775821e-07,
+      "loss": 0.9407,
+      "step": 7210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.849471132750141e-07,
+      "loss": 1.167,
+      "step": 7211
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.842498191347281e-07,
+      "loss": 1.0467,
+      "step": 7212
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.835527523316483e-07,
+      "loss": 0.8378,
+      "step": 7213
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.828559129406748e-07,
+      "loss": 1.106,
+      "step": 7214
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.821593010366812e-07,
+      "loss": 0.8599,
+      "step": 7215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.814629166945202e-07,
+      "loss": 0.8196,
+      "step": 7216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.807667599890173e-07,
+      "loss": 0.8725,
+      "step": 7217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.800708309949756e-07,
+      "loss": 0.9872,
+      "step": 7218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.793751297871713e-07,
+      "loss": 0.8901,
+      "step": 7219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.786796564403577e-07,
+      "loss": 0.945,
+      "step": 7220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.779844110292636e-07,
+      "loss": 0.9393,
+      "step": 7221
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.772893936285932e-07,
+      "loss": 0.7743,
+      "step": 7222
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.765946043130269e-07,
+      "loss": 0.729,
+      "step": 7223
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.75900043157218e-07,
+      "loss": 0.8709,
+      "step": 7224
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.752057102357977e-07,
+      "loss": 0.9245,
+      "step": 7225
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.745116056233726e-07,
+      "loss": 0.7085,
+      "step": 7226
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.738177293945245e-07,
+      "loss": 0.8551,
+      "step": 7227
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.731240816238085e-07,
+      "loss": 1.155,
+      "step": 7228
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.724306623857583e-07,
+      "loss": 0.8069,
+      "step": 7229
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.717374717548819e-07,
+      "loss": 1.0062,
+      "step": 7230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.710445098056629e-07,
+      "loss": 0.9347,
+      "step": 7231
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.703517766125585e-07,
+      "loss": 0.8643,
+      "step": 7232
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.69659272250004e-07,
+      "loss": 0.8018,
+      "step": 7233
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.689669967924084e-07,
+      "loss": 0.9335,
+      "step": 7234
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.68274950314158e-07,
+      "loss": 1.0423,
+      "step": 7235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.67583132889611e-07,
+      "loss": 1.0279,
+      "step": 7236
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.668915445931042e-07,
+      "loss": 0.8082,
+      "step": 7237
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.662001854989488e-07,
+      "loss": 0.8277,
+      "step": 7238
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.655090556814323e-07,
+      "loss": 0.8882,
+      "step": 7239
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.648181552148142e-07,
+      "loss": 1.0265,
+      "step": 7240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.641274841733333e-07,
+      "loss": 0.961,
+      "step": 7241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.634370426312016e-07,
+      "loss": 0.931,
+      "step": 7242
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.627468306626081e-07,
+      "loss": 0.7398,
+      "step": 7243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.620568483417144e-07,
+      "loss": 1.0996,
+      "step": 7244
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.613670957426596e-07,
+      "loss": 0.8032,
+      "step": 7245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.606775729395578e-07,
+      "loss": 1.2966,
+      "step": 7246
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.599882800064988e-07,
+      "loss": 0.9252,
+      "step": 7247
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.592992170175457e-07,
+      "loss": 0.8785,
+      "step": 7248
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.586103840467392e-07,
+      "loss": 0.759,
+      "step": 7249
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.579217811680939e-07,
+      "loss": 0.7976,
+      "step": 7250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.572334084556015e-07,
+      "loss": 0.8906,
+      "step": 7251
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.565452659832253e-07,
+      "loss": 0.9473,
+      "step": 7252
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.558573538249074e-07,
+      "loss": 0.9632,
+      "step": 7253
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.551696720545637e-07,
+      "loss": 0.9636,
+      "step": 7254
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.544822207460868e-07,
+      "loss": 0.8138,
+      "step": 7255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.537949999733412e-07,
+      "loss": 0.8632,
+      "step": 7256
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.531080098101698e-07,
+      "loss": 0.8853,
+      "step": 7257
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.524212503303894e-07,
+      "loss": 0.7394,
+      "step": 7258
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.517347216077931e-07,
+      "loss": 0.9825,
+      "step": 7259
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.510484237161471e-07,
+      "loss": 0.8937,
+      "step": 7260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.503623567291945e-07,
+      "loss": 0.9148,
+      "step": 7261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.496765207206532e-07,
+      "loss": 1.1823,
+      "step": 7262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.489909157642175e-07,
+      "loss": 1.0261,
+      "step": 7263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.483055419335532e-07,
+      "loss": 0.8799,
+      "step": 7264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.476203993023049e-07,
+      "loss": 0.777,
+      "step": 7265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.469354879440914e-07,
+      "loss": 1.0947,
+      "step": 7266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.462508079325066e-07,
+      "loss": 0.9287,
+      "step": 7267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.455663593411182e-07,
+      "loss": 0.8867,
+      "step": 7268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.448821422434707e-07,
+      "loss": 0.841,
+      "step": 7269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.441981567130832e-07,
+      "loss": 1.0281,
+      "step": 7270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.435144028234506e-07,
+      "loss": 0.7633,
+      "step": 7271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.428308806480412e-07,
+      "loss": 0.7702,
+      "step": 7272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.42147590260299e-07,
+      "loss": 0.7581,
+      "step": 7273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.414645317336447e-07,
+      "loss": 0.7504,
+      "step": 7274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.407817051414732e-07,
+      "loss": 0.8066,
+      "step": 7275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.400991105571527e-07,
+      "loss": 0.937,
+      "step": 7276
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.394167480540285e-07,
+      "loss": 0.9826,
+      "step": 7277
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.387346177054205e-07,
+      "loss": 0.7538,
+      "step": 7278
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.380527195846244e-07,
+      "loss": 0.8018,
+      "step": 7279
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.373710537649088e-07,
+      "loss": 0.8776,
+      "step": 7280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.366896203195187e-07,
+      "loss": 0.8101,
+      "step": 7281
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.360084193216747e-07,
+      "loss": 0.8265,
+      "step": 7282
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.353274508445728e-07,
+      "loss": 0.9107,
+      "step": 7283
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.346467149613806e-07,
+      "loss": 0.7321,
+      "step": 7284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.339662117452453e-07,
+      "loss": 0.9562,
+      "step": 7285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.332859412692842e-07,
+      "loss": 0.8093,
+      "step": 7286
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.326059036065962e-07,
+      "loss": 1.0632,
+      "step": 7287
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.31926098830248e-07,
+      "loss": 0.9873,
+      "step": 7288
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.312465270132855e-07,
+      "loss": 0.7481,
+      "step": 7289
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.305671882287294e-07,
+      "loss": 0.983,
+      "step": 7290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.298880825495747e-07,
+      "loss": 0.7828,
+      "step": 7291
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.292092100487899e-07,
+      "loss": 0.8977,
+      "step": 7292
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.285305707993207e-07,
+      "loss": 0.9454,
+      "step": 7293
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.278521648740864e-07,
+      "loss": 0.7993,
+      "step": 7294
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.271739923459831e-07,
+      "loss": 0.8439,
+      "step": 7295
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.264960532878785e-07,
+      "loss": 0.9723,
+      "step": 7296
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.258183477726179e-07,
+      "loss": 0.7659,
+      "step": 7297
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.251408758730204e-07,
+      "loss": 0.8238,
+      "step": 7298
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.244636376618816e-07,
+      "loss": 0.7856,
+      "step": 7299
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.237866332119688e-07,
+      "loss": 0.8813,
+      "step": 7300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.231098625960268e-07,
+      "loss": 0.9926,
+      "step": 7301
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.224333258867749e-07,
+      "loss": 1.1137,
+      "step": 7302
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.217570231569078e-07,
+      "loss": 0.8462,
+      "step": 7303
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.210809544790921e-07,
+      "loss": 1.1649,
+      "step": 7304
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.204051199259721e-07,
+      "loss": 0.8149,
+      "step": 7305
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.19729519570167e-07,
+      "loss": 0.9426,
+      "step": 7306
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.1905415348427e-07,
+      "loss": 0.9498,
+      "step": 7307
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.183790217408476e-07,
+      "loss": 0.7899,
+      "step": 7308
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.177041244124435e-07,
+      "loss": 1.194,
+      "step": 7309
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.170294615715759e-07,
+      "loss": 0.8114,
+      "step": 7310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.163550332907376e-07,
+      "loss": 1.0289,
+      "step": 7311
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.156808396423943e-07,
+      "loss": 1.0659,
+      "step": 7312
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.150068806989887e-07,
+      "loss": 0.7266,
+      "step": 7313
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.143331565329382e-07,
+      "loss": 1.0634,
+      "step": 7314
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.136596672166346e-07,
+      "loss": 0.9357,
+      "step": 7315
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.12986412822443e-07,
+      "loss": 0.9142,
+      "step": 7316
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.123133934227054e-07,
+      "loss": 0.7234,
+      "step": 7317
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.116406090897374e-07,
+      "loss": 1.0619,
+      "step": 7318
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.109680598958307e-07,
+      "loss": 0.9543,
+      "step": 7319
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.102957459132491e-07,
+      "loss": 0.8676,
+      "step": 7320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.096236672142332e-07,
+      "loss": 0.9199,
+      "step": 7321
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.089518238709978e-07,
+      "loss": 0.9184,
+      "step": 7322
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.082802159557337e-07,
+      "loss": 0.9006,
+      "step": 7323
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.076088435406028e-07,
+      "loss": 1.0599,
+      "step": 7324
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.069377066977455e-07,
+      "loss": 1.0387,
+      "step": 7325
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.06266805499275e-07,
+      "loss": 0.9081,
+      "step": 7326
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.055961400172808e-07,
+      "loss": 0.9926,
+      "step": 7327
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.049257103238236e-07,
+      "loss": 0.8908,
+      "step": 7328
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.042555164909424e-07,
+      "loss": 0.7537,
+      "step": 7329
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.03585558590649e-07,
+      "loss": 1.0474,
+      "step": 7330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.02915836694932e-07,
+      "loss": 0.7432,
+      "step": 7331
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.022463508757501e-07,
+      "loss": 0.8103,
+      "step": 7332
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.015771012050415e-07,
+      "loss": 0.9893,
+      "step": 7333
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.00908087754716e-07,
+      "loss": 0.9498,
+      "step": 7334
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.002393105966607e-07,
+      "loss": 0.8921,
+      "step": 7335
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.995707698027335e-07,
+      "loss": 0.9996,
+      "step": 7336
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.989024654447703e-07,
+      "loss": 0.9327,
+      "step": 7337
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.982343975945797e-07,
+      "loss": 0.8578,
+      "step": 7338
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.975665663239471e-07,
+      "loss": 1.0309,
+      "step": 7339
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.96898971704629e-07,
+      "loss": 1.0022,
+      "step": 7340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.962316138083591e-07,
+      "loss": 1.2361,
+      "step": 7341
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.95564492706845e-07,
+      "loss": 0.8481,
+      "step": 7342
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.948976084717696e-07,
+      "loss": 0.7888,
+      "step": 7343
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.942309611747881e-07,
+      "loss": 0.932,
+      "step": 7344
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.935645508875324e-07,
+      "loss": 0.828,
+      "step": 7345
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.928983776816084e-07,
+      "loss": 0.7522,
+      "step": 7346
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.922324416285969e-07,
+      "loss": 0.9776,
+      "step": 7347
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.915667428000514e-07,
+      "loss": 0.8516,
+      "step": 7348
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.909012812675016e-07,
+      "loss": 0.8838,
+      "step": 7349
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.902360571024514e-07,
+      "loss": 0.7456,
+      "step": 7350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.8957107037638e-07,
+      "loss": 0.9663,
+      "step": 7351
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.889063211607389e-07,
+      "loss": 0.8825,
+      "step": 7352
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.882418095269554e-07,
+      "loss": 0.8225,
+      "step": 7353
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.875775355464319e-07,
+      "loss": 0.8814,
+      "step": 7354
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.86913499290545e-07,
+      "loss": 0.8702,
+      "step": 7355
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.86249700830643e-07,
+      "loss": 1.0794,
+      "step": 7356
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.855861402380543e-07,
+      "loss": 0.8924,
+      "step": 7357
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.84922817584076e-07,
+      "loss": 0.9579,
+      "step": 7358
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.842597329399839e-07,
+      "loss": 0.9061,
+      "step": 7359
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.835968863770235e-07,
+      "loss": 0.8381,
+      "step": 7360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.829342779664214e-07,
+      "loss": 0.8636,
+      "step": 7361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.822719077793721e-07,
+      "loss": 0.6912,
+      "step": 7362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.816097758870487e-07,
+      "loss": 0.8808,
+      "step": 7363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.809478823605951e-07,
+      "loss": 0.9122,
+      "step": 7364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.802862272711346e-07,
+      "loss": 1.0367,
+      "step": 7365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.796248106897598e-07,
+      "loss": 0.9266,
+      "step": 7366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.789636326875417e-07,
+      "loss": 0.9236,
+      "step": 7367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.783026933355208e-07,
+      "loss": 0.9301,
+      "step": 7368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.776419927047192e-07,
+      "loss": 0.8534,
+      "step": 7369
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.769815308661254e-07,
+      "loss": 0.719,
+      "step": 7370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.763213078907087e-07,
+      "loss": 0.9282,
+      "step": 7371
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.75661323849407e-07,
+      "loss": 0.999,
+      "step": 7372
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.750015788131391e-07,
+      "loss": 0.9026,
+      "step": 7373
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.743420728527917e-07,
+      "loss": 0.807,
+      "step": 7374
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.73682806039231e-07,
+      "loss": 0.977,
+      "step": 7375
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.730237784432919e-07,
+      "loss": 0.9897,
+      "step": 7376
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.723649901357906e-07,
+      "loss": 0.9077,
+      "step": 7377
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.717064411875113e-07,
+      "loss": 0.9175,
+      "step": 7378
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.710481316692169e-07,
+      "loss": 0.873,
+      "step": 7379
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.703900616516398e-07,
+      "loss": 1.0283,
+      "step": 7380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.697322312054934e-07,
+      "loss": 0.7552,
+      "step": 7381
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.690746404014585e-07,
+      "loss": 0.9323,
+      "step": 7382
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.684172893101955e-07,
+      "loss": 0.8412,
+      "step": 7383
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.677601780023336e-07,
+      "loss": 0.8896,
+      "step": 7384
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.671033065484828e-07,
+      "loss": 1.1749,
+      "step": 7385
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.664466750192218e-07,
+      "loss": 0.9042,
+      "step": 7386
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.657902834851069e-07,
+      "loss": 1.0901,
+      "step": 7387
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.651341320166651e-07,
+      "loss": 1.088,
+      "step": 7388
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.644782206844028e-07,
+      "loss": 1.1047,
+      "step": 7389
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.638225495587954e-07,
+      "loss": 0.8567,
+      "step": 7390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.631671187102965e-07,
+      "loss": 1.1336,
+      "step": 7391
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.625119282093292e-07,
+      "loss": 0.8438,
+      "step": 7392
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.618569781262971e-07,
+      "loss": 0.8927,
+      "step": 7393
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.612022685315723e-07,
+      "loss": 0.7267,
+      "step": 7394
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.605477994955049e-07,
+      "loss": 0.8531,
+      "step": 7395
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.598935710884149e-07,
+      "loss": 0.9113,
+      "step": 7396
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.592395833806026e-07,
+      "loss": 0.8517,
+      "step": 7397
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.585858364423364e-07,
+      "loss": 0.7279,
+      "step": 7398
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.579323303438629e-07,
+      "loss": 0.8726,
+      "step": 7399
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.572790651553987e-07,
+      "loss": 0.9937,
+      "step": 7400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.566260409471406e-07,
+      "loss": 0.8522,
+      "step": 7401
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.559732577892536e-07,
+      "loss": 0.6691,
+      "step": 7402
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.553207157518806e-07,
+      "loss": 1.1136,
+      "step": 7403
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.546684149051349e-07,
+      "loss": 0.7998,
+      "step": 7404
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.540163553191096e-07,
+      "loss": 0.9884,
+      "step": 7405
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.533645370638656e-07,
+      "loss": 0.9285,
+      "step": 7406
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.527129602094427e-07,
+      "loss": 0.7079,
+      "step": 7407
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.5206162482585e-07,
+      "loss": 1.0801,
+      "step": 7408
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.514105309830774e-07,
+      "loss": 0.8318,
+      "step": 7409
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.507596787510818e-07,
+      "loss": 0.8021,
+      "step": 7410
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.501090681997991e-07,
+      "loss": 1.1928,
+      "step": 7411
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.494586993991345e-07,
+      "loss": 1.0073,
+      "step": 7412
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.488085724189742e-07,
+      "loss": 0.6994,
+      "step": 7413
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.481586873291709e-07,
+      "loss": 0.9753,
+      "step": 7414
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.475090441995569e-07,
+      "loss": 0.9284,
+      "step": 7415
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.468596430999336e-07,
+      "loss": 1.136,
+      "step": 7416
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.462104841000828e-07,
+      "loss": 0.8784,
+      "step": 7417
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.455615672697536e-07,
+      "loss": 0.8276,
+      "step": 7418
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.449128926786743e-07,
+      "loss": 1.0349,
+      "step": 7419
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.442644603965415e-07,
+      "loss": 0.8779,
+      "step": 7420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.436162704930336e-07,
+      "loss": 0.8655,
+      "step": 7421
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.429683230377955e-07,
+      "loss": 0.9799,
+      "step": 7422
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.423206181004499e-07,
+      "loss": 0.7304,
+      "step": 7423
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.416731557505929e-07,
+      "loss": 0.8815,
+      "step": 7424
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.410259360577951e-07,
+      "loss": 1.1001,
+      "step": 7425
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.403789590915981e-07,
+      "loss": 0.928,
+      "step": 7426
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.39732224921521e-07,
+      "loss": 0.9189,
+      "step": 7427
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.390857336170549e-07,
+      "loss": 0.846,
+      "step": 7428
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.384394852476665e-07,
+      "loss": 0.6707,
+      "step": 7429
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.377934798827929e-07,
+      "loss": 0.9191,
+      "step": 7430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.371477175918486e-07,
+      "loss": 0.8146,
+      "step": 7431
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.365021984442205e-07,
+      "loss": 0.9885,
+      "step": 7432
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.358569225092704e-07,
+      "loss": 0.9329,
+      "step": 7433
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.352118898563318e-07,
+      "loss": 0.9548,
+      "step": 7434
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.345671005547139e-07,
+      "loss": 1.0497,
+      "step": 7435
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.339225546736991e-07,
+      "loss": 0.9436,
+      "step": 7436
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.332782522825453e-07,
+      "loss": 0.802,
+      "step": 7437
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.326341934504806e-07,
+      "loss": 0.9326,
+      "step": 7438
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.319903782467102e-07,
+      "loss": 0.7717,
+      "step": 7439
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.313468067404117e-07,
+      "loss": 0.7179,
+      "step": 7440
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.307034790007376e-07,
+      "loss": 0.7544,
+      "step": 7441
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.300603950968122e-07,
+      "loss": 0.9263,
+      "step": 7442
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.294175550977351e-07,
+      "loss": 0.8963,
+      "step": 7443
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.287749590725797e-07,
+      "loss": 1.1683,
+      "step": 7444
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.281326070903939e-07,
+      "loss": 0.9533,
+      "step": 7445
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.274904992201967e-07,
+      "loss": 0.8576,
+      "step": 7446
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.268486355309828e-07,
+      "loss": 0.938,
+      "step": 7447
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.262070160917209e-07,
+      "loss": 0.8195,
+      "step": 7448
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.255656409713534e-07,
+      "loss": 0.7969,
+      "step": 7449
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.24924510238795e-07,
+      "loss": 0.7858,
+      "step": 7450
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.242836239629358e-07,
+      "loss": 0.9187,
+      "step": 7451
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.236429822126387e-07,
+      "loss": 0.7086,
+      "step": 7452
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.230025850567413e-07,
+      "loss": 0.8162,
+      "step": 7453
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.22362432564053e-07,
+      "loss": 0.8682,
+      "step": 7454
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.217225248033586e-07,
+      "loss": 0.9228,
+      "step": 7455
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.210828618434167e-07,
+      "loss": 0.888,
+      "step": 7456
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.204434437529593e-07,
+      "loss": 0.6331,
+      "step": 7457
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.198042706006905e-07,
+      "loss": 0.9279,
+      "step": 7458
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.191653424552899e-07,
+      "loss": 0.7514,
+      "step": 7459
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.185266593854104e-07,
+      "loss": 1.1746,
+      "step": 7460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.178882214596798e-07,
+      "loss": 0.8097,
+      "step": 7461
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.17250028746696e-07,
+      "loss": 0.8571,
+      "step": 7462
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.166120813150334e-07,
+      "loss": 0.8615,
+      "step": 7463
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.1597437923324e-07,
+      "loss": 0.8783,
+      "step": 7464
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.153369225698374e-07,
+      "loss": 0.9609,
+      "step": 7465
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.146997113933184e-07,
+      "loss": 0.9675,
+      "step": 7466
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.140627457721527e-07,
+      "loss": 0.8527,
+      "step": 7467
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.134260257747813e-07,
+      "loss": 0.9077,
+      "step": 7468
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.127895514696215e-07,
+      "loss": 1.0858,
+      "step": 7469
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.121533229250603e-07,
+      "loss": 0.9302,
+      "step": 7470
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.115173402094613e-07,
+      "loss": 1.0243,
+      "step": 7471
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.108816033911607e-07,
+      "loss": 0.8831,
+      "step": 7472
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.102461125384694e-07,
+      "loss": 0.9573,
+      "step": 7473
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.09610867719669e-07,
+      "loss": 0.9398,
+      "step": 7474
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.089758690030176e-07,
+      "loss": 0.9519,
+      "step": 7475
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.083411164567456e-07,
+      "loss": 0.9624,
+      "step": 7476
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.077066101490577e-07,
+      "loss": 0.8321,
+      "step": 7477
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.070723501481303e-07,
+      "loss": 0.8352,
+      "step": 7478
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.06438336522115e-07,
+      "loss": 0.8758,
+      "step": 7479
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.05804569339137e-07,
+      "loss": 1.0054,
+      "step": 7480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.05171048667295e-07,
+      "loss": 0.8958,
+      "step": 7481
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.045377745746597e-07,
+      "loss": 0.8075,
+      "step": 7482
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.039047471292761e-07,
+      "loss": 0.8432,
+      "step": 7483
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.032719663991638e-07,
+      "loss": 1.028,
+      "step": 7484
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.02639432452316e-07,
+      "loss": 0.8788,
+      "step": 7485
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.020071453566959e-07,
+      "loss": 0.8987,
+      "step": 7486
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.013751051802444e-07,
+      "loss": 0.9782,
+      "step": 7487
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.007433119908737e-07,
+      "loss": 0.9662,
+      "step": 7488
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.001117658564707e-07,
+      "loss": 0.824,
+      "step": 7489
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.994804668448939e-07,
+      "loss": 1.1885,
+      "step": 7490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.988494150239764e-07,
+      "loss": 0.8744,
+      "step": 7491
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.982186104615253e-07,
+      "loss": 0.8674,
+      "step": 7492
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.975880532253209e-07,
+      "loss": 0.9561,
+      "step": 7493
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.969577433831152e-07,
+      "loss": 0.7053,
+      "step": 7494
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.963276810026357e-07,
+      "loss": 0.7363,
+      "step": 7495
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.95697866151582e-07,
+      "loss": 0.9197,
+      "step": 7496
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.950682988976296e-07,
+      "loss": 0.7064,
+      "step": 7497
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.944389793084228e-07,
+      "loss": 0.8171,
+      "step": 7498
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.938099074515833e-07,
+      "loss": 0.8865,
+      "step": 7499
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.931810833947047e-07,
+      "loss": 0.8677,
+      "step": 7500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.92552507205355e-07,
+      "loss": 0.8853,
+      "step": 7501
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.919241789510728e-07,
+      "loss": 0.7557,
+      "step": 7502
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.912960986993726e-07,
+      "loss": 0.841,
+      "step": 7503
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.906682665177423e-07,
+      "loss": 0.9461,
+      "step": 7504
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.900406824736424e-07,
+      "loss": 0.9437,
+      "step": 7505
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.89413346634506e-07,
+      "loss": 0.8269,
+      "step": 7506
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.887862590677401e-07,
+      "loss": 0.9287,
+      "step": 7507
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.881594198407259e-07,
+      "loss": 1.0388,
+      "step": 7508
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.87532829020818e-07,
+      "loss": 0.8535,
+      "step": 7509
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.869064866753419e-07,
+      "loss": 0.9615,
+      "step": 7510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.862803928715985e-07,
+      "loss": 0.9567,
+      "step": 7511
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.856545476768618e-07,
+      "loss": 0.8097,
+      "step": 7512
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.850289511583798e-07,
+      "loss": 1.1117,
+      "step": 7513
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.844036033833709e-07,
+      "loss": 0.8382,
+      "step": 7514
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.837785044190295e-07,
+      "loss": 0.9604,
+      "step": 7515
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.831536543325227e-07,
+      "loss": 0.9503,
+      "step": 7516
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.825290531909912e-07,
+      "loss": 0.8373,
+      "step": 7517
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.819047010615466e-07,
+      "loss": 0.8863,
+      "step": 7518
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.812805980112766e-07,
+      "loss": 0.9506,
+      "step": 7519
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.806567441072406e-07,
+      "loss": 0.9264,
+      "step": 7520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.800331394164731e-07,
+      "loss": 0.9686,
+      "step": 7521
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.794097840059782e-07,
+      "loss": 0.8393,
+      "step": 7522
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.787866779427366e-07,
+      "loss": 0.8671,
+      "step": 7523
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.781638212937005e-07,
+      "loss": 0.9978,
+      "step": 7524
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.775412141257972e-07,
+      "loss": 0.9438,
+      "step": 7525
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.769188565059238e-07,
+      "loss": 0.9153,
+      "step": 7526
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.762967485009534e-07,
+      "loss": 1.0263,
+      "step": 7527
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.756748901777318e-07,
+      "loss": 1.0928,
+      "step": 7528
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.750532816030782e-07,
+      "loss": 0.8947,
+      "step": 7529
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.744319228437827e-07,
+      "loss": 0.8573,
+      "step": 7530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.73810813966611e-07,
+      "loss": 0.7961,
+      "step": 7531
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.731899550383014e-07,
+      "loss": 0.6579,
+      "step": 7532
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.725693461255661e-07,
+      "loss": 0.7531,
+      "step": 7533
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.719489872950877e-07,
+      "loss": 1.1253,
+      "step": 7534
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.713288786135243e-07,
+      "loss": 0.8856,
+      "step": 7535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.707090201475067e-07,
+      "loss": 0.8989,
+      "step": 7536
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.700894119636395e-07,
+      "loss": 0.8987,
+      "step": 7537
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.694700541284977e-07,
+      "loss": 0.9426,
+      "step": 7538
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.688509467086327e-07,
+      "loss": 0.9201,
+      "step": 7539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.682320897705668e-07,
+      "loss": 0.9321,
+      "step": 7540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.676134833807976e-07,
+      "loss": 0.8725,
+      "step": 7541
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.669951276057921e-07,
+      "loss": 0.797,
+      "step": 7542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.663770225119936e-07,
+      "loss": 0.925,
+      "step": 7543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.657591681658179e-07,
+      "loss": 1.0634,
+      "step": 7544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.651415646336536e-07,
+      "loss": 0.9552,
+      "step": 7545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.645242119818606e-07,
+      "loss": 0.9419,
+      "step": 7546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.639071102767745e-07,
+      "loss": 0.8971,
+      "step": 7547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.632902595847026e-07,
+      "loss": 0.7732,
+      "step": 7548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.626736599719267e-07,
+      "loss": 0.792,
+      "step": 7549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.62057311504698e-07,
+      "loss": 0.7365,
+      "step": 7550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.614412142492446e-07,
+      "loss": 1.0805,
+      "step": 7551
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.608253682717667e-07,
+      "loss": 1.0099,
+      "step": 7552
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.602097736384345e-07,
+      "loss": 0.9444,
+      "step": 7553
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.595944304153964e-07,
+      "loss": 0.9189,
+      "step": 7554
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.589793386687691e-07,
+      "loss": 0.6577,
+      "step": 7555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.583644984646459e-07,
+      "loss": 0.825,
+      "step": 7556
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.577499098690885e-07,
+      "loss": 0.8252,
+      "step": 7557
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.571355729481378e-07,
+      "loss": 0.7697,
+      "step": 7558
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.565214877678014e-07,
+      "loss": 0.8379,
+      "step": 7559
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.55907654394065e-07,
+      "loss": 0.9417,
+      "step": 7560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.552940728928823e-07,
+      "loss": 1.0009,
+      "step": 7561
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.546807433301858e-07,
+      "loss": 0.9677,
+      "step": 7562
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.540676657718752e-07,
+      "loss": 0.7718,
+      "step": 7563
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.534548402838275e-07,
+      "loss": 0.7881,
+      "step": 7564
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.528422669318876e-07,
+      "loss": 1.0669,
+      "step": 7565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.522299457818806e-07,
+      "loss": 0.7585,
+      "step": 7566
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.516178768995975e-07,
+      "loss": 1.0751,
+      "step": 7567
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.510060603508071e-07,
+      "loss": 0.962,
+      "step": 7568
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.503944962012459e-07,
+      "loss": 0.7428,
+      "step": 7569
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.497831845166304e-07,
+      "loss": 1.2438,
+      "step": 7570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.491721253626428e-07,
+      "loss": 1.0361,
+      "step": 7571
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.485613188049436e-07,
+      "loss": 0.8778,
+      "step": 7572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.479507649091609e-07,
+      "loss": 1.1807,
+      "step": 7573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.47340463740903e-07,
+      "loss": 0.8352,
+      "step": 7574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.467304153657431e-07,
+      "loss": 1.153,
+      "step": 7575
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.461206198492329e-07,
+      "loss": 0.7898,
+      "step": 7576
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.455110772568927e-07,
+      "loss": 0.8553,
+      "step": 7577
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.449017876542209e-07,
+      "loss": 0.8296,
+      "step": 7578
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.44292751106683e-07,
+      "loss": 0.6918,
+      "step": 7579
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.436839676797218e-07,
+      "loss": 0.9482,
+      "step": 7580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.430754374387485e-07,
+      "loss": 0.943,
+      "step": 7581
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.424671604491529e-07,
+      "loss": 0.8859,
+      "step": 7582
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.418591367762919e-07,
+      "loss": 1.0033,
+      "step": 7583
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.41251366485499e-07,
+      "loss": 0.764,
+      "step": 7584
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.406438496420768e-07,
+      "loss": 0.9505,
+      "step": 7585
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.400365863113067e-07,
+      "loss": 0.7624,
+      "step": 7586
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.394295765584356e-07,
+      "loss": 0.807,
+      "step": 7587
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.388228204486894e-07,
+      "loss": 0.9547,
+      "step": 7588
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.382163180472604e-07,
+      "loss": 0.9457,
+      "step": 7589
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.376100694193218e-07,
+      "loss": 0.9385,
+      "step": 7590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.370040746300112e-07,
+      "loss": 1.0483,
+      "step": 7591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.363983337444453e-07,
+      "loss": 0.956,
+      "step": 7592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.357928468277084e-07,
+      "loss": 1.0805,
+      "step": 7593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.351876139448627e-07,
+      "loss": 0.9277,
+      "step": 7594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.345826351609384e-07,
+      "loss": 0.8721,
+      "step": 7595
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.339779105409419e-07,
+      "loss": 0.9782,
+      "step": 7596
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.333734401498486e-07,
+      "loss": 0.9442,
+      "step": 7597
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.327692240526117e-07,
+      "loss": 0.9312,
+      "step": 7598
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.321652623141522e-07,
+      "loss": 0.7836,
+      "step": 7599
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.315615549993671e-07,
+      "loss": 0.8885,
+      "step": 7600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.309581021731219e-07,
+      "loss": 1.1486,
+      "step": 7601
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.303549039002614e-07,
+      "loss": 0.8918,
+      "step": 7602
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.297519602455968e-07,
+      "loss": 0.7073,
+      "step": 7603
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.291492712739156e-07,
+      "loss": 0.779,
+      "step": 7604
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.285468370499742e-07,
+      "loss": 0.8643,
+      "step": 7605
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.27944657638508e-07,
+      "loss": 0.9395,
+      "step": 7606
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.273427331042178e-07,
+      "loss": 0.889,
+      "step": 7607
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.267410635117826e-07,
+      "loss": 0.9408,
+      "step": 7608
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.261396489258492e-07,
+      "loss": 1.0832,
+      "step": 7609
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.255384894110427e-07,
+      "loss": 0.9211,
+      "step": 7610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.249375850319552e-07,
+      "loss": 0.8805,
+      "step": 7611
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.243369358531556e-07,
+      "loss": 0.9317,
+      "step": 7612
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.237365419391812e-07,
+      "loss": 1.0597,
+      "step": 7613
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.231364033545475e-07,
+      "loss": 0.9285,
+      "step": 7614
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.225365201637365e-07,
+      "loss": 1.0229,
+      "step": 7615
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.219368924312078e-07,
+      "loss": 0.9293,
+      "step": 7616
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.213375202213887e-07,
+      "loss": 0.8987,
+      "step": 7617
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.207384035986847e-07,
+      "loss": 1.0629,
+      "step": 7618
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.201395426274687e-07,
+      "loss": 0.9628,
+      "step": 7619
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.195409373720893e-07,
+      "loss": 0.8356,
+      "step": 7620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.189425878968662e-07,
+      "loss": 0.7758,
+      "step": 7621
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.183444942660928e-07,
+      "loss": 0.819,
+      "step": 7622
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.177466565440328e-07,
+      "loss": 1.0032,
+      "step": 7623
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.171490747949245e-07,
+      "loss": 0.8427,
+      "step": 7624
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.165517490829781e-07,
+      "loss": 0.7415,
+      "step": 7625
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.159546794723771e-07,
+      "loss": 0.7944,
+      "step": 7626
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.153578660272747e-07,
+      "loss": 1.0042,
+      "step": 7627
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.147613088117996e-07,
+      "loss": 1.0276,
+      "step": 7628
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.141650078900514e-07,
+      "loss": 0.8412,
+      "step": 7629
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.135689633261038e-07,
+      "loss": 0.8229,
+      "step": 7630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.129731751840003e-07,
+      "loss": 0.9255,
+      "step": 7631
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.123776435277583e-07,
+      "loss": 0.8498,
+      "step": 7632
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.117823684213685e-07,
+      "loss": 0.8842,
+      "step": 7633
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.111873499287935e-07,
+      "loss": 0.8557,
+      "step": 7634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.105925881139664e-07,
+      "loss": 0.8867,
+      "step": 7635
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.09998083040795e-07,
+      "loss": 0.9138,
+      "step": 7636
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.094038347731596e-07,
+      "loss": 0.8848,
+      "step": 7637
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.088098433749117e-07,
+      "loss": 0.727,
+      "step": 7638
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.082161089098753e-07,
+      "loss": 0.7655,
+      "step": 7639
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.07622631441847e-07,
+      "loss": 0.8224,
+      "step": 7640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.070294110345961e-07,
+      "loss": 0.8282,
+      "step": 7641
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.064364477518655e-07,
+      "loss": 0.9875,
+      "step": 7642
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.058437416573664e-07,
+      "loss": 0.7536,
+      "step": 7643
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.052512928147869e-07,
+      "loss": 0.8827,
+      "step": 7644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.046591012877848e-07,
+      "loss": 1.0034,
+      "step": 7645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.040671671399923e-07,
+      "loss": 0.8943,
+      "step": 7646
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.03475490435011e-07,
+      "loss": 1.0488,
+      "step": 7647
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.028840712364171e-07,
+      "loss": 0.786,
+      "step": 7648
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.022929096077586e-07,
+      "loss": 0.6305,
+      "step": 7649
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.017020056125567e-07,
+      "loss": 0.9802,
+      "step": 7650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.011113593143027e-07,
+      "loss": 0.8576,
+      "step": 7651
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.005209707764615e-07,
+      "loss": 0.8194,
+      "step": 7652
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.99930840062471e-07,
+      "loss": 1.0372,
+      "step": 7653
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.993409672357414e-07,
+      "loss": 1.059,
+      "step": 7654
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.987513523596526e-07,
+      "loss": 0.7694,
+      "step": 7655
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.981619954975597e-07,
+      "loss": 0.869,
+      "step": 7656
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.97572896712789e-07,
+      "loss": 0.7695,
+      "step": 7657
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.969840560686397e-07,
+      "loss": 0.8373,
+      "step": 7658
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.963954736283815e-07,
+      "loss": 0.8951,
+      "step": 7659
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.958071494552578e-07,
+      "loss": 1.2095,
+      "step": 7660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.952190836124847e-07,
+      "loss": 0.837,
+      "step": 7661
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.946312761632501e-07,
+      "loss": 0.865,
+      "step": 7662
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.940437271707126e-07,
+      "loss": 0.6568,
+      "step": 7663
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.934564366980049e-07,
+      "loss": 1.0285,
+      "step": 7664
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.928694048082313e-07,
+      "loss": 1.1463,
+      "step": 7665
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.922826315644692e-07,
+      "loss": 0.7961,
+      "step": 7666
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.916961170297658e-07,
+      "loss": 0.7931,
+      "step": 7667
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.911098612671427e-07,
+      "loss": 0.7146,
+      "step": 7668
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.90523864339593e-07,
+      "loss": 0.842,
+      "step": 7669
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.899381263100831e-07,
+      "loss": 0.9911,
+      "step": 7670
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.893526472415487e-07,
+      "loss": 0.9649,
+      "step": 7671
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.887674271969003e-07,
+      "loss": 0.96,
+      "step": 7672
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.881824662390197e-07,
+      "loss": 1.0115,
+      "step": 7673
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.875977644307618e-07,
+      "loss": 1.0381,
+      "step": 7674
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.870133218349515e-07,
+      "loss": 0.7622,
+      "step": 7675
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.864291385143871e-07,
+      "loss": 0.8103,
+      "step": 7676
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8584521453184e-07,
+      "loss": 0.8705,
+      "step": 7677
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.852615499500529e-07,
+      "loss": 1.0694,
+      "step": 7678
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.846781448317391e-07,
+      "loss": 1.0454,
+      "step": 7679
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.840949992395862e-07,
+      "loss": 0.8624,
+      "step": 7680
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.835121132362534e-07,
+      "loss": 1.0411,
+      "step": 7681
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.829294868843727e-07,
+      "loss": 1.1551,
+      "step": 7682
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.823471202465448e-07,
+      "loss": 0.9614,
+      "step": 7683
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.817650133853465e-07,
+      "loss": 0.9048,
+      "step": 7684
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.81183166363325e-07,
+      "loss": 0.8626,
+      "step": 7685
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.806015792430007e-07,
+      "loss": 1.0661,
+      "step": 7686
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.800202520868623e-07,
+      "loss": 0.9711,
+      "step": 7687
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.794391849573763e-07,
+      "loss": 0.7792,
+      "step": 7688
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.788583779169771e-07,
+      "loss": 1.0243,
+      "step": 7689
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.782778310280728e-07,
+      "loss": 0.8962,
+      "step": 7690
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.776975443530409e-07,
+      "loss": 1.2037,
+      "step": 7691
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.77117517954237e-07,
+      "loss": 0.8311,
+      "step": 7692
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.76537751893982e-07,
+      "loss": 0.956,
+      "step": 7693
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.759582462345734e-07,
+      "loss": 0.8604,
+      "step": 7694
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.753790010382768e-07,
+      "loss": 1.0829,
+      "step": 7695
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.748000163673348e-07,
+      "loss": 1.0475,
+      "step": 7696
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.742212922839574e-07,
+      "loss": 0.9868,
+      "step": 7697
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.736428288503301e-07,
+      "loss": 0.7862,
+      "step": 7698
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.730646261286057e-07,
+      "loss": 0.7392,
+      "step": 7699
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.724866841809155e-07,
+      "loss": 0.9327,
+      "step": 7700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.719090030693576e-07,
+      "loss": 0.8517,
+      "step": 7701
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.713315828560046e-07,
+      "loss": 0.991,
+      "step": 7702
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.707544236028981e-07,
+      "loss": 0.9132,
+      "step": 7703
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.701775253720572e-07,
+      "loss": 1.1499,
+      "step": 7704
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.696008882254668e-07,
+      "loss": 0.8372,
+      "step": 7705
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.690245122250884e-07,
+      "loss": 0.8291,
+      "step": 7706
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.68448397432851e-07,
+      "loss": 0.824,
+      "step": 7707
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.678725439106613e-07,
+      "loss": 0.811,
+      "step": 7708
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.672969517203927e-07,
+      "loss": 0.8469,
+      "step": 7709
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.667216209238934e-07,
+      "loss": 0.7224,
+      "step": 7710
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.661465515829814e-07,
+      "loss": 0.8492,
+      "step": 7711
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.655717437594497e-07,
+      "loss": 1.2933,
+      "step": 7712
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.6499719751506e-07,
+      "loss": 1.0846,
+      "step": 7713
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.644229129115482e-07,
+      "loss": 0.7748,
+      "step": 7714
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.638488900106192e-07,
+      "loss": 0.8871,
+      "step": 7715
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.632751288739546e-07,
+      "loss": 0.7457,
+      "step": 7716
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.62701629563203e-07,
+      "loss": 0.7554,
+      "step": 7717
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.621283921399878e-07,
+      "loss": 0.7913,
+      "step": 7718
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.615554166659016e-07,
+      "loss": 0.6993,
+      "step": 7719
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.609827032025136e-07,
+      "loss": 0.8644,
+      "step": 7720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.604102518113588e-07,
+      "loss": 0.7616,
+      "step": 7721
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.598380625539493e-07,
+      "loss": 0.9878,
+      "step": 7722
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.592661354917644e-07,
+      "loss": 1.0197,
+      "step": 7723
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.586944706862609e-07,
+      "loss": 0.872,
+      "step": 7724
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.581230681988612e-07,
+      "loss": 0.8438,
+      "step": 7725
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.575519280909644e-07,
+      "loss": 0.728,
+      "step": 7726
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.56981050423937e-07,
+      "loss": 0.9545,
+      "step": 7727
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.564104352591232e-07,
+      "loss": 0.8749,
+      "step": 7728
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.55840082657833e-07,
+      "loss": 0.9407,
+      "step": 7729
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.552699926813519e-07,
+      "loss": 0.9454,
+      "step": 7730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.547001653909341e-07,
+      "loss": 0.926,
+      "step": 7731
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.54130600847811e-07,
+      "loss": 0.7255,
+      "step": 7732
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.535612991131795e-07,
+      "loss": 0.8992,
+      "step": 7733
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.529922602482122e-07,
+      "loss": 0.6745,
+      "step": 7734
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.524234843140505e-07,
+      "loss": 0.9892,
+      "step": 7735
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.518549713718127e-07,
+      "loss": 0.7968,
+      "step": 7736
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.512867214825823e-07,
+      "loss": 0.8141,
+      "step": 7737
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.507187347074201e-07,
+      "loss": 0.9462,
+      "step": 7738
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.501510111073532e-07,
+      "loss": 0.8778,
+      "step": 7739
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.49583550743387e-07,
+      "loss": 1.0532,
+      "step": 7740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.490163536764925e-07,
+      "loss": 0.8353,
+      "step": 7741
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.484494199676168e-07,
+      "loss": 0.9773,
+      "step": 7742
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.478827496776747e-07,
+      "loss": 0.922,
+      "step": 7743
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.473163428675573e-07,
+      "loss": 0.8361,
+      "step": 7744
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.467501995981234e-07,
+      "loss": 1.0802,
+      "step": 7745
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.461843199302059e-07,
+      "loss": 0.8799,
+      "step": 7746
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.45618703924607e-07,
+      "loss": 1.0208,
+      "step": 7747
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.450533516421045e-07,
+      "loss": 0.7965,
+      "step": 7748
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.444882631434434e-07,
+      "loss": 0.8207,
+      "step": 7749
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.439234384893442e-07,
+      "loss": 0.7703,
+      "step": 7750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.433588777404945e-07,
+      "loss": 0.7776,
+      "step": 7751
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.427945809575598e-07,
+      "loss": 0.7275,
+      "step": 7752
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.422305482011709e-07,
+      "loss": 0.9844,
+      "step": 7753
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.416667795319346e-07,
+      "loss": 0.7748,
+      "step": 7754
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.411032750104272e-07,
+      "loss": 0.8922,
+      "step": 7755
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.405400346971981e-07,
+      "loss": 1.0189,
+      "step": 7756
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.399770586527663e-07,
+      "loss": 0.8838,
+      "step": 7757
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.394143469376241e-07,
+      "loss": 0.9165,
+      "step": 7758
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.388518996122345e-07,
+      "loss": 0.9276,
+      "step": 7759
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.382897167370335e-07,
+      "loss": 0.9046,
+      "step": 7760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.37727798372426e-07,
+      "loss": 0.949,
+      "step": 7761
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.371661445787907e-07,
+      "loss": 1.0182,
+      "step": 7762
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.366047554164775e-07,
+      "loss": 0.9391,
+      "step": 7763
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.360436309458085e-07,
+      "loss": 1.0426,
+      "step": 7764
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.354827712270747e-07,
+      "loss": 1.1461,
+      "step": 7765
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.34922176320541e-07,
+      "loss": 1.0342,
+      "step": 7766
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.343618462864439e-07,
+      "loss": 1.1428,
+      "step": 7767
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.33801781184991e-07,
+      "loss": 0.8704,
+      "step": 7768
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.332419810763601e-07,
+      "loss": 1.0631,
+      "step": 7769
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.326824460207022e-07,
+      "loss": 0.8724,
+      "step": 7770
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.321231760781397e-07,
+      "loss": 1.0705,
+      "step": 7771
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.315641713087662e-07,
+      "loss": 0.9096,
+      "step": 7772
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.310054317726459e-07,
+      "loss": 0.9032,
+      "step": 7773
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.304469575298154e-07,
+      "loss": 0.8894,
+      "step": 7774
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.298887486402834e-07,
+      "loss": 0.8002,
+      "step": 7775
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.293308051640296e-07,
+      "loss": 0.9611,
+      "step": 7776
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.28773127161004e-07,
+      "loss": 0.8668,
+      "step": 7777
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.282157146911295e-07,
+      "loss": 1.0063,
+      "step": 7778
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.276585678142999e-07,
+      "loss": 0.7314,
+      "step": 7779
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.271016865903816e-07,
+      "loss": 1.1008,
+      "step": 7780
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.265450710792102e-07,
+      "loss": 0.8581,
+      "step": 7781
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.259887213405939e-07,
+      "loss": 0.8065,
+      "step": 7782
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.254326374343133e-07,
+      "loss": 0.8406,
+      "step": 7783
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.248768194201196e-07,
+      "loss": 0.7957,
+      "step": 7784
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.243212673577345e-07,
+      "loss": 0.7749,
+      "step": 7785
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.237659813068525e-07,
+      "loss": 0.793,
+      "step": 7786
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.232109613271393e-07,
+      "loss": 1.1189,
+      "step": 7787
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.226562074782322e-07,
+      "loss": 0.9058,
+      "step": 7788
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.221017198197381e-07,
+      "loss": 0.9853,
+      "step": 7789
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.215474984112374e-07,
+      "loss": 0.9841,
+      "step": 7790
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.209935433122811e-07,
+      "loss": 1.0752,
+      "step": 7791
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.204398545823923e-07,
+      "loss": 0.747,
+      "step": 7792
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.198864322810636e-07,
+      "loss": 0.8322,
+      "step": 7793
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.193332764677608e-07,
+      "loss": 0.7585,
+      "step": 7794
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.187803872019202e-07,
+      "loss": 0.7714,
+      "step": 7795
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.18227764542951e-07,
+      "loss": 1.0007,
+      "step": 7796
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.176754085502305e-07,
+      "loss": 0.9184,
+      "step": 7797
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.171233192831102e-07,
+      "loss": 0.855,
+      "step": 7798
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.165714968009121e-07,
+      "loss": 0.8483,
+      "step": 7799
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.160199411629301e-07,
+      "loss": 0.7193,
+      "step": 7800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.154686524284277e-07,
+      "loss": 1.0693,
+      "step": 7801
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.149176306566408e-07,
+      "loss": 1.0563,
+      "step": 7802
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.143668759067777e-07,
+      "loss": 0.9073,
+      "step": 7803
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.13816388238017e-07,
+      "loss": 0.9395,
+      "step": 7804
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.13266167709507e-07,
+      "loss": 0.8265,
+      "step": 7805
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.127162143803702e-07,
+      "loss": 0.8431,
+      "step": 7806
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.121665283096985e-07,
+      "loss": 0.8535,
+      "step": 7807
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.116171095565565e-07,
+      "loss": 0.8402,
+      "step": 7808
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.110679581799778e-07,
+      "loss": 0.7987,
+      "step": 7809
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.105190742389694e-07,
+      "loss": 0.7508,
+      "step": 7810
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.099704577925085e-07,
+      "loss": 0.873,
+      "step": 7811
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.094221088995453e-07,
+      "loss": 0.9124,
+      "step": 7812
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.08874027618998e-07,
+      "loss": 0.8566,
+      "step": 7813
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.083262140097589e-07,
+      "loss": 0.9592,
+      "step": 7814
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.0777866813069e-07,
+      "loss": 0.8614,
+      "step": 7815
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.072313900406263e-07,
+      "loss": 0.9067,
+      "step": 7816
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.06684379798371e-07,
+      "loss": 0.8461,
+      "step": 7817
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.06137637462701e-07,
+      "loss": 0.7451,
+      "step": 7818
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.05591163092364e-07,
+      "loss": 0.7938,
+      "step": 7819
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.050449567460795e-07,
+      "loss": 1.0725,
+      "step": 7820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.044990184825354e-07,
+      "loss": 0.8589,
+      "step": 7821
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.039533483603937e-07,
+      "loss": 1.0993,
+      "step": 7822
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.034079464382876e-07,
+      "loss": 1.1152,
+      "step": 7823
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.028628127748184e-07,
+      "loss": 1.0804,
+      "step": 7824
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.02317947428562e-07,
+      "loss": 0.9036,
+      "step": 7825
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.017733504580639e-07,
+      "loss": 0.983,
+      "step": 7826
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.012290219218415e-07,
+      "loss": 0.9709,
+      "step": 7827
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.006849618783819e-07,
+      "loss": 0.9229,
+      "step": 7828
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.001411703861448e-07,
+      "loss": 1.0064,
+      "step": 7829
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.995976475035607e-07,
+      "loss": 0.8803,
+      "step": 7830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.990543932890312e-07,
+      "loss": 0.9092,
+      "step": 7831
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.985114078009283e-07,
+      "loss": 1.1705,
+      "step": 7832
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.979686910975962e-07,
+      "loss": 1.1209,
+      "step": 7833
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.974262432373494e-07,
+      "loss": 0.8647,
+      "step": 7834
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.968840642784752e-07,
+      "loss": 0.8547,
+      "step": 7835
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.963421542792289e-07,
+      "loss": 0.775,
+      "step": 7836
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.958005132978393e-07,
+      "loss": 0.8283,
+      "step": 7837
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.95259141392506e-07,
+      "loss": 0.9989,
+      "step": 7838
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.947180386214001e-07,
+      "loss": 0.9493,
+      "step": 7839
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.941772050426612e-07,
+      "loss": 0.7996,
+      "step": 7840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.936366407144029e-07,
+      "loss": 1.0084,
+      "step": 7841
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.930963456947086e-07,
+      "loss": 1.0563,
+      "step": 7842
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.925563200416338e-07,
+      "loss": 0.7978,
+      "step": 7843
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.920165638132028e-07,
+      "loss": 0.9246,
+      "step": 7844
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.914770770674131e-07,
+      "loss": 0.8063,
+      "step": 7845
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.909378598622325e-07,
+      "loss": 0.9705,
+      "step": 7846
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.903989122556004e-07,
+      "loss": 0.9802,
+      "step": 7847
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.898602343054255e-07,
+      "loss": 0.7477,
+      "step": 7848
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.893218260695894e-07,
+      "loss": 1.0556,
+      "step": 7849
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.887836876059436e-07,
+      "loss": 0.8678,
+      "step": 7850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.882458189723124e-07,
+      "loss": 0.9698,
+      "step": 7851
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.877082202264878e-07,
+      "loss": 0.8349,
+      "step": 7852
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.871708914262356e-07,
+      "loss": 0.9338,
+      "step": 7853
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.866338326292922e-07,
+      "loss": 0.9724,
+      "step": 7854
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.860970438933647e-07,
+      "loss": 0.7446,
+      "step": 7855
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.855605252761295e-07,
+      "loss": 0.7485,
+      "step": 7856
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.850242768352363e-07,
+      "loss": 1.1415,
+      "step": 7857
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.844882986283053e-07,
+      "loss": 0.8878,
+      "step": 7858
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.839525907129282e-07,
+      "loss": 0.752,
+      "step": 7859
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.834171531466645e-07,
+      "loss": 0.8947,
+      "step": 7860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.828819859870482e-07,
+      "loss": 0.7646,
+      "step": 7861
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.823470892915829e-07,
+      "loss": 1.107,
+      "step": 7862
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.81812463117744e-07,
+      "loss": 1.018,
+      "step": 7863
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.812781075229754e-07,
+      "loss": 0.9532,
+      "step": 7864
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.807440225646947e-07,
+      "loss": 0.9927,
+      "step": 7865
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.802102083002888e-07,
+      "loss": 1.074,
+      "step": 7866
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.796766647871168e-07,
+      "loss": 0.8229,
+      "step": 7867
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.79143392082507e-07,
+      "loss": 0.9011,
+      "step": 7868
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.786103902437593e-07,
+      "loss": 0.8655,
+      "step": 7869
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.780776593281455e-07,
+      "loss": 0.974,
+      "step": 7870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.775451993929082e-07,
+      "loss": 0.8319,
+      "step": 7871
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.770130104952583e-07,
+      "loss": 0.8254,
+      "step": 7872
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.764810926923803e-07,
+      "loss": 1.0475,
+      "step": 7873
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.759494460414289e-07,
+      "loss": 1.0635,
+      "step": 7874
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7541807059953e-07,
+      "loss": 1.1328,
+      "step": 7875
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7488696642377873e-07,
+      "loss": 0.9103,
+      "step": 7876
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7435613357124264e-07,
+      "loss": 0.9521,
+      "step": 7877
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7382557209895985e-07,
+      "loss": 1.0136,
+      "step": 7878
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.732952820639397e-07,
+      "loss": 0.7237,
+      "step": 7879
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727652635231606e-07,
+      "loss": 1.0052,
+      "step": 7880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7223551653357346e-07,
+      "loss": 0.8823,
+      "step": 7881
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.717060411520999e-07,
+      "loss": 0.8817,
+      "step": 7882
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7117683743563225e-07,
+      "loss": 0.8353,
+      "step": 7883
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7064790544103255e-07,
+      "loss": 0.9258,
+      "step": 7884
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7011924522513473e-07,
+      "loss": 0.9198,
+      "step": 7885
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6959085684474336e-07,
+      "loss": 0.9288,
+      "step": 7886
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.690627403566347e-07,
+      "loss": 0.858,
+      "step": 7887
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.685348958175526e-07,
+      "loss": 0.7908,
+      "step": 7888
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.680073232842168e-07,
+      "loss": 0.974,
+      "step": 7889
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.674800228133126e-07,
+      "loss": 0.8138,
+      "step": 7890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6695299446149996e-07,
+      "loss": 1.1245,
+      "step": 7891
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.664262382854057e-07,
+      "loss": 0.9561,
+      "step": 7892
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.65899754341633e-07,
+      "loss": 0.8453,
+      "step": 7893
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6537354268675015e-07,
+      "loss": 0.8212,
+      "step": 7894
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6484760337729984e-07,
+      "loss": 1.1104,
+      "step": 7895
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.643219364697925e-07,
+      "loss": 0.86,
+      "step": 7896
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.637965420207133e-07,
+      "loss": 0.8827,
+      "step": 7897
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6327142008651415e-07,
+      "loss": 0.8113,
+      "step": 7898
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.627465707236205e-07,
+      "loss": 0.8522,
+      "step": 7899
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6222199398842524e-07,
+      "loss": 0.976,
+      "step": 7900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.61697689937297e-07,
+      "loss": 1.1883,
+      "step": 7901
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6117365862657036e-07,
+      "loss": 1.0324,
+      "step": 7902
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6064990011255357e-07,
+      "loss": 0.6996,
+      "step": 7903
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.601264144515224e-07,
+      "loss": 1.0644,
+      "step": 7904
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5960320169972834e-07,
+      "loss": 0.8483,
+      "step": 7905
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.590802619133879e-07,
+      "loss": 1.3283,
+      "step": 7906
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.585575951486931e-07,
+      "loss": 0.9576,
+      "step": 7907
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.580352014618017e-07,
+      "loss": 0.7944,
+      "step": 7908
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.57513080908848e-07,
+      "loss": 0.8426,
+      "step": 7909
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.569912335459316e-07,
+      "loss": 0.7742,
+      "step": 7910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.564696594291263e-07,
+      "loss": 1.0146,
+      "step": 7911
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5594835861447357e-07,
+      "loss": 0.926,
+      "step": 7912
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.554273311579893e-07,
+      "loss": 1.1329,
+      "step": 7913
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5490657711565596e-07,
+      "loss": 0.8905,
+      "step": 7914
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.543860965434299e-07,
+      "loss": 0.9037,
+      "step": 7915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5386588949723496e-07,
+      "loss": 0.7112,
+      "step": 7916
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.533459560329698e-07,
+      "loss": 0.7502,
+      "step": 7917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.528262962064991e-07,
+      "loss": 1.1333,
+      "step": 7918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.523069100736621e-07,
+      "loss": 0.8818,
+      "step": 7919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5178779769026425e-07,
+      "loss": 0.9118,
+      "step": 7920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.512689591120872e-07,
+      "loss": 0.941,
+      "step": 7921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5075039439487774e-07,
+      "loss": 1.0004,
+      "step": 7922
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.502321035943573e-07,
+      "loss": 0.8705,
+      "step": 7923
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4971408676621395e-07,
+      "loss": 0.9889,
+      "step": 7924
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4919634396611153e-07,
+      "loss": 0.9638,
+      "step": 7925
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4867887524967923e-07,
+      "loss": 0.7883,
+      "step": 7926
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4816168067252014e-07,
+      "loss": 1.0417,
+      "step": 7927
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.476447602902053e-07,
+      "loss": 0.792,
+      "step": 7928
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.471281141582798e-07,
+      "loss": 0.8504,
+      "step": 7929
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.46611742332256e-07,
+      "loss": 0.797,
+      "step": 7930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.460956448676188e-07,
+      "loss": 1.1923,
+      "step": 7931
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4557982181982083e-07,
+      "loss": 1.102,
+      "step": 7932
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450642732442904e-07,
+      "loss": 0.7777,
+      "step": 7933
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4454899919642066e-07,
+      "loss": 0.9094,
+      "step": 7934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.440339997315792e-07,
+      "loss": 1.0591,
+      "step": 7935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.435192749051008e-07,
+      "loss": 0.9828,
+      "step": 7936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.430048247722952e-07,
+      "loss": 0.8461,
+      "step": 7937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.424906493884379e-07,
+      "loss": 0.8665,
+      "step": 7938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4197674880877893e-07,
+      "loss": 1.0419,
+      "step": 7939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.414631230885339e-07,
+      "loss": 0.9791,
+      "step": 7940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4094977228289525e-07,
+      "loss": 0.8655,
+      "step": 7941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4043669644702055e-07,
+      "loss": 0.7628,
+      "step": 7942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.399238956360407e-07,
+      "loss": 0.9215,
+      "step": 7943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3941136990505405e-07,
+      "loss": 0.9428,
+      "step": 7944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.388991193091344e-07,
+      "loss": 0.778,
+      "step": 7945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3838714390332115e-07,
+      "loss": 0.9111,
+      "step": 7946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.378754437426269e-07,
+      "loss": 0.7701,
+      "step": 7947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3736401888203204e-07,
+      "loss": 1.0501,
+      "step": 7948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3685286937649225e-07,
+      "loss": 0.8272,
+      "step": 7949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.363419952809277e-07,
+      "loss": 0.8043,
+      "step": 7950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.358313966502336e-07,
+      "loss": 0.8903,
+      "step": 7951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3532107353927166e-07,
+      "loss": 0.7699,
+      "step": 7952
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3481102600287846e-07,
+      "loss": 0.8218,
+      "step": 7953
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.34301254095857e-07,
+      "loss": 0.7829,
+      "step": 7954
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3379175787298264e-07,
+      "loss": 0.899,
+      "step": 7955
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.332825373890007e-07,
+      "loss": 0.9409,
+      "step": 7956
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.3277359269862773e-07,
+      "loss": 0.7571,
+      "step": 7957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.3226492385654857e-07,
+      "loss": 0.8899,
+      "step": 7958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.3175653091741965e-07,
+      "loss": 1.0018,
+      "step": 7959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.3124841393586845e-07,
+      "loss": 0.8903,
+      "step": 7960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.307405729664927e-07,
+      "loss": 0.8082,
+      "step": 7961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.3023300806385815e-07,
+      "loss": 0.7627,
+      "step": 7962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.297257192825037e-07,
+      "loss": 1.1314,
+      "step": 7963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.292187066769372e-07,
+      "loss": 0.7789,
+      "step": 7964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.28711970301638e-07,
+      "loss": 0.8608,
+      "step": 7965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2820551021105325e-07,
+      "loss": 0.8686,
+      "step": 7966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.276993264596031e-07,
+      "loss": 0.9389,
+      "step": 7967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2719341910167663e-07,
+      "loss": 0.9042,
+      "step": 7968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.266877881916341e-07,
+      "loss": 0.7352,
+      "step": 7969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2618243378380475e-07,
+      "loss": 1.2006,
+      "step": 7970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2567735593248933e-07,
+      "loss": 0.9778,
+      "step": 7971
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2517255469195825e-07,
+      "loss": 0.8841,
+      "step": 7972
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2466803011645306e-07,
+      "loss": 0.8116,
+      "step": 7973
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.241637822601834e-07,
+      "loss": 0.912,
+      "step": 7974
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2365981117733156e-07,
+      "loss": 0.8129,
+      "step": 7975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.231561169220496e-07,
+      "loss": 0.8204,
+      "step": 7976
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.226526995484592e-07,
+      "loss": 0.941,
+      "step": 7977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.2214955911065177e-07,
+      "loss": 0.8823,
+      "step": 7978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.216466956626904e-07,
+      "loss": 0.8758,
+      "step": 7979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.211441092586076e-07,
+      "loss": 0.9029,
+      "step": 7980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.206417999524067e-07,
+      "loss": 0.9425,
+      "step": 7981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.201397677980596e-07,
+      "loss": 0.8014,
+      "step": 7982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.196380128495105e-07,
+      "loss": 0.8771,
+      "step": 7983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1913653516067295e-07,
+      "loss": 0.8513,
+      "step": 7984
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1863533478543116e-07,
+      "loss": 1.0987,
+      "step": 7985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1813441177763767e-07,
+      "loss": 0.8571,
+      "step": 7986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1763376619111717e-07,
+      "loss": 0.9496,
+      "step": 7987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1713339807966453e-07,
+      "loss": 0.7818,
+      "step": 7988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.166333074970445e-07,
+      "loss": 1.0026,
+      "step": 7989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.161334944969908e-07,
+      "loss": 0.8878,
+      "step": 7990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1563395913320836e-07,
+      "loss": 0.9175,
+      "step": 7991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1513470145937296e-07,
+      "loss": 1.0157,
+      "step": 7992
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1463572152913e-07,
+      "loss": 0.733,
+      "step": 7993
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1413701939609327e-07,
+      "loss": 0.6826,
+      "step": 7994
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.136385951138496e-07,
+      "loss": 0.9611,
+      "step": 7995
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1314044873595447e-07,
+      "loss": 0.7076,
+      "step": 7996
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1264258031593403e-07,
+      "loss": 1.1905,
+      "step": 7997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.121449899072829e-07,
+      "loss": 0.8855,
+      "step": 7998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1164767756346786e-07,
+      "loss": 1.0137,
+      "step": 7999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.111506433379256e-07,
+      "loss": 1.059,
+      "step": 8000
+    },
+    {
+      "epoch": 1.66,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0341,
+      "eval_samples_per_second": 352.2,
+      "eval_steps_per_second": 44.033,
+      "step": 8000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.106538872840624e-07,
+      "loss": 0.9999,
+      "step": 8001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1015740945525346e-07,
+      "loss": 0.9226,
+      "step": 8002
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.0966120990484625e-07,
+      "loss": 0.9423,
+      "step": 8003
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.09165288686157e-07,
+      "loss": 0.8778,
+      "step": 8004
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.0866964585247344e-07,
+      "loss": 1.1924,
+      "step": 8005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0817428145705104e-07,
+      "loss": 0.9885,
+      "step": 8006
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0767919555311686e-07,
+      "loss": 0.8103,
+      "step": 8007
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0718438819386795e-07,
+      "loss": 1.0332,
+      "step": 8008
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.066898594324725e-07,
+      "loss": 0.9772,
+      "step": 8009
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.06195609322066e-07,
+      "loss": 1.1248,
+      "step": 8010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.05701637915756e-07,
+      "loss": 0.9158,
+      "step": 8011
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0520794526661995e-07,
+      "loss": 0.8789,
+      "step": 8012
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0471453142770553e-07,
+      "loss": 0.8423,
+      "step": 8013
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.042213964520289e-07,
+      "loss": 1.0217,
+      "step": 8014
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.037285403925781e-07,
+      "loss": 1.0847,
+      "step": 8015
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0323596330231037e-07,
+      "loss": 1.0449,
+      "step": 8016
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0274366523415376e-07,
+      "loss": 0.7196,
+      "step": 8017
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.022516462410042e-07,
+      "loss": 0.9101,
+      "step": 8018
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.017599063757301e-07,
+      "loss": 0.9085,
+      "step": 8019
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0126844569116826e-07,
+      "loss": 0.9099,
+      "step": 8020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0077726424012775e-07,
+      "loss": 0.9464,
+      "step": 8021
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.0028636207538295e-07,
+      "loss": 0.9051,
+      "step": 8022
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9979573924968437e-07,
+      "loss": 1.0067,
+      "step": 8023
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9930539581574754e-07,
+      "loss": 0.9715,
+      "step": 8024
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9881533182626097e-07,
+      "loss": 1.0535,
+      "step": 8025
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9832554733387983e-07,
+      "loss": 1.0969,
+      "step": 8026
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9783604239123473e-07,
+      "loss": 0.6979,
+      "step": 8027
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.973468170509202e-07,
+      "loss": 0.8041,
+      "step": 8028
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9685787136550533e-07,
+      "loss": 0.9548,
+      "step": 8029
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9636920538752495e-07,
+      "loss": 1.0363,
+      "step": 8030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9588081916948883e-07,
+      "loss": 0.9346,
+      "step": 8031
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9539271276387256e-07,
+      "loss": 0.9631,
+      "step": 8032
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9490488622312395e-07,
+      "loss": 0.7549,
+      "step": 8033
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9441733959965797e-07,
+      "loss": 0.7081,
+      "step": 8034
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9393007294586446e-07,
+      "loss": 1.0971,
+      "step": 8035
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9344308631409785e-07,
+      "loss": 0.885,
+      "step": 8036
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.929563797566863e-07,
+      "loss": 0.9741,
+      "step": 8037
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9246995332592463e-07,
+      "loss": 1.0041,
+      "step": 8038
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9198380707408133e-07,
+      "loss": 0.7751,
+      "step": 8039
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.914979410533913e-07,
+      "loss": 0.7968,
+      "step": 8040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9101235531606216e-07,
+      "loss": 0.7517,
+      "step": 8041
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.905270499142678e-07,
+      "loss": 1.0712,
+      "step": 8042
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9004202490015716e-07,
+      "loss": 0.7683,
+      "step": 8043
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.895572803258439e-07,
+      "loss": 0.7281,
+      "step": 8044
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.890728162434154e-07,
+      "loss": 0.8172,
+      "step": 8045
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.885886327049253e-07,
+      "loss": 0.8298,
+      "step": 8046
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881047297624013e-07,
+      "loss": 0.8632,
+      "step": 8047
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8762110746783717e-07,
+      "loss": 1.0661,
+      "step": 8048
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.871377658731994e-07,
+      "loss": 0.8849,
+      "step": 8049
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8665470503042066e-07,
+      "loss": 0.7799,
+      "step": 8050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8617192499140883e-07,
+      "loss": 0.6922,
+      "step": 8051
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8568942580803644e-07,
+      "loss": 1.1892,
+      "step": 8052
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8520720753214964e-07,
+      "loss": 0.9505,
+      "step": 8053
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8472527021556e-07,
+      "loss": 0.7492,
+      "step": 8054
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8424361391005515e-07,
+      "loss": 0.9206,
+      "step": 8055
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.837622386673866e-07,
+      "loss": 0.8665,
+      "step": 8056
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.832811445392793e-07,
+      "loss": 0.9514,
+      "step": 8057
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.82800331577425e-07,
+      "loss": 0.8602,
+      "step": 8058
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8231979983348996e-07,
+      "loss": 1.0227,
+      "step": 8059
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.818395493591046e-07,
+      "loss": 0.9757,
+      "step": 8060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8135958020587356e-07,
+      "loss": 0.7419,
+      "step": 8061
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.808798924253677e-07,
+      "loss": 0.9683,
+      "step": 8062
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8040048606913134e-07,
+      "loss": 0.7708,
+      "step": 8063
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7992136118867573e-07,
+      "loss": 0.874,
+      "step": 8064
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.794425178354832e-07,
+      "loss": 0.8225,
+      "step": 8065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7896395606100374e-07,
+      "loss": 0.9948,
+      "step": 8066
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7848567591666184e-07,
+      "loss": 0.7983,
+      "step": 8067
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.78007677453846e-07,
+      "loss": 0.8968,
+      "step": 8068
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7752996072391886e-07,
+      "loss": 0.7875,
+      "step": 8069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7705252577820904e-07,
+      "loss": 0.7836,
+      "step": 8070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7657537266801946e-07,
+      "loss": 1.0832,
+      "step": 8071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7609850144461845e-07,
+      "loss": 0.8907,
+      "step": 8072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7562191215924667e-07,
+      "loss": 0.67,
+      "step": 8073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7514560486311155e-07,
+      "loss": 0.888,
+      "step": 8074
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.746695796073958e-07,
+      "loss": 0.7664,
+      "step": 8075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.741938364432451e-07,
+      "loss": 1.0003,
+      "step": 8076
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7371837542178034e-07,
+      "loss": 0.6665,
+      "step": 8077
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7324319659408755e-07,
+      "loss": 0.9255,
+      "step": 8078
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.72768300011227e-07,
+      "loss": 0.8078,
+      "step": 8079
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.722936857242244e-07,
+      "loss": 0.7932,
+      "step": 8080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.718193537840784e-07,
+      "loss": 0.9053,
+      "step": 8081
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7134530424175386e-07,
+      "loss": 1.0169,
+      "step": 8082
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.708715371481903e-07,
+      "loss": 0.8263,
+      "step": 8083
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.703980525542918e-07,
+      "loss": 0.7992,
+      "step": 8084
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.699248505109356e-07,
+      "loss": 1.0458,
+      "step": 8085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6945193106896535e-07,
+      "loss": 0.8447,
+      "step": 8086
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.689792942791984e-07,
+      "loss": 1.0788,
+      "step": 8087
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6850694019241817e-07,
+      "loss": 0.8415,
+      "step": 8088
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.68034868859379e-07,
+      "loss": 0.9973,
+      "step": 8089
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6756308033080633e-07,
+      "loss": 0.7332,
+      "step": 8090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.670915746573923e-07,
+      "loss": 0.7902,
+      "step": 8091
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6662035188980037e-07,
+      "loss": 0.7117,
+      "step": 8092
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.66149412078664e-07,
+      "loss": 0.9244,
+      "step": 8093
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.656787552745858e-07,
+      "loss": 0.918,
+      "step": 8094
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.65208381528137e-07,
+      "loss": 0.9612,
+      "step": 8095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.647382908898593e-07,
+      "loss": 0.9057,
+      "step": 8096
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.642684834102642e-07,
+      "loss": 1.0122,
+      "step": 8097
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6379895913983317e-07,
+      "loss": 0.789,
+      "step": 8098
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.633297181290153e-07,
+      "loss": 0.8546,
+      "step": 8099
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.628607604282311e-07,
+      "loss": 0.8189,
+      "step": 8100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.623920860878701e-07,
+      "loss": 0.8409,
+      "step": 8101
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.619236951582916e-07,
+      "loss": 1.1355,
+      "step": 8102
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.6145558768982336e-07,
+      "loss": 0.9093,
+      "step": 8103
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.6098776373276375e-07,
+      "loss": 0.8066,
+      "step": 8104
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.6052022333738065e-07,
+      "loss": 0.8624,
+      "step": 8105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.600529665539122e-07,
+      "loss": 0.9899,
+      "step": 8106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5958599343256313e-07,
+      "loss": 1.1209,
+      "step": 8107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.591193040235108e-07,
+      "loss": 0.751,
+      "step": 8108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5865289837690063e-07,
+      "loss": 0.9349,
+      "step": 8109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.581867765428486e-07,
+      "loss": 0.754,
+      "step": 8110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.577209385714386e-07,
+      "loss": 1.0161,
+      "step": 8111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5725538451272497e-07,
+      "loss": 0.8187,
+      "step": 8112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.567901144167317e-07,
+      "loss": 0.948,
+      "step": 8113
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5632512833345275e-07,
+      "loss": 0.8115,
+      "step": 8114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.558604263128495e-07,
+      "loss": 0.8064,
+      "step": 8115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5539600840485474e-07,
+      "loss": 0.9677,
+      "step": 8116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5493187465937014e-07,
+      "loss": 0.977,
+      "step": 8117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.544680251262675e-07,
+      "loss": 0.9543,
+      "step": 8118
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.54004459855386e-07,
+      "loss": 0.6864,
+      "step": 8119
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.535411788965367e-07,
+      "loss": 1.0175,
+      "step": 8120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5307818229949884e-07,
+      "loss": 0.9304,
+      "step": 8121
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5261547011402193e-07,
+      "loss": 0.9371,
+      "step": 8122
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5215304238982326e-07,
+      "loss": 0.8502,
+      "step": 8123
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5169089917659136e-07,
+      "loss": 0.8415,
+      "step": 8124
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5122904052398317e-07,
+      "loss": 1.061,
+      "step": 8125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.507674664816264e-07,
+      "loss": 0.7724,
+      "step": 8126
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5030617709911563e-07,
+      "loss": 1.2068,
+      "step": 8127
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.498451724260169e-07,
+      "loss": 0.8585,
+      "step": 8128
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.493844525118652e-07,
+      "loss": 0.8791,
+      "step": 8129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4892401740616574e-07,
+      "loss": 1.2181,
+      "step": 8130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4846386715839086e-07,
+      "loss": 1.2071,
+      "step": 8131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.480040018179843e-07,
+      "loss": 1.1786,
+      "step": 8132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.475444214343583e-07,
+      "loss": 0.8151,
+      "step": 8133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.470851260568959e-07,
+      "loss": 0.9152,
+      "step": 8134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.466261157349464e-07,
+      "loss": 1.0187,
+      "step": 8135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.461673905178316e-07,
+      "loss": 0.8563,
+      "step": 8136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.457089504548412e-07,
+      "loss": 0.976,
+      "step": 8137
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4525079559523533e-07,
+      "loss": 0.7005,
+      "step": 8138
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.447929259882416e-07,
+      "loss": 0.9026,
+      "step": 8139
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4433534168305813e-07,
+      "loss": 0.9514,
+      "step": 8140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4387804272885313e-07,
+      "loss": 1.1044,
+      "step": 8141
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4342102917476303e-07,
+      "loss": 0.7633,
+      "step": 8142
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4296430106989354e-07,
+      "loss": 0.9279,
+      "step": 8143
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.425078584633203e-07,
+      "loss": 0.7961,
+      "step": 8144
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.420517014040879e-07,
+      "loss": 0.761,
+      "step": 8145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.415958299412111e-07,
+      "loss": 0.8186,
+      "step": 8146
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4114024412367193e-07,
+      "loss": 0.9123,
+      "step": 8147
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4068494400042404e-07,
+      "loss": 0.724,
+      "step": 8148
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4022992962038925e-07,
+      "loss": 1.0796,
+      "step": 8149
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.397752010324591e-07,
+      "loss": 0.781,
+      "step": 8150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3932075828549314e-07,
+      "loss": 0.8432,
+      "step": 8151
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3886660142832203e-07,
+      "loss": 0.7944,
+      "step": 8152
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3841273050974477e-07,
+      "loss": 0.7252,
+      "step": 8153
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.379591455785299e-07,
+      "loss": 0.9669,
+      "step": 8154
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3750584668341453e-07,
+      "loss": 0.8581,
+      "step": 8155
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.37052833873106e-07,
+      "loss": 0.7845,
+      "step": 8156
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.366001071962804e-07,
+      "loss": 0.8936,
+      "step": 8157
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.361476667015838e-07,
+      "loss": 0.7537,
+      "step": 8158
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.356955124376296e-07,
+      "loss": 1.0799,
+      "step": 8159
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.352436444530027e-07,
+      "loss": 0.7489,
+      "step": 8160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.347920627962559e-07,
+      "loss": 1.2789,
+      "step": 8161
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3434076751591225e-07,
+      "loss": 0.8822,
+      "step": 8162
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3388975866046246e-07,
+      "loss": 0.8805,
+      "step": 8163
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.33439036278368e-07,
+      "loss": 0.8482,
+      "step": 8164
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.329886004180586e-07,
+      "loss": 0.891,
+      "step": 8165
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3253845112793467e-07,
+      "loss": 0.7625,
+      "step": 8166
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.320885884563627e-07,
+      "loss": 0.9745,
+      "step": 8167
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3163901245168195e-07,
+      "loss": 0.87,
+      "step": 8168
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3118972316219875e-07,
+      "loss": 0.9544,
+      "step": 8169
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3074072063618985e-07,
+      "loss": 1.0106,
+      "step": 8170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302920049218997e-07,
+      "loss": 0.7971,
+      "step": 8171
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.29843576067543e-07,
+      "loss": 0.7658,
+      "step": 8172
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2939543412130357e-07,
+      "loss": 1.1107,
+      "step": 8173
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2894757913133457e-07,
+      "loss": 0.8406,
+      "step": 8174
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.285000111457571e-07,
+      "loss": 1.0326,
+      "step": 8175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.280527302126628e-07,
+      "loss": 0.8816,
+      "step": 8176
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2760573638011217e-07,
+      "loss": 0.9683,
+      "step": 8177
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2715902969613477e-07,
+      "loss": 0.8516,
+      "step": 8178
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2671261020872854e-07,
+      "loss": 0.7938,
+      "step": 8179
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.262664779658615e-07,
+      "loss": 0.9976,
+      "step": 8180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.258206330154706e-07,
+      "loss": 1.0757,
+      "step": 8181
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2537507540546217e-07,
+      "loss": 0.9211,
+      "step": 8182
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.24929805183711e-07,
+      "loss": 0.825,
+      "step": 8183
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.24484822398061e-07,
+      "loss": 0.9945,
+      "step": 8184
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2404012709632566e-07,
+      "loss": 0.8818,
+      "step": 8185
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2359571932628884e-07,
+      "loss": 0.9889,
+      "step": 8186
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2315159913570025e-07,
+      "loss": 0.8909,
+      "step": 8187
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.227077665722811e-07,
+      "loss": 0.9768,
+      "step": 8188
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.222642216837215e-07,
+      "loss": 1.0625,
+      "step": 8189
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.218209645176811e-07,
+      "loss": 0.9741,
+      "step": 8190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.21377995121786e-07,
+      "loss": 0.902,
+      "step": 8191
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.209353135436343e-07,
+      "loss": 0.9376,
+      "step": 8192
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2049291983079176e-07,
+      "loss": 0.846,
+      "step": 8193
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.200508140307945e-07,
+      "loss": 1.2197,
+      "step": 8194
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.196089961911455e-07,
+      "loss": 0.9164,
+      "step": 8195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.191674663593187e-07,
+      "loss": 1.0211,
+      "step": 8196
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1872622458275624e-07,
+      "loss": 0.9487,
+      "step": 8197
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.182852709088702e-07,
+      "loss": 1.0808,
+      "step": 8198
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.178446053850399e-07,
+      "loss": 0.89,
+      "step": 8199
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.174042280586152e-07,
+      "loss": 0.7864,
+      "step": 8200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.16964138976915e-07,
+      "loss": 0.7512,
+      "step": 8201
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1652433818722704e-07,
+      "loss": 0.954,
+      "step": 8202
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1608482573680684e-07,
+      "loss": 0.8508,
+      "step": 8203
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.156456016728807e-07,
+      "loss": 1.0839,
+      "step": 8204
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1520666604264314e-07,
+      "loss": 0.9729,
+      "step": 8205
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1476801889325814e-07,
+      "loss": 0.9791,
+      "step": 8206
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1432966027185763e-07,
+      "loss": 1.0023,
+      "step": 8207
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1389159022554323e-07,
+      "loss": 0.9817,
+      "step": 8208
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1345380880138575e-07,
+      "loss": 0.9138,
+      "step": 8209
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1301631604642546e-07,
+      "loss": 0.9273,
+      "step": 8210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1257911200766974e-07,
+      "loss": 0.7316,
+      "step": 8211
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.121421967320964e-07,
+      "loss": 0.8015,
+      "step": 8212
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.117055702666526e-07,
+      "loss": 0.7432,
+      "step": 8213
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.11269232658254e-07,
+      "loss": 0.7637,
+      "step": 8214
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1083318395378355e-07,
+      "loss": 0.8206,
+      "step": 8215
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1039742420009576e-07,
+      "loss": 0.7094,
+      "step": 8216
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0996195344401313e-07,
+      "loss": 0.7499,
+      "step": 8217
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0952677173232693e-07,
+      "loss": 0.8856,
+      "step": 8218
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.09091879111796e-07,
+      "loss": 0.8248,
+      "step": 8219
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.086572756291518e-07,
+      "loss": 0.8374,
+      "step": 8220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.082229613310905e-07,
+      "loss": 1.1268,
+      "step": 8221
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0778893626428083e-07,
+      "loss": 0.9044,
+      "step": 8222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.073552004753565e-07,
+      "loss": 1.2934,
+      "step": 8223
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0692175401092483e-07,
+      "loss": 0.7397,
+      "step": 8224
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0648859691755826e-07,
+      "loss": 0.9173,
+      "step": 8225
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.060557292417999e-07,
+      "loss": 0.9042,
+      "step": 8226
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0562315103016025e-07,
+      "loss": 0.9114,
+      "step": 8227
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0519086232912206e-07,
+      "loss": 0.9121,
+      "step": 8228
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.047588631851326e-07,
+      "loss": 0.8754,
+      "step": 8229
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0432715364461163e-07,
+      "loss": 0.9912,
+      "step": 8230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0389573375394476e-07,
+      "loss": 0.7548,
+      "step": 8231
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.034646035594896e-07,
+      "loss": 0.9581,
+      "step": 8232
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0303376310757003e-07,
+      "loss": 0.9557,
+      "step": 8233
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.026032124444807e-07,
+      "loss": 0.8762,
+      "step": 8234
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.021729516164824e-07,
+      "loss": 0.8008,
+      "step": 8235
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.017429806698093e-07,
+      "loss": 0.8546,
+      "step": 8236
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.013132996506599e-07,
+      "loss": 0.8714,
+      "step": 8237
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.008839086052042e-07,
+      "loss": 0.8654,
+      "step": 8238
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.00454807579579e-07,
+      "loss": 1.0182,
+      "step": 8239
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0002599661989307e-07,
+      "loss": 0.8565,
+      "step": 8240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.995974757722205e-07,
+      "loss": 1.0045,
+      "step": 8241
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.991692450826072e-07,
+      "loss": 0.828,
+      "step": 8242
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9874130459706474e-07,
+      "loss": 1.1619,
+      "step": 8243
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.983136543615772e-07,
+      "loss": 0.9915,
+      "step": 8244
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9788629442209423e-07,
+      "loss": 0.7929,
+      "step": 8245
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.974592248245368e-07,
+      "loss": 0.9426,
+      "step": 8246
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.970324456147915e-07,
+      "loss": 0.948,
+      "step": 8247
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9660595683871826e-07,
+      "loss": 0.8337,
+      "step": 8248
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.961797585421415e-07,
+      "loss": 0.9395,
+      "step": 8249
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9575385077085724e-07,
+      "loss": 0.8488,
+      "step": 8250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9532823357062787e-07,
+      "loss": 1.0389,
+      "step": 8251
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.949029069871878e-07,
+      "loss": 0.8083,
+      "step": 8252
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.944778710662365e-07,
+      "loss": 0.9471,
+      "step": 8253
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.940531258534457e-07,
+      "loss": 1.0087,
+      "step": 8254
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9362867139445224e-07,
+      "loss": 0.876,
+      "step": 8255
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.932045077348661e-07,
+      "loss": 0.9568,
+      "step": 8256
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.92780634920262e-07,
+      "loss": 0.8679,
+      "step": 8257
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9235705299618587e-07,
+      "loss": 0.8319,
+      "step": 8258
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9193376200815003e-07,
+      "loss": 0.8398,
+      "step": 8259
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.915107620016394e-07,
+      "loss": 1.2952,
+      "step": 8260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.910880530221036e-07,
+      "loss": 0.9001,
+      "step": 8261
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.90665635114964e-07,
+      "loss": 1.0217,
+      "step": 8262
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.90243508325607e-07,
+      "loss": 0.8097,
+      "step": 8263
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.898216726993932e-07,
+      "loss": 1.0769,
+      "step": 8264
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.894001282816464e-07,
+      "loss": 0.6935,
+      "step": 8265
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8897887511766307e-07,
+      "loss": 0.7199,
+      "step": 8266
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8855791325270496e-07,
+      "loss": 1.0554,
+      "step": 8267
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.881372427320069e-07,
+      "loss": 0.8693,
+      "step": 8268
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8771686360076774e-07,
+      "loss": 1.0646,
+      "step": 8269
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.872967759041587e-07,
+      "loss": 0.827,
+      "step": 8270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.868769796873166e-07,
+      "loss": 0.8817,
+      "step": 8271
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8645747499535046e-07,
+      "loss": 1.0038,
+      "step": 8272
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8603826187333414e-07,
+      "loss": 0.8198,
+      "step": 8273
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.85619340366314e-07,
+      "loss": 1.0184,
+      "step": 8274
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.852007105193006e-07,
+      "loss": 1.0134,
+      "step": 8275
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.847823723772781e-07,
+      "loss": 0.7844,
+      "step": 8276
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.843643259851958e-07,
+      "loss": 0.9915,
+      "step": 8277
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8394657138797307e-07,
+      "loss": 0.9971,
+      "step": 8278
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8352910863049643e-07,
+      "loss": 1.0649,
+      "step": 8279
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.831119377576246e-07,
+      "loss": 0.8886,
+      "step": 8280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8269505881418033e-07,
+      "loss": 0.9326,
+      "step": 8281
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.822784718449586e-07,
+      "loss": 0.831,
+      "step": 8282
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.818621768947202e-07,
+      "loss": 0.9506,
+      "step": 8283
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8144617400819797e-07,
+      "loss": 0.8313,
+      "step": 8284
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8103046323008984e-07,
+      "loss": 1.0113,
+      "step": 8285
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.806150446050645e-07,
+      "loss": 0.9598,
+      "step": 8286
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.801999181777581e-07,
+      "loss": 0.8973,
+      "step": 8287
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7978508399277746e-07,
+      "loss": 0.8961,
+      "step": 8288
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.793705420946948e-07,
+      "loss": 0.7348,
+      "step": 8289
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.789562925280532e-07,
+      "loss": 0.8287,
+      "step": 8290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7854233533736375e-07,
+      "loss": 0.9395,
+      "step": 8291
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.781286705671068e-07,
+      "loss": 0.7509,
+      "step": 8292
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.777152982617295e-07,
+      "loss": 0.8911,
+      "step": 8293
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.77302218465649e-07,
+      "loss": 0.9557,
+      "step": 8294
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.768894312232507e-07,
+      "loss": 0.8702,
+      "step": 8295
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7647693657888963e-07,
+      "loss": 0.7279,
+      "step": 8296
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.760647345768865e-07,
+      "loss": 0.9198,
+      "step": 8297
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.756528252615334e-07,
+      "loss": 0.9519,
+      "step": 8298
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7524120867708967e-07,
+      "loss": 0.9801,
+      "step": 8299
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.748298848677845e-07,
+      "loss": 0.853,
+      "step": 8300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7441885387781276e-07,
+      "loss": 0.8844,
+      "step": 8301
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.740081157513411e-07,
+      "loss": 1.2271,
+      "step": 8302
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.735976705325026e-07,
+      "loss": 0.9797,
+      "step": 8303
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7318751826540033e-07,
+      "loss": 0.7968,
+      "step": 8304
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7277765899410435e-07,
+      "loss": 0.7908,
+      "step": 8305
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.723680927626544e-07,
+      "loss": 0.7797,
+      "step": 8306
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7195881961505795e-07,
+      "loss": 0.8094,
+      "step": 8307
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7154983959529223e-07,
+      "loss": 1.1067,
+      "step": 8308
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.71141152747301e-07,
+      "loss": 1.3726,
+      "step": 8309
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7073275911499853e-07,
+      "loss": 0.9544,
+      "step": 8310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.70324658742266e-07,
+      "loss": 0.7113,
+      "step": 8311
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6991685167295464e-07,
+      "loss": 1.203,
+      "step": 8312
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6950933795088216e-07,
+      "loss": 0.9046,
+      "step": 8313
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6910211761983684e-07,
+      "loss": 0.858,
+      "step": 8314
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6869519072357365e-07,
+      "loss": 1.0448,
+      "step": 8315
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6828855730581826e-07,
+      "loss": 0.7881,
+      "step": 8316
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6788221741026177e-07,
+      "loss": 0.7479,
+      "step": 8317
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.674761710805659e-07,
+      "loss": 0.7672,
+      "step": 8318
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.670704183603604e-07,
+      "loss": 0.8514,
+      "step": 8319
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.666649592932444e-07,
+      "loss": 0.8064,
+      "step": 8320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.662597939227824e-07,
+      "loss": 0.8233,
+      "step": 8321
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6585492229251095e-07,
+      "loss": 1.1701,
+      "step": 8322
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6545034444593285e-07,
+      "loss": 0.8552,
+      "step": 8323
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.650460604265204e-07,
+      "loss": 0.8876,
+      "step": 8324
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6464207027771337e-07,
+      "loss": 0.9872,
+      "step": 8325
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.642383740429205e-07,
+      "loss": 1.0416,
+      "step": 8326
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.638349717655194e-07,
+      "loss": 0.9748,
+      "step": 8327
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.634318634888554e-07,
+      "loss": 0.7033,
+      "step": 8328
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6302904925624217e-07,
+      "loss": 0.903,
+      "step": 8329
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.626265291109624e-07,
+      "loss": 0.9092,
+      "step": 8330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6222430309626653e-07,
+      "loss": 0.8734,
+      "step": 8331
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.618223712553747e-07,
+      "loss": 1.0344,
+      "step": 8332
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.614207336314729e-07,
+      "loss": 0.9636,
+      "step": 8333
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.610193902677177e-07,
+      "loss": 0.8611,
+      "step": 8334
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6061834120723395e-07,
+      "loss": 0.8371,
+      "step": 8335
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6021758649311443e-07,
+      "loss": 0.903,
+      "step": 8336
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.598171261684191e-07,
+      "loss": 1.1283,
+      "step": 8337
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.594169602761779e-07,
+      "loss": 0.765,
+      "step": 8338
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.59017088859389e-07,
+      "loss": 1.0879,
+      "step": 8339
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.586175119610188e-07,
+      "loss": 0.8187,
+      "step": 8340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5821822962400057e-07,
+      "loss": 1.0289,
+      "step": 8341
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5781924189123797e-07,
+      "loss": 0.951,
+      "step": 8342
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.574205488056024e-07,
+      "loss": 0.9965,
+      "step": 8343
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.570221504099336e-07,
+      "loss": 1.0664,
+      "step": 8344
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.566240467470383e-07,
+      "loss": 0.8509,
+      "step": 8345
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5622623785969336e-07,
+      "loss": 1.0041,
+      "step": 8346
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.558287237906438e-07,
+      "loss": 1.0232,
+      "step": 8347
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5543150458260224e-07,
+      "loss": 1.1568,
+      "step": 8348
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.550345802782491e-07,
+      "loss": 0.7803,
+      "step": 8349
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.546379509202347e-07,
+      "loss": 0.9046,
+      "step": 8350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5424161655117616e-07,
+      "loss": 0.748,
+      "step": 8351
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5384557721366084e-07,
+      "loss": 0.867,
+      "step": 8352
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.534498329502414e-07,
+      "loss": 1.1901,
+      "step": 8353
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5305438380344206e-07,
+      "loss": 0.7008,
+      "step": 8354
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5265922981575286e-07,
+      "loss": 0.9918,
+      "step": 8355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5226437102963386e-07,
+      "loss": 0.7941,
+      "step": 8356
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5186980748751143e-07,
+      "loss": 1.0346,
+      "step": 8357
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5147553923178235e-07,
+      "loss": 0.8304,
+      "step": 8358
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.510815663048104e-07,
+      "loss": 0.7131,
+      "step": 8359
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5068788874892837e-07,
+      "loss": 0.9564,
+      "step": 8360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5029450660643603e-07,
+      "loss": 0.8812,
+      "step": 8361
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.499014199196027e-07,
+      "loss": 1.1033,
+      "step": 8362
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4950862873066583e-07,
+      "loss": 0.877,
+      "step": 8363
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.491161330818307e-07,
+      "loss": 1.1383,
+      "step": 8364
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.487239330152705e-07,
+      "loss": 0.8328,
+      "step": 8365
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.483320285731273e-07,
+      "loss": 0.6989,
+      "step": 8366
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4794041979751103e-07,
+      "loss": 0.859,
+      "step": 8367
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.47549106730501e-07,
+      "loss": 0.778,
+      "step": 8368
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.471580894141426e-07,
+      "loss": 0.8627,
+      "step": 8369
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4676736789045097e-07,
+      "loss": 0.8972,
+      "step": 8370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.463769422014094e-07,
+      "loss": 0.8376,
+      "step": 8371
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4598681238896904e-07,
+      "loss": 1.051,
+      "step": 8372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.45596978495049e-07,
+      "loss": 0.7442,
+      "step": 8373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.452074405615369e-07,
+      "loss": 0.9254,
+      "step": 8374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.448181986302888e-07,
+      "loss": 0.7524,
+      "step": 8375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.444292527431289e-07,
+      "loss": 0.8102,
+      "step": 8376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4404060294184883e-07,
+      "loss": 1.039,
+      "step": 8377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.436522492682095e-07,
+      "loss": 0.7539,
+      "step": 8378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.432641917639392e-07,
+      "loss": 0.8763,
+      "step": 8379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4287643047073526e-07,
+      "loss": 0.823,
+      "step": 8380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4248896543026134e-07,
+      "loss": 0.7923,
+      "step": 8381
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.421017966841518e-07,
+      "loss": 0.759,
+      "step": 8382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.41714924274007e-07,
+      "loss": 0.7043,
+      "step": 8383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.413283482413974e-07,
+      "loss": 0.649,
+      "step": 8384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.409420686278594e-07,
+      "loss": 0.7221,
+      "step": 8385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4055608547489915e-07,
+      "loss": 0.884,
+      "step": 8386
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.401703988239908e-07,
+      "loss": 0.9378,
+      "step": 8387
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.397850087165766e-07,
+      "loss": 0.8673,
+      "step": 8388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.39399915194066e-07,
+      "loss": 1.0816,
+      "step": 8389
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.390151182978373e-07,
+      "loss": 0.7852,
+      "step": 8390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3863061806923703e-07,
+      "loss": 0.9791,
+      "step": 8391
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3824641454958017e-07,
+      "loss": 0.8425,
+      "step": 8392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3786250778014874e-07,
+      "loss": 1.0891,
+      "step": 8393
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3747889780219367e-07,
+      "loss": 0.7327,
+      "step": 8394
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3709558465693405e-07,
+      "loss": 0.9099,
+      "step": 8395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3671256838555688e-07,
+      "loss": 0.7309,
+      "step": 8396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3632984902921695e-07,
+      "loss": 0.7133,
+      "step": 8397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3594742662903734e-07,
+      "loss": 0.8115,
+      "step": 8398
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3556530122610953e-07,
+      "loss": 0.7641,
+      "step": 8399
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3518347286149332e-07,
+      "loss": 0.8102,
+      "step": 8400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3480194157621493e-07,
+      "loss": 0.7791,
+      "step": 8401
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3442070741127087e-07,
+      "loss": 0.887,
+      "step": 8402
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3403977040762437e-07,
+      "loss": 0.9863,
+      "step": 8403
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.336591306062077e-07,
+      "loss": 0.964,
+      "step": 8404
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3327878804791947e-07,
+      "loss": 0.9146,
+      "step": 8405
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3289874277362832e-07,
+      "loss": 0.9268,
+      "step": 8406
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3251899482416928e-07,
+      "loss": 0.8979,
+      "step": 8407
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3213954424034766e-07,
+      "loss": 0.8608,
+      "step": 8408
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.317603910629339e-07,
+      "loss": 0.7617,
+      "step": 8409
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.313815353326687e-07,
+      "loss": 0.8252,
+      "step": 8410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3100297709025985e-07,
+      "loss": 0.7954,
+      "step": 8411
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3062471637638382e-07,
+      "loss": 0.8687,
+      "step": 8412
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.302467532316841e-07,
+      "loss": 0.8955,
+      "step": 8413
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2986908769677317e-07,
+      "loss": 0.7862,
+      "step": 8414
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2949171981223095e-07,
+      "loss": 0.7831,
+      "step": 8415
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2911464961860663e-07,
+      "loss": 0.9513,
+      "step": 8416
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2873787715641448e-07,
+      "loss": 0.906,
+      "step": 8417
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.283614024661398e-07,
+      "loss": 1.1093,
+      "step": 8418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2798522558823488e-07,
+      "loss": 0.7689,
+      "step": 8419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2760934656311938e-07,
+      "loss": 0.9273,
+      "step": 8420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2723376543118236e-07,
+      "loss": 0.8903,
+      "step": 8421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2685848223277882e-07,
+      "loss": 1.0645,
+      "step": 8422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2648349700823356e-07,
+      "loss": 1.0127,
+      "step": 8423
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2610880979783865e-07,
+      "loss": 0.8048,
+      "step": 8424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2573442064185456e-07,
+      "loss": 0.8486,
+      "step": 8425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.253603295805088e-07,
+      "loss": 1.0231,
+      "step": 8426
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2498653665399749e-07,
+      "loss": 0.7626,
+      "step": 8427
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2461304190248456e-07,
+      "loss": 1.2366,
+      "step": 8428
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2423984536610288e-07,
+      "loss": 0.9989,
+      "step": 8429
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2386694708495136e-07,
+      "loss": 0.821,
+      "step": 8430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2349434709909832e-07,
+      "loss": 0.9894,
+      "step": 8431
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.231220454485794e-07,
+      "loss": 0.8654,
+      "step": 8432
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.227500421733989e-07,
+      "loss": 1.0305,
+      "step": 8433
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2237833731352786e-07,
+      "loss": 0.8464,
+      "step": 8434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2200693090890634e-07,
+      "loss": 1.1446,
+      "step": 8435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2163582299944173e-07,
+      "loss": 0.8687,
+      "step": 8436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2126501362501016e-07,
+      "loss": 0.9822,
+      "step": 8437
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2089450282545443e-07,
+      "loss": 0.9081,
+      "step": 8438
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2052429064058565e-07,
+      "loss": 0.8929,
+      "step": 8439
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.201543771101837e-07,
+      "loss": 1.1023,
+      "step": 8440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.197847622739958e-07,
+      "loss": 0.864,
+      "step": 8441
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1941544617173682e-07,
+      "loss": 0.7626,
+      "step": 8442
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1904642884308935e-07,
+      "loss": 0.9156,
+      "step": 8443
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1867771032770466e-07,
+      "loss": 0.8261,
+      "step": 8444
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.183092906652021e-07,
+      "loss": 0.843,
+      "step": 8445
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1794116989516733e-07,
+      "loss": 0.832,
+      "step": 8446
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1757334805715535e-07,
+      "loss": 0.844,
+      "step": 8447
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.172058251906882e-07,
+      "loss": 0.9562,
+      "step": 8448
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1683860133525734e-07,
+      "loss": 0.9896,
+      "step": 8449
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1647167653031986e-07,
+      "loss": 0.8887,
+      "step": 8450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1610505081530186e-07,
+      "loss": 0.8009,
+      "step": 8451
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.157387242295975e-07,
+      "loss": 0.9511,
+      "step": 8452
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.15372696812569e-07,
+      "loss": 0.7975,
+      "step": 8453
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1500696860354518e-07,
+      "loss": 0.9414,
+      "step": 8454
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1464153964182364e-07,
+      "loss": 0.7216,
+      "step": 8455
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.142764099666703e-07,
+      "loss": 0.8764,
+      "step": 8456
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.139115796173181e-07,
+      "loss": 0.856,
+      "step": 8457
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1354704863296735e-07,
+      "loss": 0.9402,
+      "step": 8458
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1318281705278742e-07,
+      "loss": 1.0123,
+      "step": 8459
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1281888491591493e-07,
+      "loss": 0.6571,
+      "step": 8460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.12455252261455e-07,
+      "loss": 0.8604,
+      "step": 8461
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.120919191284787e-07,
+      "loss": 0.8741,
+      "step": 8462
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1172888555602677e-07,
+      "loss": 0.7081,
+      "step": 8463
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1136615158310736e-07,
+      "loss": 0.892,
+      "step": 8464
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1100371724869595e-07,
+      "loss": 0.8351,
+      "step": 8465
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1064158259173606e-07,
+      "loss": 0.9477,
+      "step": 8466
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1027974765113888e-07,
+      "loss": 1.0376,
+      "step": 8467
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.099182124657837e-07,
+      "loss": 1.1415,
+      "step": 8468
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0955697707451804e-07,
+      "loss": 1.029,
+      "step": 8469
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0919604151615557e-07,
+      "loss": 1.0601,
+      "step": 8470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0883540582947958e-07,
+      "loss": 0.8298,
+      "step": 8471
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0847507005323973e-07,
+      "loss": 1.0209,
+      "step": 8472
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0811503422615475e-07,
+      "loss": 1.042,
+      "step": 8473
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0775529838690998e-07,
+      "loss": 1.0013,
+      "step": 8474
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.073958625741589e-07,
+      "loss": 0.8389,
+      "step": 8475
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.070367268265232e-07,
+      "loss": 0.9519,
+      "step": 8476
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.066778911825924e-07,
+      "loss": 0.9829,
+      "step": 8477
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0631935568092264e-07,
+      "loss": 0.7217,
+      "step": 8478
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0596112036003843e-07,
+      "loss": 0.9597,
+      "step": 8479
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0560318525843235e-07,
+      "loss": 0.7858,
+      "step": 8480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0524555041456529e-07,
+      "loss": 0.7716,
+      "step": 8481
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0488821586686412e-07,
+      "loss": 0.676,
+      "step": 8482
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0453118165372488e-07,
+      "loss": 0.98,
+      "step": 8483
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.041744478135098e-07,
+      "loss": 0.8541,
+      "step": 8484
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0381801438455196e-07,
+      "loss": 0.8107,
+      "step": 8485
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.03461881405148e-07,
+      "loss": 0.9009,
+      "step": 8486
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0310604891356565e-07,
+      "loss": 0.8193,
+      "step": 8487
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.027505169480387e-07,
+      "loss": 1.1523,
+      "step": 8488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0239528554676957e-07,
+      "loss": 0.975,
+      "step": 8489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0204035474792704e-07,
+      "loss": 0.9217,
+      "step": 8490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0168572458964862e-07,
+      "loss": 1.1346,
+      "step": 8491
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.013313951100395e-07,
+      "loss": 0.9395,
+      "step": 8492
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0097736634717256e-07,
+      "loss": 0.8031,
+      "step": 8493
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0062363833908737e-07,
+      "loss": 0.7065,
+      "step": 8494
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0027021112379284e-07,
+      "loss": 0.8936,
+      "step": 8495
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9991708473926396e-07,
+      "loss": 0.97,
+      "step": 8496
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9956425922344535e-07,
+      "loss": 0.8256,
+      "step": 8497
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9921173461424636e-07,
+      "loss": 0.8703,
+      "step": 8498
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.98859510949547e-07,
+      "loss": 0.9512,
+      "step": 8499
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9850758826719306e-07,
+      "loss": 0.7937,
+      "step": 8500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9815596660499924e-07,
+      "loss": 0.8825,
+      "step": 8501
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9780464600074666e-07,
+      "loss": 0.8541,
+      "step": 8502
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9745362649218478e-07,
+      "loss": 0.9076,
+      "step": 8503
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.971029081170308e-07,
+      "loss": 1.0537,
+      "step": 8504
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9675249091296987e-07,
+      "loss": 0.8136,
+      "step": 8505
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9640237491765323e-07,
+      "loss": 0.9571,
+      "step": 8506
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.960525601687011e-07,
+      "loss": 1.1271,
+      "step": 8507
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9570304670370142e-07,
+      "loss": 0.9708,
+      "step": 8508
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9535383456020983e-07,
+      "loss": 1.0329,
+      "step": 8509
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.950049237757483e-07,
+      "loss": 0.9904,
+      "step": 8510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9465631438780717e-07,
+      "loss": 0.8341,
+      "step": 8511
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9430800643384516e-07,
+      "loss": 0.8754,
+      "step": 8512
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9395999995128798e-07,
+      "loss": 1.0296,
+      "step": 8513
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9361229497752807e-07,
+      "loss": 0.7381,
+      "step": 8514
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9326489154992688e-07,
+      "loss": 0.7567,
+      "step": 8515
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9291778970581287e-07,
+      "loss": 0.8849,
+      "step": 8516
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9257098948248254e-07,
+      "loss": 0.8237,
+      "step": 8517
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.922244909171984e-07,
+      "loss": 0.7484,
+      "step": 8518
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9187829404719236e-07,
+      "loss": 0.796,
+      "step": 8519
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9153239890966334e-07,
+      "loss": 0.9524,
+      "step": 8520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9118680554177824e-07,
+      "loss": 0.9991,
+      "step": 8521
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9084151398066972e-07,
+      "loss": 0.742,
+      "step": 8522
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9049652426344012e-07,
+      "loss": 0.8947,
+      "step": 8523
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9015183642715872e-07,
+      "loss": 0.9968,
+      "step": 8524
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.898074505088623e-07,
+      "loss": 1.0003,
+      "step": 8525
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8946336654555419e-07,
+      "loss": 0.8205,
+      "step": 8526
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.891195845742072e-07,
+      "loss": 0.7508,
+      "step": 8527
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8877610463175976e-07,
+      "loss": 1.006,
+      "step": 8528
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8843292675512003e-07,
+      "loss": 0.7673,
+      "step": 8529
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8809005098116118e-07,
+      "loss": 1.0111,
+      "step": 8530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8774747734672571e-07,
+      "loss": 0.8303,
+      "step": 8531
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.874052058886232e-07,
+      "loss": 0.9373,
+      "step": 8532
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.870632366436309e-07,
+      "loss": 1.0064,
+      "step": 8533
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8672156964849273e-07,
+      "loss": 1.1016,
+      "step": 8534
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8638020493992102e-07,
+      "loss": 0.9433,
+      "step": 8535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8603914255459575e-07,
+      "loss": 0.8858,
+      "step": 8536
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8569838252916426e-07,
+      "loss": 0.8296,
+      "step": 8537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8535792490023996e-07,
+      "loss": 0.9773,
+      "step": 8538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.850177697044062e-07,
+      "loss": 0.8625,
+      "step": 8539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.846779169782118e-07,
+      "loss": 0.7268,
+      "step": 8540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8433836675817516e-07,
+      "loss": 0.803,
+      "step": 8541
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8399911908077948e-07,
+      "loss": 0.8632,
+      "step": 8542
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8366017398247758e-07,
+      "loss": 0.8865,
+      "step": 8543
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8332153149968866e-07,
+      "loss": 0.7746,
+      "step": 8544
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8298319166880096e-07,
+      "loss": 1.0519,
+      "step": 8545
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8264515452616803e-07,
+      "loss": 0.7311,
+      "step": 8546
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8230742010811185e-07,
+      "loss": 0.8401,
+      "step": 8547
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8196998845092237e-07,
+      "loss": 1.0609,
+      "step": 8548
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.816328595908573e-07,
+      "loss": 0.82,
+      "step": 8549
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8129603356413958e-07,
+      "loss": 0.9378,
+      "step": 8550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.80959510406962e-07,
+      "loss": 0.9337,
+      "step": 8551
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8062329015548394e-07,
+      "loss": 0.9158,
+      "step": 8552
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8028737284583252e-07,
+      "loss": 1.1609,
+      "step": 8553
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7995175851410084e-07,
+      "loss": 0.8224,
+      "step": 8554
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7961644719635205e-07,
+      "loss": 1.0372,
+      "step": 8555
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7928143892861472e-07,
+      "loss": 0.9045,
+      "step": 8556
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7894673374688563e-07,
+      "loss": 1.0657,
+      "step": 8557
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.786123316871281e-07,
+      "loss": 0.9859,
+      "step": 8558
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.78278232785275e-07,
+      "loss": 0.962,
+      "step": 8559
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7794443707722396e-07,
+      "loss": 0.9386,
+      "step": 8560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.776109445988423e-07,
+      "loss": 0.8462,
+      "step": 8561
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7727775538596202e-07,
+      "loss": 0.7558,
+      "step": 8562
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.769448694743868e-07,
+      "loss": 0.9324,
+      "step": 8563
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.766122868998834e-07,
+      "loss": 0.8704,
+      "step": 8564
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7628000769818852e-07,
+      "loss": 1.0899,
+      "step": 8565
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7594803190500464e-07,
+      "loss": 0.9742,
+      "step": 8566
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7561635955600387e-07,
+      "loss": 1.0721,
+      "step": 8567
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.752849906868237e-07,
+      "loss": 0.7708,
+      "step": 8568
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7495392533306965e-07,
+      "loss": 1.0112,
+      "step": 8569
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7462316353031428e-07,
+      "loss": 0.9187,
+      "step": 8570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.742927053140988e-07,
+      "loss": 0.6863,
+      "step": 8571
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.739625507199305e-07,
+      "loss": 0.8122,
+      "step": 8572
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7363269978328466e-07,
+      "loss": 0.9236,
+      "step": 8573
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.733031525396026e-07,
+      "loss": 0.897,
+      "step": 8574
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7297390902429633e-07,
+      "loss": 1.0524,
+      "step": 8575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7264496927274087e-07,
+      "loss": 0.8717,
+      "step": 8576
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7231633332028262e-07,
+      "loss": 0.824,
+      "step": 8577
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7198800120223136e-07,
+      "loss": 1.096,
+      "step": 8578
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.716599729538685e-07,
+      "loss": 1.1118,
+      "step": 8579
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.713322486104395e-07,
+      "loss": 1.0268,
+      "step": 8580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.710048282071589e-07,
+      "loss": 0.8554,
+      "step": 8581
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7067771177920654e-07,
+      "loss": 0.8209,
+      "step": 8582
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7035089936173298e-07,
+      "loss": 0.7647,
+      "step": 8583
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7002439098985313e-07,
+      "loss": 0.9887,
+      "step": 8584
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6969818669865123e-07,
+      "loss": 0.9583,
+      "step": 8585
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6937228652317593e-07,
+      "loss": 0.6103,
+      "step": 8586
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6904669049844755e-07,
+      "loss": 0.7916,
+      "step": 8587
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6872139865944946e-07,
+      "loss": 1.209,
+      "step": 8588
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.68396411041136e-07,
+      "loss": 1.0049,
+      "step": 8589
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6807172767842494e-07,
+      "loss": 1.0435,
+      "step": 8590
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6774734860620567e-07,
+      "loss": 0.9117,
+      "step": 8591
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6742327385933098e-07,
+      "loss": 0.9153,
+      "step": 8592
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6709950347262403e-07,
+      "loss": 0.96,
+      "step": 8593
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6677603748087234e-07,
+      "loss": 0.8912,
+      "step": 8594
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6645287591883406e-07,
+      "loss": 1.1597,
+      "step": 8595
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.661300188212318e-07,
+      "loss": 0.7291,
+      "step": 8596
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6580746622275743e-07,
+      "loss": 0.987,
+      "step": 8597
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6548521815806728e-07,
+      "loss": 0.7126,
+      "step": 8598
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6516327466178927e-07,
+      "loss": 0.6394,
+      "step": 8599
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.648416357685144e-07,
+      "loss": 0.8386,
+      "step": 8600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.645203015128044e-07,
+      "loss": 0.8251,
+      "step": 8601
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6419927192918427e-07,
+      "loss": 0.7738,
+      "step": 8602
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.638785470521511e-07,
+      "loss": 0.7871,
+      "step": 8603
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6355812691616534e-07,
+      "loss": 0.7759,
+      "step": 8604
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6323801155565677e-07,
+      "loss": 0.9888,
+      "step": 8605
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6291820100502054e-07,
+      "loss": 0.919,
+      "step": 8606
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6259869529862214e-07,
+      "loss": 0.8985,
+      "step": 8607
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6227949447079082e-07,
+      "loss": 1.0473,
+      "step": 8608
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.619605985558258e-07,
+      "loss": 0.7107,
+      "step": 8609
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.616420075879913e-07,
+      "loss": 0.6823,
+      "step": 8610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.613237216015213e-07,
+      "loss": 1.059,
+      "step": 8611
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6100574063061414e-07,
+      "loss": 0.8368,
+      "step": 8612
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.606880647094384e-07,
+      "loss": 0.9022,
+      "step": 8613
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6037069387212656e-07,
+      "loss": 0.9481,
+      "step": 8614
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6005362815278157e-07,
+      "loss": 0.7456,
+      "step": 8615
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5973686758547157e-07,
+      "loss": 0.8968,
+      "step": 8616
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5942041220423263e-07,
+      "loss": 0.7977,
+      "step": 8617
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5910426204306693e-07,
+      "loss": 0.993,
+      "step": 8618
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.587884171359466e-07,
+      "loss": 0.9736,
+      "step": 8619
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5847287751680751e-07,
+      "loss": 0.7948,
+      "step": 8620
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5815764321955484e-07,
+      "loss": 1.028,
+      "step": 8621
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5784271427806052e-07,
+      "loss": 1.0659,
+      "step": 8622
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5752809072616448e-07,
+      "loss": 1.0306,
+      "step": 8623
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5721377259767168e-07,
+      "loss": 0.9231,
+      "step": 8624
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5689975992635608e-07,
+      "loss": 0.8913,
+      "step": 8625
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5658605274595873e-07,
+      "loss": 0.7793,
+      "step": 8626
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.562726510901873e-07,
+      "loss": 0.845,
+      "step": 8627
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5595955499271618e-07,
+      "loss": 0.7636,
+      "step": 8628
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.5564676448718784e-07,
+      "loss": 0.6148,
+      "step": 8629
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.553342796072117e-07,
+      "loss": 0.8556,
+      "step": 8630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5502210038636488e-07,
+      "loss": 0.9265,
+      "step": 8631
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5471022685818958e-07,
+      "loss": 0.9458,
+      "step": 8632
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5439865905619766e-07,
+      "loss": 0.7518,
+      "step": 8633
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5408739701386665e-07,
+      "loss": 1.0239,
+      "step": 8634
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5377644076464215e-07,
+      "loss": 0.9473,
+      "step": 8635
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5346579034193575e-07,
+      "loss": 0.7061,
+      "step": 8636
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.531554457791271e-07,
+      "loss": 0.8725,
+      "step": 8637
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.528454071095622e-07,
+      "loss": 0.8304,
+      "step": 8638
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5253567436655601e-07,
+      "loss": 0.8278,
+      "step": 8639
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5222624758338767e-07,
+      "loss": 0.721,
+      "step": 8640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5191712679330615e-07,
+      "loss": 0.7633,
+      "step": 8641
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5160831202952596e-07,
+      "loss": 0.9015,
+      "step": 8642
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5129980332522953e-07,
+      "loss": 0.8693,
+      "step": 8643
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.50991600713566e-07,
+      "loss": 0.8314,
+      "step": 8644
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5068370422765155e-07,
+      "loss": 0.7139,
+      "step": 8645
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.5037611390056938e-07,
+      "loss": 0.8171,
+      "step": 8646
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.500688297653714e-07,
+      "loss": 0.856,
+      "step": 8647
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4976185185507352e-07,
+      "loss": 1.0599,
+      "step": 8648
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4945518020266137e-07,
+      "loss": 0.9084,
+      "step": 8649
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4914881484108656e-07,
+      "loss": 0.959,
+      "step": 8650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4884275580326878e-07,
+      "loss": 0.7219,
+      "step": 8651
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4853700312209306e-07,
+      "loss": 0.9555,
+      "step": 8652
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482315568304128e-07,
+      "loss": 0.7575,
+      "step": 8653
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792641696104803e-07,
+      "loss": 0.9906,
+      "step": 8654
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.476215835467869e-07,
+      "loss": 0.8127,
+      "step": 8655
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4731705662038253e-07,
+      "loss": 0.9157,
+      "step": 8656
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4701283621455708e-07,
+      "loss": 0.9892,
+      "step": 8657
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4670892236199872e-07,
+      "loss": 0.8128,
+      "step": 8658
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4640531509536336e-07,
+      "loss": 0.9506,
+      "step": 8659
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4610201444727288e-07,
+      "loss": 0.9985,
+      "step": 8660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4579902045031725e-07,
+      "loss": 0.8248,
+      "step": 8661
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4549633313705336e-07,
+      "loss": 0.886,
+      "step": 8662
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.451939525400049e-07,
+      "loss": 0.8659,
+      "step": 8663
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4489187869166255e-07,
+      "loss": 0.9402,
+      "step": 8664
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4459011162448365e-07,
+      "loss": 1.0588,
+      "step": 8665
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4428865137089363e-07,
+      "loss": 0.8897,
+      "step": 8666
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4398749796328492e-07,
+      "loss": 0.9975,
+      "step": 8667
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4368665143401493e-07,
+      "loss": 1.0206,
+      "step": 8668
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4338611181541084e-07,
+      "loss": 0.8947,
+      "step": 8669
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4308587913976512e-07,
+      "loss": 0.9003,
+      "step": 8670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4278595343933797e-07,
+      "loss": 0.8393,
+      "step": 8671
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4248633474635598e-07,
+      "loss": 0.9458,
+      "step": 8672
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4218702309301335e-07,
+      "loss": 0.7382,
+      "step": 8673
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4188801851147136e-07,
+      "loss": 0.738,
+      "step": 8674
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.41589321033858e-07,
+      "loss": 0.9004,
+      "step": 8675
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4129093069226762e-07,
+      "loss": 0.8197,
+      "step": 8676
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4099284751876318e-07,
+      "loss": 1.1169,
+      "step": 8677
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4069507154537275e-07,
+      "loss": 0.7801,
+      "step": 8678
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4039760280409375e-07,
+      "loss": 0.9528,
+      "step": 8679
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.401004413268876e-07,
+      "loss": 0.9906,
+      "step": 8680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3980358714568475e-07,
+      "loss": 0.7829,
+      "step": 8681
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3950704029238272e-07,
+      "loss": 1.1169,
+      "step": 8682
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.392108007988453e-07,
+      "loss": 0.7517,
+      "step": 8683
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3891486869690308e-07,
+      "loss": 0.9378,
+      "step": 8684
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3861924401835358e-07,
+      "loss": 1.1041,
+      "step": 8685
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3832392679496243e-07,
+      "loss": 0.8494,
+      "step": 8686
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3802891705846154e-07,
+      "loss": 0.7902,
+      "step": 8687
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3773421484054893e-07,
+      "loss": 0.9179,
+      "step": 8688
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.374398201728906e-07,
+      "loss": 0.8913,
+      "step": 8689
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.371457330871192e-07,
+      "loss": 0.8196,
+      "step": 8690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3685195361483515e-07,
+      "loss": 0.8807,
+      "step": 8691
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3655848178760355e-07,
+      "loss": 0.8562,
+      "step": 8692
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3626531763695916e-07,
+      "loss": 0.8431,
+      "step": 8693
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3597246119440176e-07,
+      "loss": 1.0611,
+      "step": 8694
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3567991249139987e-07,
+      "loss": 0.6829,
+      "step": 8695
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3538767155938603e-07,
+      "loss": 0.8018,
+      "step": 8696
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3509573842976274e-07,
+      "loss": 0.9725,
+      "step": 8697
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3480411313389794e-07,
+      "loss": 0.9501,
+      "step": 8698
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3451279570312724e-07,
+      "loss": 0.8841,
+      "step": 8699
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3422178616875158e-07,
+      "loss": 0.9094,
+      "step": 8700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.339310845620406e-07,
+      "loss": 0.8765,
+      "step": 8701
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3364069091423004e-07,
+      "loss": 0.9552,
+      "step": 8702
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.333506052565232e-07,
+      "loss": 0.7815,
+      "step": 8703
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3306082762008853e-07,
+      "loss": 0.7942,
+      "step": 8704
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3277135803606377e-07,
+      "loss": 0.8211,
+      "step": 8705
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3248219653555173e-07,
+      "loss": 0.9814,
+      "step": 8706
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.321933431496235e-07,
+      "loss": 0.8976,
+      "step": 8707
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3190479790931564e-07,
+      "loss": 1.0227,
+      "step": 8708
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3161656084563235e-07,
+      "loss": 1.0066,
+      "step": 8709
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3132863198954514e-07,
+      "loss": 0.8993,
+      "step": 8710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.31041011371992e-07,
+      "loss": 0.8675,
+      "step": 8711
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.307536990238768e-07,
+      "loss": 0.8188,
+      "step": 8712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3046669497607222e-07,
+      "loss": 0.986,
+      "step": 8713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3017999925941659e-07,
+      "loss": 0.9272,
+      "step": 8714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2989361190471526e-07,
+      "loss": 0.9961,
+      "step": 8715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2960753294274063e-07,
+      "loss": 0.7278,
+      "step": 8716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2932176240423145e-07,
+      "loss": 0.807,
+      "step": 8717
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.290363003198941e-07,
+      "loss": 0.945,
+      "step": 8718
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2875114672040177e-07,
+      "loss": 0.9455,
+      "step": 8719
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.284663016363936e-07,
+      "loss": 1.0067,
+      "step": 8720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2818176509847613e-07,
+      "loss": 0.7686,
+      "step": 8721
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.278975371372232e-07,
+      "loss": 1.0013,
+      "step": 8722
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2761361778317538e-07,
+      "loss": 0.9014,
+      "step": 8723
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2733000706683896e-07,
+      "loss": 0.6,
+      "step": 8724
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2704670501868854e-07,
+      "loss": 0.8149,
+      "step": 8725
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2676371166916412e-07,
+      "loss": 0.9656,
+      "step": 8726
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.264810270486747e-07,
+      "loss": 0.8104,
+      "step": 8727
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2619865118759332e-07,
+      "loss": 0.6914,
+      "step": 8728
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2591658411626173e-07,
+      "loss": 0.9976,
+      "step": 8729
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.256348258649883e-07,
+      "loss": 0.9094,
+      "step": 8730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2535337646404789e-07,
+      "loss": 0.8177,
+      "step": 8731
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.250722359436819e-07,
+      "loss": 0.908,
+      "step": 8732
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2479140433409886e-07,
+      "loss": 0.8359,
+      "step": 8733
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2451088166547396e-07,
+      "loss": 0.8347,
+      "step": 8734
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2423066796795045e-07,
+      "loss": 1.1402,
+      "step": 8735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.239507632716359e-07,
+      "loss": 0.9373,
+      "step": 8736
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2367116760660658e-07,
+      "loss": 0.9795,
+      "step": 8737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.233918810029051e-07,
+      "loss": 1.045,
+      "step": 8738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2311290349054115e-07,
+      "loss": 0.9265,
+      "step": 8739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2283423509949009e-07,
+      "loss": 0.8812,
+      "step": 8740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2255587585969498e-07,
+      "loss": 0.9008,
+      "step": 8741
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2227782580106582e-07,
+      "loss": 0.8863,
+      "step": 8742
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2200008495347947e-07,
+      "loss": 0.8548,
+      "step": 8743
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.21722653346778e-07,
+      "loss": 0.8163,
+      "step": 8744
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2144553101077194e-07,
+      "loss": 0.9187,
+      "step": 8745
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2116871797523842e-07,
+      "loss": 0.8088,
+      "step": 8746
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2089221426992103e-07,
+      "loss": 0.8712,
+      "step": 8747
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.206160199245293e-07,
+      "loss": 0.8527,
+      "step": 8748
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2034013496874085e-07,
+      "loss": 1.0134,
+      "step": 8749
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2006455943219962e-07,
+      "loss": 0.9545,
+      "step": 8750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.197892933445156e-07,
+      "loss": 0.7825,
+      "step": 8751
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1951433673526679e-07,
+      "loss": 0.7066,
+      "step": 8752
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.192396896339969e-07,
+      "loss": 0.9271,
+      "step": 8753
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1896535207021697e-07,
+      "loss": 0.8163,
+      "step": 8754
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1869132407340377e-07,
+      "loss": 0.8796,
+      "step": 8755
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1841760567300275e-07,
+      "loss": 0.9534,
+      "step": 8756
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1814419689842404e-07,
+      "loss": 0.9137,
+      "step": 8757
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1787109777904614e-07,
+      "loss": 1.0394,
+      "step": 8758
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1759830834421259e-07,
+      "loss": 0.9068,
+      "step": 8759
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1732582862323559e-07,
+      "loss": 0.801,
+      "step": 8760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1705365864539241e-07,
+      "loss": 0.651,
+      "step": 8761
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1678179843992831e-07,
+      "loss": 1.0716,
+      "step": 8762
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1651024803605326e-07,
+      "loss": 0.8412,
+      "step": 8763
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1623900746294725e-07,
+      "loss": 0.8577,
+      "step": 8764
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1596807674975363e-07,
+      "loss": 0.9258,
+      "step": 8765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1569745592558446e-07,
+      "loss": 0.7716,
+      "step": 8766
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1542714501951745e-07,
+      "loss": 0.9074,
+      "step": 8767
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.151571440605984e-07,
+      "loss": 0.9609,
+      "step": 8768
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.148874530778381e-07,
+      "loss": 0.8188,
+      "step": 8769
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1461807210021568e-07,
+      "loss": 0.765,
+      "step": 8770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1434900115667435e-07,
+      "loss": 0.8972,
+      "step": 8771
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1408024027612796e-07,
+      "loss": 0.7995,
+      "step": 8772
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1381178948745341e-07,
+      "loss": 0.8662,
+      "step": 8773
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1354364881949664e-07,
+      "loss": 0.8001,
+      "step": 8774
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1327581830106792e-07,
+      "loss": 0.9153,
+      "step": 8775
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1300829796094725e-07,
+      "loss": 0.8645,
+      "step": 8776
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1274108782787829e-07,
+      "loss": 0.94,
+      "step": 8777
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.124741879305744e-07,
+      "loss": 0.8206,
+      "step": 8778
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1220759829771165e-07,
+      "loss": 0.9419,
+      "step": 8779
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1194131895793747e-07,
+      "loss": 1.0499,
+      "step": 8780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1167534993986194e-07,
+      "loss": 0.8817,
+      "step": 8781
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1140969127206457e-07,
+      "loss": 0.8633,
+      "step": 8782
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1114434298308885e-07,
+      "loss": 0.9625,
+      "step": 8783
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1087930510144795e-07,
+      "loss": 0.7988,
+      "step": 8784
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1061457765561944e-07,
+      "loss": 0.8008,
+      "step": 8785
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1035016067404857e-07,
+      "loss": 1.0448,
+      "step": 8786
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1008605418514628e-07,
+      "loss": 0.9878,
+      "step": 8787
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0982225821729153e-07,
+      "loss": 0.9366,
+      "step": 8788
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0955877279882898e-07,
+      "loss": 1.0422,
+      "step": 8789
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0929559795806998e-07,
+      "loss": 0.8562,
+      "step": 8790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0903273372329225e-07,
+      "loss": 0.7615,
+      "step": 8791
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0877018012274154e-07,
+      "loss": 1.0317,
+      "step": 8792
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.085079371846286e-07,
+      "loss": 0.9895,
+      "step": 8793
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0824600493713155e-07,
+      "loss": 1.067,
+      "step": 8794
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0798438340839423e-07,
+      "loss": 0.8959,
+      "step": 8795
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0772307262652947e-07,
+      "loss": 1.2524,
+      "step": 8796
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0746207261961349e-07,
+      "loss": 1.1436,
+      "step": 8797
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0720138341569186e-07,
+      "loss": 0.8371,
+      "step": 8798
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0694100504277449e-07,
+      "loss": 1.0194,
+      "step": 8799
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0668093752884068e-07,
+      "loss": 0.7493,
+      "step": 8800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0642118090183272e-07,
+      "loss": 0.9571,
+      "step": 8801
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0616173518966333e-07,
+      "loss": 0.8823,
+      "step": 8802
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0590260042020783e-07,
+      "loss": 0.8216,
+      "step": 8803
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0564377662131231e-07,
+      "loss": 0.8041,
+      "step": 8804
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0538526382078618e-07,
+      "loss": 0.7058,
+      "step": 8805
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0512706204640721e-07,
+      "loss": 0.9056,
+      "step": 8806
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0486917132591822e-07,
+      "loss": 1.0106,
+      "step": 8807
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0461159168703072e-07,
+      "loss": 0.8406,
+      "step": 8808
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0435432315742055e-07,
+      "loss": 1.207,
+      "step": 8809
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0409736576473227e-07,
+      "loss": 0.939,
+      "step": 8810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0384071953657481e-07,
+      "loss": 0.7596,
+      "step": 8811
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.035843845005261e-07,
+      "loss": 0.9864,
+      "step": 8812
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.033283606841281e-07,
+      "loss": 0.9213,
+      "step": 8813
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0307264811489148e-07,
+      "loss": 0.8678,
+      "step": 8814
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0281724682029125e-07,
+      "loss": 0.9581,
+      "step": 8815
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0256215682777181e-07,
+      "loss": 0.7242,
+      "step": 8816
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0230737816474155e-07,
+      "loss": 0.9333,
+      "step": 8817
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0205291085857693e-07,
+      "loss": 1.3792,
+      "step": 8818
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0179875493662005e-07,
+      "loss": 0.9088,
+      "step": 8819
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0154491042618075e-07,
+      "loss": 0.8225,
+      "step": 8820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.012913773545332e-07,
+      "loss": 0.9644,
+      "step": 8821
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0103815574892062e-07,
+      "loss": 0.7724,
+      "step": 8822
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0078524563655123e-07,
+      "loss": 0.964,
+      "step": 8823
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0053264704460063e-07,
+      "loss": 0.8344,
+      "step": 8824
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0028036000021012e-07,
+      "loss": 0.9131,
+      "step": 8825
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0002838453048769e-07,
+      "loss": 0.9014,
+      "step": 8826
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.977672066250864e-08,
+      "loss": 0.9679,
+      "step": 8827
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.952536842331406e-08,
+      "loss": 0.8236,
+      "step": 8828
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.927432783991164e-08,
+      "loss": 0.7802,
+      "step": 8829
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.902359893927548e-08,
+      "loss": 0.9159,
+      "step": 8830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.87731817483467e-08,
+      "loss": 1.0462,
+      "step": 8831
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.852307629403278e-08,
+      "loss": 0.7632,
+      "step": 8832
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.827328260320689e-08,
+      "loss": 0.7517,
+      "step": 8833
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.802380070270988e-08,
+      "loss": 0.9895,
+      "step": 8834
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.777463061934833e-08,
+      "loss": 1.0161,
+      "step": 8835
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.752577237989613e-08,
+      "loss": 1.0208,
+      "step": 8836
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.72772260110919e-08,
+      "loss": 0.7952,
+      "step": 8837
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.702899153964295e-08,
+      "loss": 0.9096,
+      "step": 8838
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.678106899222161e-08,
+      "loss": 1.0175,
+      "step": 8839
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.653345839546723e-08,
+      "loss": 0.8577,
+      "step": 8840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.628615977598554e-08,
+      "loss": 1.0704,
+      "step": 8841
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.603917316034893e-08,
+      "loss": 1.169,
+      "step": 8842
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.579249857509553e-08,
+      "loss": 0.9591,
+      "step": 8843
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.554613604673146e-08,
+      "loss": 0.9598,
+      "step": 8844
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.530008560172754e-08,
+      "loss": 0.9445,
+      "step": 8845
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.505434726652196e-08,
+      "loss": 1.0151,
+      "step": 8846
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.480892106751992e-08,
+      "loss": 0.9573,
+      "step": 8847
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.456380703109203e-08,
+      "loss": 0.8878,
+      "step": 8848
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.431900518357516e-08,
+      "loss": 0.9656,
+      "step": 8849
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.4074515551274e-08,
+      "loss": 0.9191,
+      "step": 8850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.383033816045916e-08,
+      "loss": 1.0429,
+      "step": 8851
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.3586473037367e-08,
+      "loss": 0.89,
+      "step": 8852
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.33429202082009e-08,
+      "loss": 0.6837,
+      "step": 8853
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.309967969913024e-08,
+      "loss": 0.742,
+      "step": 8854
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.285675153629181e-08,
+      "loss": 0.8368,
+      "step": 8855
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.261413574578837e-08,
+      "loss": 0.9928,
+      "step": 8856
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.237183235368807e-08,
+      "loss": 0.9835,
+      "step": 8857
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.212984138602675e-08,
+      "loss": 0.7858,
+      "step": 8858
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.188816286880696e-08,
+      "loss": 0.8413,
+      "step": 8859
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.16467968279966e-08,
+      "loss": 0.6813,
+      "step": 8860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.140574328952989e-08,
+      "loss": 0.8608,
+      "step": 8861
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.11650022793088e-08,
+      "loss": 0.8713,
+      "step": 8862
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.092457382320063e-08,
+      "loss": 0.7107,
+      "step": 8863
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.068445794703938e-08,
+      "loss": 0.9134,
+      "step": 8864
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.044465467662543e-08,
+      "loss": 0.8225,
+      "step": 8865
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.020516403772583e-08,
+      "loss": 0.9227,
+      "step": 8866
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.996598605607365e-08,
+      "loss": 0.7421,
+      "step": 8867
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.9727120757369e-08,
+      "loss": 1.0443,
+      "step": 8868
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.948856816727701e-08,
+      "loss": 0.9589,
+      "step": 8869
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.925032831143087e-08,
+      "loss": 0.8965,
+      "step": 8870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.901240121542908e-08,
+      "loss": 0.7594,
+      "step": 8871
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.877478690483754e-08,
+      "loss": 0.8825,
+      "step": 8872
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.853748540518714e-08,
+      "loss": 0.9479,
+      "step": 8873
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.830049674197615e-08,
+      "loss": 0.9289,
+      "step": 8874
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.806382094066889e-08,
+      "loss": 0.9374,
+      "step": 8875
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.782745802669667e-08,
+      "loss": 0.8648,
+      "step": 8876
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.759140802545585e-08,
+      "loss": 0.8144,
+      "step": 8877
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.735567096231045e-08,
+      "loss": 0.962,
+      "step": 8878
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.712024686259057e-08,
+      "loss": 1.2144,
+      "step": 8879
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.688513575159229e-08,
+      "loss": 0.8417,
+      "step": 8880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.665033765457808e-08,
+      "loss": 0.905,
+      "step": 8881
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.641585259677742e-08,
+      "loss": 0.9808,
+      "step": 8882
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.618168060338517e-08,
+      "loss": 0.8279,
+      "step": 8883
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.594782169956416e-08,
+      "loss": 0.8421,
+      "step": 8884
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.571427591044067e-08,
+      "loss": 1.055,
+      "step": 8885
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.548104326111128e-08,
+      "loss": 0.8414,
+      "step": 8886
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.524812377663526e-08,
+      "loss": 1.2104,
+      "step": 8887
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.501551748204062e-08,
+      "loss": 0.8782,
+      "step": 8888
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.478322440232034e-08,
+      "loss": 1.0471,
+      "step": 8889
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.45512445624348e-08,
+      "loss": 0.7771,
+      "step": 8890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.431957798730972e-08,
+      "loss": 0.9645,
+      "step": 8891
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.408822470183853e-08,
+      "loss": 0.8644,
+      "step": 8892
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.385718473087867e-08,
+      "loss": 1.0227,
+      "step": 8893
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.362645809925695e-08,
+      "loss": 0.7503,
+      "step": 8894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.339604483176388e-08,
+      "loss": 0.9093,
+      "step": 8895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.3165944953158e-08,
+      "loss": 0.8579,
+      "step": 8896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.293615848816249e-08,
+      "loss": 0.902,
+      "step": 8897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.270668546146897e-08,
+      "loss": 1.0601,
+      "step": 8898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.247752589773405e-08,
+      "loss": 0.9935,
+      "step": 8899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.224867982158102e-08,
+      "loss": 0.8943,
+      "step": 8900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.202014725759821e-08,
+      "loss": 0.7451,
+      "step": 8901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.179192823034332e-08,
+      "loss": 0.795,
+      "step": 8902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.156402276433706e-08,
+      "loss": 0.8263,
+      "step": 8903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.13364308840685e-08,
+      "loss": 0.778,
+      "step": 8904
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.110915261399176e-08,
+      "loss": 1.0232,
+      "step": 8905
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.088218797852864e-08,
+      "loss": 0.7922,
+      "step": 8906
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.065553700206596e-08,
+      "loss": 0.8218,
+      "step": 8907
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.04291997089579e-08,
+      "loss": 0.9154,
+      "step": 8908
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.020317612352334e-08,
+      "loss": 0.772,
+      "step": 8909
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.997746627004986e-08,
+      "loss": 0.8498,
+      "step": 8910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.975207017278874e-08,
+      "loss": 0.86,
+      "step": 8911
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.952698785595958e-08,
+      "loss": 0.9616,
+      "step": 8912
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.93022193437467e-08,
+      "loss": 0.7272,
+      "step": 8913
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.907776466030248e-08,
+      "loss": 1.0239,
+      "step": 8914
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.885362382974393e-08,
+      "loss": 0.9132,
+      "step": 8915
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.862979687615513e-08,
+      "loss": 0.9538,
+      "step": 8916
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.84062838235855e-08,
+      "loss": 0.9429,
+      "step": 8917
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.818308469605284e-08,
+      "loss": 0.8057,
+      "step": 8918
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.79601995175393e-08,
+      "loss": 0.7427,
+      "step": 8919
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.773762831199371e-08,
+      "loss": 1.1314,
+      "step": 8920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.751537110333095e-08,
+      "loss": 0.7672,
+      "step": 8921
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.729342791543359e-08,
+      "loss": 1.1193,
+      "step": 8922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.707179877214854e-08,
+      "loss": 0.7619,
+      "step": 8923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.685048369729042e-08,
+      "loss": 0.8398,
+      "step": 8924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.662948271463887e-08,
+      "loss": 0.7717,
+      "step": 8925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.640879584794125e-08,
+      "loss": 1.0374,
+      "step": 8926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.618842312090957e-08,
+      "loss": 0.805,
+      "step": 8927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.596836455722322e-08,
+      "loss": 0.982,
+      "step": 8928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.574862018052731e-08,
+      "loss": 0.9168,
+      "step": 8929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.552919001443359e-08,
+      "loss": 0.9428,
+      "step": 8930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.531007408251989e-08,
+      "loss": 0.9952,
+      "step": 8931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.509127240833002e-08,
+      "loss": 0.902,
+      "step": 8932
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.487278501537387e-08,
+      "loss": 0.9845,
+      "step": 8933
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.465461192712864e-08,
+      "loss": 0.8084,
+      "step": 8934
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.443675316703624e-08,
+      "loss": 0.775,
+      "step": 8935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.421920875850663e-08,
+      "loss": 0.9385,
+      "step": 8936
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.400197872491343e-08,
+      "loss": 0.9315,
+      "step": 8937
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.378506308959932e-08,
+      "loss": 0.7924,
+      "step": 8938
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.356846187587096e-08,
+      "loss": 0.9144,
+      "step": 8939
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.33521751070031e-08,
+      "loss": 0.9805,
+      "step": 8940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.313620280623479e-08,
+      "loss": 0.7276,
+      "step": 8941
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.29205449967728e-08,
+      "loss": 0.9552,
+      "step": 8942
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.27052017017893e-08,
+      "loss": 0.8099,
+      "step": 8943
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.249017294442306e-08,
+      "loss": 0.8926,
+      "step": 8944
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.227545874777863e-08,
+      "loss": 1.0869,
+      "step": 8945
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.206105913492756e-08,
+      "loss": 0.6811,
+      "step": 8946
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.184697412890639e-08,
+      "loss": 1.0924,
+      "step": 8947
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.163320375271943e-08,
+      "loss": 0.9437,
+      "step": 8948
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.141974802933493e-08,
+      "loss": 0.8825,
+      "step": 8949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.120660698169023e-08,
+      "loss": 0.9821,
+      "step": 8950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.099378063268635e-08,
+      "loss": 1.0697,
+      "step": 8951
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.0781269005192e-08,
+      "loss": 1.0297,
+      "step": 8952
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.056907212204089e-08,
+      "loss": 1.025,
+      "step": 8953
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.035719000603413e-08,
+      "loss": 0.7977,
+      "step": 8954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.014562267993818e-08,
+      "loss": 1.2075,
+      "step": 8955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.993437016648618e-08,
+      "loss": 0.7931,
+      "step": 8956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.972343248837665e-08,
+      "loss": 1.104,
+      "step": 8957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.951280966827578e-08,
+      "loss": 0.8299,
+      "step": 8958
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.930250172881413e-08,
+      "loss": 1.0534,
+      "step": 8959
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.909250869258931e-08,
+      "loss": 0.7543,
+      "step": 8960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.888283058216527e-08,
+      "loss": 0.8495,
+      "step": 8961
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.86734674200723e-08,
+      "loss": 0.711,
+      "step": 8962
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.846441922880575e-08,
+      "loss": 0.8729,
+      "step": 8963
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.825568603082832e-08,
+      "loss": 0.8515,
+      "step": 8964
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.804726784856808e-08,
+      "loss": 0.9051,
+      "step": 8965
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.783916470442009e-08,
+      "loss": 0.967,
+      "step": 8966
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.763137662074447e-08,
+      "loss": 1.0997,
+      "step": 8967
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.742390361986772e-08,
+      "loss": 1.0068,
+      "step": 8968
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.721674572408365e-08,
+      "loss": 0.7651,
+      "step": 8969
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.700990295565112e-08,
+      "loss": 0.9311,
+      "step": 8970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.680337533679503e-08,
+      "loss": 0.91,
+      "step": 8971
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.659716288970696e-08,
+      "loss": 1.1007,
+      "step": 8972
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.639126563654419e-08,
+      "loss": 1.0034,
+      "step": 8973
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6185683599431e-08,
+      "loss": 0.8243,
+      "step": 8974
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.59804168004564e-08,
+      "loss": 1.0651,
+      "step": 8975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577546526167677e-08,
+      "loss": 0.7798,
+      "step": 8976
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557082900511413e-08,
+      "loss": 1.1291,
+      "step": 8977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536650805275657e-08,
+      "loss": 0.8984,
+      "step": 8978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516250242655819e-08,
+      "loss": 0.7518,
+      "step": 8979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.495881214843947e-08,
+      "loss": 0.9959,
+      "step": 8980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.475543724028687e-08,
+      "loss": 1.0238,
+      "step": 8981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.455237772395328e-08,
+      "loss": 0.9584,
+      "step": 8982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.43496336212569e-08,
+      "loss": 0.9091,
+      "step": 8983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.41472049539833e-08,
+      "loss": 0.8959,
+      "step": 8984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.394509174388274e-08,
+      "loss": 0.8909,
+      "step": 8985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.374329401267287e-08,
+      "loss": 0.8105,
+      "step": 8986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.354181178203633e-08,
+      "loss": 0.8629,
+      "step": 8987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.334064507362247e-08,
+      "loss": 0.8369,
+      "step": 8988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.313979390904668e-08,
+      "loss": 0.7838,
+      "step": 8989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.2939258309891e-08,
+      "loss": 0.758,
+      "step": 8990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.273903829770189e-08,
+      "loss": 0.6805,
+      "step": 8991
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.25391338939938e-08,
+      "loss": 0.9571,
+      "step": 8992
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.233954512024586e-08,
+      "loss": 1.2119,
+      "step": 8993
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.214027199790462e-08,
+      "loss": 0.8437,
+      "step": 8994
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.194131454838125e-08,
+      "loss": 0.9279,
+      "step": 8995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.174267279305368e-08,
+      "loss": 0.8195,
+      "step": 8996
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.15443467532668e-08,
+      "loss": 0.6663,
+      "step": 8997
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.134633645032994e-08,
+      "loss": 0.9441,
+      "step": 8998
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.114864190551939e-08,
+      "loss": 1.0173,
+      "step": 8999
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.095126314007781e-08,
+      "loss": 0.8508,
+      "step": 9000
+    },
+    {
+      "epoch": 1.87,
+      "eval_loss": NaN,
+      "eval_runtime": 15.0106,
+      "eval_samples_per_second": 352.75,
+      "eval_steps_per_second": 44.102,
+      "step": 9000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.075420017521293e-08,
+      "loss": 0.8437,
+      "step": 9001
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.055745303210015e-08,
+      "loss": 0.8408,
+      "step": 9002
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.036102173187885e-08,
+      "loss": 0.8385,
+      "step": 9003
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.016490629565586e-08,
+      "loss": 0.826,
+      "step": 9004
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.996910674450395e-08,
+      "loss": 1.0228,
+      "step": 9005
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9773623099462e-08,
+      "loss": 0.9783,
+      "step": 9006
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.957845538153417e-08,
+      "loss": 0.8994,
+      "step": 9007
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9383603611691705e-08,
+      "loss": 0.8814,
+      "step": 9008
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9189067810870854e-08,
+      "loss": 0.8656,
+      "step": 9009
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.8994847999975235e-08,
+      "loss": 0.7666,
+      "step": 9010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.880094419987314e-08,
+      "loss": 1.0052,
+      "step": 9011
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.860735643139958e-08,
+      "loss": 0.7893,
+      "step": 9012
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.841408471535592e-08,
+      "loss": 1.0014,
+      "step": 9013
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.822112907250887e-08,
+      "loss": 0.8081,
+      "step": 9014
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.802848952359152e-08,
+      "loss": 1.1453,
+      "step": 9015
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.783616608930298e-08,
+      "loss": 1.0877,
+      "step": 9016
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7644158790308706e-08,
+      "loss": 0.728,
+      "step": 9017
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.745246764723955e-08,
+      "loss": 0.9963,
+      "step": 9018
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7261092680692685e-08,
+      "loss": 0.7619,
+      "step": 9019
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7070033911231336e-08,
+      "loss": 0.8087,
+      "step": 9020
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.6879291359385074e-08,
+      "loss": 1.1866,
+      "step": 9021
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.668886504564852e-08,
+      "loss": 0.7834,
+      "step": 9022
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.6498754990483624e-08,
+      "loss": 0.8715,
+      "step": 9023
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.630896121431739e-08,
+      "loss": 1.0776,
+      "step": 9024
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.6119483737543496e-08,
+      "loss": 0.8734,
+      "step": 9025
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.593032258052066e-08,
+      "loss": 0.7871,
+      "step": 9026
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.574147776357463e-08,
+      "loss": 1.0677,
+      "step": 9027
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.5552949306996836e-08,
+      "loss": 0.9138,
+      "step": 9028
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.536473723104474e-08,
+      "loss": 0.6629,
+      "step": 9029
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.517684155594116e-08,
+      "loss": 0.9081,
+      "step": 9030
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.498926230187595e-08,
+      "loss": 0.8881,
+      "step": 9031
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.480199948900433e-08,
+      "loss": 1.1874,
+      "step": 9032
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.4615053137447854e-08,
+      "loss": 0.9569,
+      "step": 9033
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.442842326729347e-08,
+      "loss": 0.8031,
+      "step": 9034
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.4242109898594796e-08,
+      "loss": 0.9998,
+      "step": 9035
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.405611305137148e-08,
+      "loss": 0.7781,
+      "step": 9036
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.3870432745608545e-08,
+      "loss": 0.9311,
+      "step": 9037
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.368506900125736e-08,
+      "loss": 0.8615,
+      "step": 9038
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.3500021838234994e-08,
+      "loss": 0.8552,
+      "step": 9039
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.331529127642521e-08,
+      "loss": 1.0891,
+      "step": 9040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.3130877335677455e-08,
+      "loss": 0.9291,
+      "step": 9041
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.294678003580622e-08,
+      "loss": 0.9409,
+      "step": 9042
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.276299939659301e-08,
+      "loss": 0.9913,
+      "step": 9043
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.257953543778538e-08,
+      "loss": 1.0143,
+      "step": 9044
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.2396388179096864e-08,
+      "loss": 0.8186,
+      "step": 9045
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.221355764020541e-08,
+      "loss": 0.9608,
+      "step": 9046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.2031043840756964e-08,
+      "loss": 0.9215,
+      "step": 9047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.1848846800362835e-08,
+      "loss": 0.8642,
+      "step": 9048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.16669665385997e-08,
+      "loss": 0.8216,
+      "step": 9049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.148540307501026e-08,
+      "loss": 0.824,
+      "step": 9050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.130415642910391e-08,
+      "loss": 0.8007,
+      "step": 9051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.1123226620355734e-08,
+      "loss": 0.9334,
+      "step": 9052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.094261366820652e-08,
+      "loss": 0.8264,
+      "step": 9053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.0762317592063067e-08,
+      "loss": 0.8115,
+      "step": 9054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.058233841129789e-08,
+      "loss": 0.9564,
+      "step": 9055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.0402676145249846e-08,
+      "loss": 0.8889,
+      "step": 9056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.022333081322417e-08,
+      "loss": 0.9086,
+      "step": 9057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.004430243449076e-08,
+      "loss": 1.0309,
+      "step": 9058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.9865591028286913e-08,
+      "loss": 0.7675,
+      "step": 9059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.968719661381427e-08,
+      "loss": 0.8004,
+      "step": 9060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.950911921024248e-08,
+      "loss": 1.1211,
+      "step": 9061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.933135883670492e-08,
+      "loss": 0.7869,
+      "step": 9062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.91539155123023e-08,
+      "loss": 0.901,
+      "step": 9063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.897678925610072e-08,
+      "loss": 0.8745,
+      "step": 9064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8799980087132936e-08,
+      "loss": 0.7104,
+      "step": 9065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.862348802439642e-08,
+      "loss": 0.7841,
+      "step": 9066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8447313086855686e-08,
+      "loss": 0.895,
+      "step": 9067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8271455293440237e-08,
+      "loss": 0.837,
+      "step": 9068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.809591466304664e-08,
+      "loss": 1.1527,
+      "step": 9069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.792069121453646e-08,
+      "loss": 1.0853,
+      "step": 9070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.774578496673698e-08,
+      "loss": 0.8859,
+      "step": 9071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.757119593844217e-08,
+      "loss": 0.7974,
+      "step": 9072
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.739692414841201e-08,
+      "loss": 0.8891,
+      "step": 9073
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.722296961537154e-08,
+      "loss": 0.8108,
+      "step": 9074
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.704933235801245e-08,
+      "loss": 0.9053,
+      "step": 9075
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.687601239499151e-08,
+      "loss": 0.9003,
+      "step": 9076
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.67030097449328e-08,
+      "loss": 0.8375,
+      "step": 9077
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.653032442642446e-08,
+      "loss": 1.1031,
+      "step": 9078
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.635795645802232e-08,
+      "loss": 0.8382,
+      "step": 9079
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.618590585824656e-08,
+      "loss": 0.8688,
+      "step": 9080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.601417264558505e-08,
+      "loss": 0.9199,
+      "step": 9081
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.584275683848971e-08,
+      "loss": 1.1365,
+      "step": 9082
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.567165845537913e-08,
+      "loss": 1.1024,
+      "step": 9083
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.550087751463794e-08,
+      "loss": 0.7422,
+      "step": 9084
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.533041403461713e-08,
+      "loss": 0.9149,
+      "step": 9085
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.5160268033631716e-08,
+      "loss": 0.8164,
+      "step": 9086
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.49904395299654e-08,
+      "loss": 1.0618,
+      "step": 9087
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.482092854186526e-08,
+      "loss": 1.0036,
+      "step": 9088
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.465173508754539e-08,
+      "loss": 0.9272,
+      "step": 9089
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.4482859185185576e-08,
+      "loss": 0.9098,
+      "step": 9090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.4314300852931975e-08,
+      "loss": 1.1384,
+      "step": 9091
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.414606010889577e-08,
+      "loss": 0.8682,
+      "step": 9092
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.397813697115449e-08,
+      "loss": 0.8418,
+      "step": 9093
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3810531457751044e-08,
+      "loss": 0.9708,
+      "step": 9094
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3643243586695356e-08,
+      "loss": 0.9308,
+      "step": 9095
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.347627337596205e-08,
+      "loss": 1.1293,
+      "step": 9096
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.330962084349277e-08,
+      "loss": 1.1552,
+      "step": 9097
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.314328600719286e-08,
+      "loss": 0.7941,
+      "step": 9098
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.297726888493636e-08,
+      "loss": 0.9275,
+      "step": 9099
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2811569494561334e-08,
+      "loss": 1.0203,
+      "step": 9100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2646187853872196e-08,
+      "loss": 0.7914,
+      "step": 9101
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.248112398063875e-08,
+      "loss": 0.8359,
+      "step": 9102
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.231637789259779e-08,
+      "loss": 0.7969,
+      "step": 9103
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2151949607450836e-08,
+      "loss": 0.7382,
+      "step": 9104
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.198783914286608e-08,
+      "loss": 0.8708,
+      "step": 9105
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.182404651647676e-08,
+      "loss": 0.7673,
+      "step": 9106
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.166057174588245e-08,
+      "loss": 1.0287,
+      "step": 9107
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.149741484864877e-08,
+      "loss": 0.8401,
+      "step": 9108
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.1334575842307045e-08,
+      "loss": 1.0616,
+      "step": 9109
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.117205474435326e-08,
+      "loss": 0.9841,
+      "step": 9110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.100985157225145e-08,
+      "loss": 0.8877,
+      "step": 9111
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.084796634343002e-08,
+      "loss": 0.8901,
+      "step": 9112
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.068639907528371e-08,
+      "loss": 0.8253,
+      "step": 9113
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0525149785171966e-08,
+      "loss": 1.0443,
+      "step": 9114
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.036421849042227e-08,
+      "loss": 0.9038,
+      "step": 9115
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.020360520832578e-08,
+      "loss": 0.7869,
+      "step": 9116
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.004330995614103e-08,
+      "loss": 0.9789,
+      "step": 9117
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.988333275109091e-08,
+      "loss": 0.8931,
+      "step": 9118
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.9723673610365666e-08,
+      "loss": 0.7156,
+      "step": 9119
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.9564332551120574e-08,
+      "loss": 0.6838,
+      "step": 9120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.940530959047695e-08,
+      "loss": 1.0805,
+      "step": 9121
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.9246604745520774e-08,
+      "loss": 0.8642,
+      "step": 9122
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.908821803330642e-08,
+      "loss": 1.1389,
+      "step": 9123
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.893014947085127e-08,
+      "loss": 0.9444,
+      "step": 9124
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.877239907514074e-08,
+      "loss": 0.9135,
+      "step": 9125
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.8614966863123934e-08,
+      "loss": 0.889,
+      "step": 9126
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.845785285171799e-08,
+      "loss": 0.7592,
+      "step": 9127
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.83010570578044e-08,
+      "loss": 0.907,
+      "step": 9128
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.814457949823102e-08,
+      "loss": 0.872,
+      "step": 9129
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.798842018981108e-08,
+      "loss": 0.8124,
+      "step": 9130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.783257914932414e-08,
+      "loss": 0.7714,
+      "step": 9131
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7677056393515154e-08,
+      "loss": 0.7628,
+      "step": 9132
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.752185193909508e-08,
+      "loss": 0.8174,
+      "step": 9133
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7366965802740236e-08,
+      "loss": 0.8325,
+      "step": 9134
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.721239800109399e-08,
+      "loss": 0.9285,
+      "step": 9135
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.705814855076406e-08,
+      "loss": 0.8824,
+      "step": 9136
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.690421746832484e-08,
+      "loss": 0.8579,
+      "step": 9137
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.675060477031544e-08,
+      "loss": 0.7457,
+      "step": 9138
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.659731047324266e-08,
+      "loss": 0.9032,
+      "step": 9139
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.644433459357699e-08,
+      "loss": 1.1028,
+      "step": 9140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.629167714775661e-08,
+      "loss": 0.7792,
+      "step": 9141
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.61393381521834e-08,
+      "loss": 1.1013,
+      "step": 9142
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.598731762322727e-08,
+      "loss": 0.6506,
+      "step": 9143
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5835615577222146e-08,
+      "loss": 1.093,
+      "step": 9144
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.568423203046867e-08,
+      "loss": 0.8001,
+      "step": 9145
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5533166999232815e-08,
+      "loss": 1.0611,
+      "step": 9146
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.538242049974694e-08,
+      "loss": 1.0791,
+      "step": 9147
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.523199254820808e-08,
+      "loss": 0.8449,
+      "step": 9148
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5081883160780315e-08,
+      "loss": 0.8439,
+      "step": 9149
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.49320923535924e-08,
+      "loss": 0.8167,
+      "step": 9150
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.478262014273981e-08,
+      "loss": 0.9283,
+      "step": 9151
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.463346654428301e-08,
+      "loss": 0.9576,
+      "step": 9152
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.4484631574248527e-08,
+      "loss": 1.1238,
+      "step": 9153
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.4336115248628566e-08,
+      "loss": 1.0214,
+      "step": 9154
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.418791758338202e-08,
+      "loss": 1.0576,
+      "step": 9155
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.40400385944315e-08,
+      "loss": 0.8565,
+      "step": 9156
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.389247829766762e-08,
+      "loss": 0.8673,
+      "step": 9157
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.374523670894503e-08,
+      "loss": 1.0911,
+      "step": 9158
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.35983138440854e-08,
+      "loss": 0.9061,
+      "step": 9159
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.345170971887479e-08,
+      "loss": 0.9734,
+      "step": 9160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.330542434906658e-08,
+      "loss": 0.8997,
+      "step": 9161
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.315945775037854e-08,
+      "loss": 0.9204,
+      "step": 9162
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.3013809938495455e-08,
+      "loss": 0.6734,
+      "step": 9163
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.28684809290668e-08,
+      "loss": 1.211,
+      "step": 9164
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.272347073770776e-08,
+      "loss": 0.9972,
+      "step": 9165
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.25787793800002e-08,
+      "loss": 0.874,
+      "step": 9166
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.243440687149135e-08,
+      "loss": 0.8233,
+      "step": 9167
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.229035322769314e-08,
+      "loss": 0.937,
+      "step": 9168
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.214661846408518e-08,
+      "loss": 1.1607,
+      "step": 9169
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.2003202596111113e-08,
+      "loss": 0.7561,
+      "step": 9170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.186010563918129e-08,
+      "loss": 0.7525,
+      "step": 9171
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1717327608671074e-08,
+      "loss": 0.9677,
+      "step": 9172
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.15748685199222e-08,
+      "loss": 0.9215,
+      "step": 9173
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.143272838824174e-08,
+      "loss": 0.9015,
+      "step": 9174
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1290907228903155e-08,
+      "loss": 0.8,
+      "step": 9175
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1149405057144585e-08,
+      "loss": 0.937,
+      "step": 9176
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.100822188817054e-08,
+      "loss": 0.8152,
+      "step": 9177
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0867357737151213e-08,
+      "loss": 0.9334,
+      "step": 9178
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0726812619222835e-08,
+      "loss": 0.8239,
+      "step": 9179
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0586586549485984e-08,
+      "loss": 0.875,
+      "step": 9180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.044667954300862e-08,
+      "loss": 0.8406,
+      "step": 9181
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.03070916148237e-08,
+      "loss": 0.9834,
+      "step": 9182
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.016782277993024e-08,
+      "loss": 0.945,
+      "step": 9183
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.002887305329194e-08,
+      "loss": 0.867,
+      "step": 9184
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9890242449839177e-08,
+      "loss": 0.7194,
+      "step": 9185
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9751930984468045e-08,
+      "loss": 0.9219,
+      "step": 9186
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.961393867204032e-08,
+      "loss": 1.0214,
+      "step": 9187
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9476265527382473e-08,
+      "loss": 0.7348,
+      "step": 9188
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9338911565288008e-08,
+      "loss": 1.0428,
+      "step": 9189
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.920187680051545e-08,
+      "loss": 0.6975,
+      "step": 9190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9065161247789352e-08,
+      "loss": 0.8457,
+      "step": 9191
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8928764921799634e-08,
+      "loss": 1.0746,
+      "step": 9192
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8792687837202238e-08,
+      "loss": 0.8656,
+      "step": 9193
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8656930008618465e-08,
+      "loss": 0.799,
+      "step": 9194
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8521491450635652e-08,
+      "loss": 0.836,
+      "step": 9195
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.838637217780615e-08,
+      "loss": 0.7688,
+      "step": 9196
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8251572204649357e-08,
+      "loss": 0.9457,
+      "step": 9197
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8117091545648677e-08,
+      "loss": 0.7907,
+      "step": 9198
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.798293021525489e-08,
+      "loss": 0.7855,
+      "step": 9199
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.784908822788279e-08,
+      "loss": 0.8825,
+      "step": 9200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.7715565597914216e-08,
+      "loss": 0.8302,
+      "step": 9201
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.758236233969602e-08,
+      "loss": 1.0683,
+      "step": 9202
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.7449478467540756e-08,
+      "loss": 0.7978,
+      "step": 9203
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.7316913995726998e-08,
+      "loss": 0.9211,
+      "step": 9204
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.718466893849869e-08,
+      "loss": 1.0069,
+      "step": 9205
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.70527433100658e-08,
+      "loss": 0.9548,
+      "step": 9206
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6921137124603312e-08,
+      "loss": 0.7591,
+      "step": 9207
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.678985039625226e-08,
+      "loss": 1.0447,
+      "step": 9208
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6658883139120016e-08,
+      "loss": 0.8782,
+      "step": 9209
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.652823536727833e-08,
+      "loss": 0.8903,
+      "step": 9210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.639790709476564e-08,
+      "loss": 0.8285,
+      "step": 9211
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.626789833558574e-08,
+      "loss": 0.806,
+      "step": 9212
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.613820910370779e-08,
+      "loss": 0.856,
+      "step": 9213
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.600883941306731e-08,
+      "loss": 0.8743,
+      "step": 9214
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5879789277564848e-08,
+      "loss": 0.7857,
+      "step": 9215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.575105871106631e-08,
+      "loss": 0.9136,
+      "step": 9216
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5622647727404635e-08,
+      "loss": 1.0114,
+      "step": 9217
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5494556340377118e-08,
+      "loss": 0.9757,
+      "step": 9218
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5366784563747412e-08,
+      "loss": 1.1459,
+      "step": 9219
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.523933241124421e-08,
+      "loss": 0.9005,
+      "step": 9220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5112199896562882e-08,
+      "loss": 0.8591,
+      "step": 9221
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4985387033362837e-08,
+      "loss": 0.8262,
+      "step": 9222
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.485889383527118e-08,
+      "loss": 0.9362,
+      "step": 9223
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4732720315878697e-08,
+      "loss": 0.7278,
+      "step": 9224
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.460686648874322e-08,
+      "loss": 1.1004,
+      "step": 9225
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4481332367387586e-08,
+      "loss": 0.6996,
+      "step": 9226
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4356117965300684e-08,
+      "loss": 0.7828,
+      "step": 9227
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.423122329593641e-08,
+      "loss": 0.8114,
+      "step": 9228
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4106648372714702e-08,
+      "loss": 0.8243,
+      "step": 9229
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3982393209021515e-08,
+      "loss": 0.84,
+      "step": 9230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3858457818207835e-08,
+      "loss": 1.0527,
+      "step": 9231
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3734842213590015e-08,
+      "loss": 1.0589,
+      "step": 9232
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3611546408451755e-08,
+      "loss": 0.876,
+      "step": 9233
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.34885704160398e-08,
+      "loss": 0.8528,
+      "step": 9234
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.336591424956891e-08,
+      "loss": 0.8423,
+      "step": 9235
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3243577922217875e-08,
+      "loss": 0.76,
+      "step": 9236
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3121561447132178e-08,
+      "loss": 1.0827,
+      "step": 9237
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2999864837422337e-08,
+      "loss": 0.9212,
+      "step": 9238
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2878488106164553e-08,
+      "loss": 0.9119,
+      "step": 9239
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2757431266400397e-08,
+      "loss": 0.9427,
+      "step": 9240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2636694331138463e-08,
+      "loss": 0.8125,
+      "step": 9241
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2516277313350707e-08,
+      "loss": 0.8494,
+      "step": 9242
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2396180225976782e-08,
+      "loss": 0.8656,
+      "step": 9243
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2276403081920692e-08,
+      "loss": 0.8973,
+      "step": 9244
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2156945894052815e-08,
+      "loss": 0.8768,
+      "step": 9245
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.2037808675208215e-08,
+      "loss": 0.7881,
+      "step": 9246
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1918991438188983e-08,
+      "loss": 0.9221,
+      "step": 9247
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1800494195761244e-08,
+      "loss": 1.085,
+      "step": 9248
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1682316960658146e-08,
+      "loss": 0.7009,
+      "step": 9249
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1564459745577526e-08,
+      "loss": 0.8416,
+      "step": 9250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1446922563183258e-08,
+      "loss": 0.9352,
+      "step": 9251
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.132970542610424e-08,
+      "loss": 0.7569,
+      "step": 9252
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1212808346936064e-08,
+      "loss": 0.7366,
+      "step": 9253
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1096231338239013e-08,
+      "loss": 0.9904,
+      "step": 9254
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0979974412539405e-08,
+      "loss": 0.9257,
+      "step": 9255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.086403758232891e-08,
+      "loss": 0.8063,
+      "step": 9256
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.07484208600649e-08,
+      "loss": 1.0207,
+      "step": 9257
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0633124258170765e-08,
+      "loss": 0.9862,
+      "step": 9258
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0518147789034602e-08,
+      "loss": 0.775,
+      "step": 9259
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.040349146501086e-08,
+      "loss": 0.8831,
+      "step": 9260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.028915529841935e-08,
+      "loss": 0.7759,
+      "step": 9261
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0175139301545574e-08,
+      "loss": 0.8699,
+      "step": 9262
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.006144348664041e-08,
+      "loss": 0.937,
+      "step": 9263
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9948067865920405e-08,
+      "loss": 0.9638,
+      "step": 9264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9835012451567825e-08,
+      "loss": 0.9003,
+      "step": 9265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9722277255730614e-08,
+      "loss": 1.0386,
+      "step": 9266
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9609862290522084e-08,
+      "loss": 0.8489,
+      "step": 9267
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.949776756802091e-08,
+      "loss": 0.8438,
+      "step": 9268
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.938599310027245e-08,
+      "loss": 1.1895,
+      "step": 9269
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9274538899285765e-08,
+      "loss": 0.8883,
+      "step": 9270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9163404977037613e-08,
+      "loss": 0.8033,
+      "step": 9271
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.905259134546844e-08,
+      "loss": 0.7185,
+      "step": 9272
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8942098016486054e-08,
+      "loss": 0.9923,
+      "step": 9273
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8831925001962293e-08,
+      "loss": 0.7994,
+      "step": 9274
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8722072313735684e-08,
+      "loss": 0.8793,
+      "step": 9275
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8612539963609454e-08,
+      "loss": 0.9501,
+      "step": 9276
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8503327963353188e-08,
+      "loss": 0.9242,
+      "step": 9277
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.839443632470117e-08,
+      "loss": 0.9017,
+      "step": 9278
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.82858650593547e-08,
+      "loss": 0.9541,
+      "step": 9279
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.817761417897912e-08,
+      "loss": 0.9085,
+      "step": 9280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8069683695206117e-08,
+      "loss": 0.7357,
+      "step": 9281
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7962073619632757e-08,
+      "loss": 0.9553,
+      "step": 9282
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7854783963821784e-08,
+      "loss": 0.7799,
+      "step": 9283
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7747814739301316e-08,
+      "loss": 0.762,
+      "step": 9284
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7641165957565485e-08,
+      "loss": 0.9379,
+      "step": 9285
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7534837630073798e-08,
+      "loss": 0.8686,
+      "step": 9286
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7428829768250775e-08,
+      "loss": 0.7248,
+      "step": 9287
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7323142383487312e-08,
+      "loss": 0.8154,
+      "step": 9288
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7217775487138988e-08,
+      "loss": 0.6687,
+      "step": 9289
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7112729090528078e-08,
+      "loss": 1.3784,
+      "step": 9290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7008003204941557e-08,
+      "loss": 0.929,
+      "step": 9291
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.690359784163242e-08,
+      "loss": 0.8915,
+      "step": 9292
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6799513011818368e-08,
+      "loss": 1.1319,
+      "step": 9293
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6695748726684112e-08,
+      "loss": 1.0473,
+      "step": 9294
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.659230499737874e-08,
+      "loss": 0.8772,
+      "step": 9295
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6489181835017353e-08,
+      "loss": 0.8931,
+      "step": 9296
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6386379250680427e-08,
+      "loss": 0.9262,
+      "step": 9297
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6283897255414125e-08,
+      "loss": 1.0051,
+      "step": 9298
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6181735860229973e-08,
+      "loss": 1.0091,
+      "step": 9299
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6079895076105522e-08,
+      "loss": 0.9767,
+      "step": 9300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5978374913983685e-08,
+      "loss": 0.8504,
+      "step": 9301
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5877175384772068e-08,
+      "loss": 0.8838,
+      "step": 9302
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5776296499345312e-08,
+      "loss": 1.0039,
+      "step": 9303
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5675738268542406e-08,
+      "loss": 0.7915,
+      "step": 9304
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5575500703168712e-08,
+      "loss": 1.1115,
+      "step": 9305
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5475583813994277e-08,
+      "loss": 0.8267,
+      "step": 9306
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5375987611755184e-08,
+      "loss": 0.7797,
+      "step": 9307
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5276712107153535e-08,
+      "loss": 0.9741,
+      "step": 9308
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5177757310856132e-08,
+      "loss": 0.8075,
+      "step": 9309
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.5079123233495807e-08,
+      "loss": 0.9932,
+      "step": 9310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.498080988567041e-08,
+      "loss": 0.8655,
+      "step": 9311
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4882817277944494e-08,
+      "loss": 0.7531,
+      "step": 9312
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4785145420846303e-08,
+      "loss": 0.8019,
+      "step": 9313
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.468779432487144e-08,
+      "loss": 0.9279,
+      "step": 9314
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4590764000480205e-08,
+      "loss": 1.1701,
+      "step": 9315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4494054458098261e-08,
+      "loss": 0.7454,
+      "step": 9316
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4397665708116958e-08,
+      "loss": 0.7786,
+      "step": 9317
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.430159776089368e-08,
+      "loss": 0.9736,
+      "step": 9318
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4205850626750505e-08,
+      "loss": 0.9243,
+      "step": 9319
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4110424315975534e-08,
+      "loss": 1.0202,
+      "step": 9320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4015318838822566e-08,
+      "loss": 1.0844,
+      "step": 9321
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3920534205510427e-08,
+      "loss": 1.0493,
+      "step": 9322
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3826070426223636e-08,
+      "loss": 0.9519,
+      "step": 9323
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.373192751111274e-08,
+      "loss": 0.9299,
+      "step": 9324
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3638105470293315e-08,
+      "loss": 0.9274,
+      "step": 9325
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3544604313845966e-08,
+      "loss": 0.9223,
+      "step": 9326
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3451424051817984e-08,
+      "loss": 1.1522,
+      "step": 9327
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3358564694221365e-08,
+      "loss": 0.9063,
+      "step": 9328
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3266026251033791e-08,
+      "loss": 0.9807,
+      "step": 9329
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3173808732198645e-08,
+      "loss": 0.8201,
+      "step": 9330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3081912147624331e-08,
+      "loss": 0.8861,
+      "step": 9331
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2990336507185618e-08,
+      "loss": 0.8967,
+      "step": 9332
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2899081820722302e-08,
+      "loss": 0.9104,
+      "step": 9333
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2808148098039207e-08,
+      "loss": 0.9634,
+      "step": 9334
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2717535348907516e-08,
+      "loss": 0.7919,
+      "step": 9335
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2627243583063442e-08,
+      "loss": 1.0665,
+      "step": 9336
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2537272810208889e-08,
+      "loss": 0.8108,
+      "step": 9337
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2447623040011458e-08,
+      "loss": 0.8049,
+      "step": 9338
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2358294282103445e-08,
+      "loss": 0.789,
+      "step": 9339
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2269286546083502e-08,
+      "loss": 0.8283,
+      "step": 9340
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2180599841515649e-08,
+      "loss": 0.9602,
+      "step": 9341
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2092234177928929e-08,
+      "loss": 1.1291,
+      "step": 9342
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2004189564818747e-08,
+      "loss": 0.9021,
+      "step": 9343
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.19164660116452e-08,
+      "loss": 0.8367,
+      "step": 9344
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.182906352783375e-08,
+      "loss": 0.8584,
+      "step": 9345
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1741982122776552e-08,
+      "loss": 1.0024,
+      "step": 9346
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1655221805829786e-08,
+      "loss": 1.0859,
+      "step": 9347
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1568782586316662e-08,
+      "loss": 0.8683,
+      "step": 9348
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1482664473524418e-08,
+      "loss": 1.0169,
+      "step": 9349
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1396867476706318e-08,
+      "loss": 0.7798,
+      "step": 9350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.131139160508199e-08,
+      "loss": 0.811,
+      "step": 9351
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1226236867835082e-08,
+      "loss": 0.8437,
+      "step": 9352
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1141403274115947e-08,
+      "loss": 0.9147,
+      "step": 9353
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1056890833039623e-08,
+      "loss": 0.8451,
+      "step": 9354
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.097269955368685e-08,
+      "loss": 0.9183,
+      "step": 9355
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.088882944510472e-08,
+      "loss": 0.8147,
+      "step": 9356
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0805280516304362e-08,
+      "loss": 1.1017,
+      "step": 9357
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0722052776263259e-08,
+      "loss": 0.9082,
+      "step": 9358
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0639146233924257e-08,
+      "loss": 0.9572,
+      "step": 9359
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0556560898195566e-08,
+      "loss": 0.8201,
+      "step": 9360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.047429677795142e-08,
+      "loss": 0.6659,
+      "step": 9361
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0392353882030414e-08,
+      "loss": 1.0687,
+      "step": 9362
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.031073221923784e-08,
+      "loss": 0.6886,
+      "step": 9363
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.022943179834368e-08,
+      "loss": 0.6952,
+      "step": 9364
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0148452628083948e-08,
+      "loss": 1.0174,
+      "step": 9365
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0067794717159351e-08,
+      "loss": 0.909,
+      "step": 9366
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.987458074236954e-09,
+      "loss": 0.8347,
+      "step": 9367
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.907442707948854e-09,
+      "loss": 0.9306,
+      "step": 9368
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.827748626892508e-09,
+      "loss": 0.8206,
+      "step": 9369
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.748375839631396e-09,
+      "loss": 0.9232,
+      "step": 9370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.669324354693698e-09,
+      "loss": 1.0115,
+      "step": 9371
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.590594180573952e-09,
+      "loss": 0.8248,
+      "step": 9372
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.512185325731392e-09,
+      "loss": 1.0107,
+      "step": 9373
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.434097798590946e-09,
+      "loss": 0.7921,
+      "step": 9374
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.356331607543567e-09,
+      "loss": 1.0732,
+      "step": 9375
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.278886760944904e-09,
+      "loss": 0.8632,
+      "step": 9376
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.201763267116303e-09,
+      "loss": 0.8779,
+      "step": 9377
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.1249611343448e-09,
+      "loss": 1.0984,
+      "step": 9378
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.048480370882794e-09,
+      "loss": 0.7815,
+      "step": 9379
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.972320984948379e-09,
+      "loss": 0.9998,
+      "step": 9380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.89648298472434e-09,
+      "loss": 0.9949,
+      "step": 9381
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.820966378360162e-09,
+      "loss": 0.8523,
+      "step": 9382
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.745771173969352e-09,
+      "loss": 0.6665,
+      "step": 9383
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.670897379632114e-09,
+      "loss": 0.8034,
+      "step": 9384
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.596345003393347e-09,
+      "loss": 0.8166,
+      "step": 9385
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.522114053263975e-09,
+      "loss": 0.9982,
+      "step": 9386
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.44820453721995e-09,
+      "loss": 0.8456,
+      "step": 9387
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.374616463202922e-09,
+      "loss": 0.8333,
+      "step": 9388
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.301349839119899e-09,
+      "loss": 0.8086,
+      "step": 9389
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.228404672843248e-09,
+      "loss": 1.0816,
+      "step": 9390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.155780972211036e-09,
+      "loss": 0.9094,
+      "step": 9391
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.083478745027018e-09,
+      "loss": 0.7904,
+      "step": 9392
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.011497999059313e-09,
+      "loss": 0.8643,
+      "step": 9393
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.939838742042737e-09,
+      "loss": 0.9973,
+      "step": 9394
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.868500981676795e-09,
+      "loss": 0.8734,
+      "step": 9395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.797484725627357e-09,
+      "loss": 1.0016,
+      "step": 9396
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.726789981524318e-09,
+      "loss": 1.1504,
+      "step": 9397
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.656416756964269e-09,
+      "loss": 0.8022,
+      "step": 9398
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.586365059508493e-09,
+      "loss": 0.7792,
+      "step": 9399
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.516634896684638e-09,
+      "loss": 0.9681,
+      "step": 9400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.447226275984709e-09,
+      "loss": 0.699,
+      "step": 9401
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.378139204866741e-09,
+      "loss": 1.1605,
+      "step": 9402
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.309373690754129e-09,
+      "loss": 1.1659,
+      "step": 9403
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.240929741035629e-09,
+      "loss": 0.7989,
+      "step": 9404
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.172807363066025e-09,
+      "loss": 0.9939,
+      "step": 9405
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.105006564164462e-09,
+      "loss": 1.0761,
+      "step": 9406
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.037527351616446e-09,
+      "loss": 0.7914,
+      "step": 9407
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.97036973267251e-09,
+      "loss": 0.9987,
+      "step": 9408
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.903533714548549e-09,
+      "loss": 1.0748,
+      "step": 9409
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.837019304426484e-09,
+      "loss": 0.8787,
+      "step": 9410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.770826509453265e-09,
+      "loss": 0.7561,
+      "step": 9411
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.704955336741203e-09,
+      "loss": 0.8824,
+      "step": 9412
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.63940579336797e-09,
+      "loss": 0.8156,
+      "step": 9413
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.574177886376598e-09,
+      "loss": 0.9266,
+      "step": 9414
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.509271622776813e-09,
+      "loss": 0.8748,
+      "step": 9415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.444687009541706e-09,
+      "loss": 0.9234,
+      "step": 9416
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.380424053611722e-09,
+      "loss": 0.8841,
+      "step": 9417
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.31648276189134e-09,
+      "loss": 0.9828,
+      "step": 9418
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.252863141251397e-09,
+      "loss": 0.9662,
+      "step": 9419
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.189565198527425e-09,
+      "loss": 1.2253,
+      "step": 9420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.126588940521316e-09,
+      "loss": 0.8046,
+      "step": 9421
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.063934373999325e-09,
+      "loss": 1.2102,
+      "step": 9422
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.001601505694398e-09,
+      "loss": 0.9985,
+      "step": 9423
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9395903423035136e-09,
+      "loss": 1.0472,
+      "step": 9424
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.877900890489673e-09,
+      "loss": 0.9826,
+      "step": 9425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.816533156881909e-09,
+      "loss": 0.9709,
+      "step": 9426
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.755487148074279e-09,
+      "loss": 0.9924,
+      "step": 9427
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694762870625536e-09,
+      "loss": 0.9969,
+      "step": 9428
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.634360331061128e-09,
+      "loss": 0.973,
+      "step": 9429
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.574279535870863e-09,
+      "loss": 0.7249,
+      "step": 9430
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.514520491510577e-09,
+      "loss": 0.8367,
+      "step": 9431
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.455083204401468e-09,
+      "loss": 0.9923,
+      "step": 9432
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.395967680930092e-09,
+      "loss": 0.9364,
+      "step": 9433
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.337173927448036e-09,
+      "loss": 0.6087,
+      "step": 9434
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.278701950272913e-09,
+      "loss": 1.1157,
+      "step": 9435
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.22055175568803e-09,
+      "loss": 0.9118,
+      "step": 9436
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.162723349940723e-09,
+      "loss": 1.0963,
+      "step": 9437
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1052167392453555e-09,
+      "loss": 0.722,
+      "step": 9438
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.048031929780317e-09,
+      "loss": 1.0757,
+      "step": 9439
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.991168927691026e-09,
+      "loss": 1.169,
+      "step": 9440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.934627739086595e-09,
+      "loss": 1.0266,
+      "step": 9441
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8784083700428305e-09,
+      "loss": 0.8524,
+      "step": 9442
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.822510826600235e-09,
+      "loss": 0.9018,
+      "step": 9443
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.766935114765336e-09,
+      "loss": 0.912,
+      "step": 9444
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.711681240509358e-09,
+      "loss": 0.8718,
+      "step": 9445
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.656749209769551e-09,
+      "loss": 0.8661,
+      "step": 9446
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.602139028448193e-09,
+      "loss": 0.8341,
+      "step": 9447
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.547850702413259e-09,
+      "loss": 0.8357,
+      "step": 9448
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.493884237498081e-09,
+      "loss": 1.2808,
+      "step": 9449
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.440239639501353e-09,
+      "loss": 0.9535,
+      "step": 9450
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.386916914187133e-09,
+      "loss": 0.9033,
+      "step": 9451
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.333916067284838e-09,
+      "loss": 0.8978,
+      "step": 9452
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.281237104489577e-09,
+      "loss": 0.8539,
+      "step": 9453
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.22888003146149e-09,
+      "loss": 0.753,
+      "step": 9454
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.176844853826411e-09,
+      "loss": 1.0204,
+      "step": 9455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.125131577175867e-09,
+      "loss": 0.8878,
+      "step": 9456
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.073740207066079e-09,
+      "loss": 0.855,
+      "step": 9457
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.022670749018964e-09,
+      "loss": 0.6972,
+      "step": 9458
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.971923208522133e-09,
+      "loss": 0.8768,
+      "step": 9459
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.921497591028555e-09,
+      "loss": 1.011,
+      "step": 9460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.871393901955899e-09,
+      "loss": 1.0701,
+      "step": 9461
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.8216121466885246e-09,
+      "loss": 0.8742,
+      "step": 9462
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7721523305748184e-09,
+      "loss": 0.9479,
+      "step": 9463
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.723014458929863e-09,
+      "loss": 0.8709,
+      "step": 9464
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.6741985370327692e-09,
+      "loss": 1.1206,
+      "step": 9465
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.6257045701296733e-09,
+      "loss": 0.9244,
+      "step": 9466
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.5775325634304077e-09,
+      "loss": 0.699,
+      "step": 9467
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.5296825221118324e-09,
+      "loss": 0.9128,
+      "step": 9468
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.482154451314834e-09,
+      "loss": 1.1884,
+      "step": 9469
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.434948356146328e-09,
+      "loss": 1.0082,
+      "step": 9470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.3880642416792563e-09,
+      "loss": 0.9822,
+      "step": 9471
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.3415021129505894e-09,
+      "loss": 0.8013,
+      "step": 9472
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 9616,
+  "num_train_epochs": 2,
+  "save_steps": 256,
+  "total_flos": 1.101418869958692e+19,
+  "trial_name": null,
+  "trial_params": null
+}