diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,43198 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999305024671624,
+  "global_step": 7194,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998609952738393e-05,
+      "loss": 13.6974,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.998609952738393e-05,
+      "loss": 13.1563,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997219905476786e-05,
+      "loss": 12.9935,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99582985821518e-05,
+      "loss": 13.5701,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.994439810953573e-05,
+      "loss": 11.9934,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.993049763691967e-05,
+      "loss": 10.7499,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99165971643036e-05,
+      "loss": 9.5844,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.990269669168753e-05,
+      "loss": 9.4138,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.988879621907146e-05,
+      "loss": 7.9144,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.987489574645539e-05,
+      "loss": 7.5599,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.986099527383932e-05,
+      "loss": 7.0566,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.984709480122325e-05,
+      "loss": 6.8375,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.983319432860718e-05,
+      "loss": 6.4292,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.981929385599111e-05,
+      "loss": 6.1716,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.980539338337504e-05,
+      "loss": 5.9142,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.979149291075897e-05,
+      "loss": 5.7691,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.97775924381429e-05,
+      "loss": 5.6627,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.976369196552683e-05,
+      "loss": 5.417,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.974979149291076e-05,
+      "loss": 5.2857,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.973589102029469e-05,
+      "loss": 5.0671,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.972199054767862e-05,
+      "loss": 5.0442,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.970809007506255e-05,
+      "loss": 4.9064,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.969418960244648e-05,
+      "loss": 4.9802,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.968028912983042e-05,
+      "loss": 4.8083,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.966638865721435e-05,
+      "loss": 4.7511,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.965248818459828e-05,
+      "loss": 4.6491,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.963858771198221e-05,
+      "loss": 4.631,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.962468723936614e-05,
+      "loss": 4.5066,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.961078676675007e-05,
+      "loss": 4.4718,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9596886294134e-05,
+      "loss": 4.3555,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.958298582151793e-05,
+      "loss": 4.4173,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.956908534890186e-05,
+      "loss": 4.3991,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.955518487628581e-05,
+      "loss": 4.3836,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.954128440366974e-05,
+      "loss": 4.274,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.952738393105367e-05,
+      "loss": 4.2836,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.95134834584376e-05,
+      "loss": 4.236,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.949958298582153e-05,
+      "loss": 4.187,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.948568251320546e-05,
+      "loss": 4.1469,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.947178204058939e-05,
+      "loss": 4.1026,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.945788156797332e-05,
+      "loss": 4.0298,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944398109535725e-05,
+      "loss": 3.9922,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.943008062274118e-05,
+      "loss": 4.0254,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.941618015012511e-05,
+      "loss": 3.8997,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.940227967750904e-05,
+      "loss": 3.9286,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.938837920489297e-05,
+      "loss": 3.9405,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.93744787322769e-05,
+      "loss": 3.8761,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.936057825966083e-05,
+      "loss": 3.8862,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.934667778704476e-05,
+      "loss": 3.8217,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.933277731442869e-05,
+      "loss": 3.7614,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.931887684181262e-05,
+      "loss": 3.7561,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.930497636919656e-05,
+      "loss": 3.7029,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.929107589658049e-05,
+      "loss": 3.7194,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.927717542396442e-05,
+      "loss": 3.7607,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.926327495134835e-05,
+      "loss": 3.664,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.924937447873228e-05,
+      "loss": 3.6929,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.923547400611621e-05,
+      "loss": 3.6286,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.922157353350014e-05,
+      "loss": 3.5846,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.920767306088407e-05,
+      "loss": 3.5923,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9193772588268e-05,
+      "loss": 3.5427,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.917987211565193e-05,
+      "loss": 3.516,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.916597164303588e-05,
+      "loss": 3.5374,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.91520711704198e-05,
+      "loss": 3.5246,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.913817069780374e-05,
+      "loss": 3.5253,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.912427022518767e-05,
+      "loss": 3.4852,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.91103697525716e-05,
+      "loss": 3.4818,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.909646927995553e-05,
+      "loss": 3.4612,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.908256880733946e-05,
+      "loss": 3.4454,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.906866833472339e-05,
+      "loss": 3.4039,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.905476786210732e-05,
+      "loss": 3.406,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.904086738949125e-05,
+      "loss": 3.4099,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.902696691687518e-05,
+      "loss": 3.3857,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.901306644425911e-05,
+      "loss": 3.3768,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.899916597164304e-05,
+      "loss": 3.3559,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.898526549902697e-05,
+      "loss": 3.3653,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.89713650264109e-05,
+      "loss": 3.3531,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.895746455379483e-05,
+      "loss": 3.295,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.894356408117876e-05,
+      "loss": 3.2913,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.89296636085627e-05,
+      "loss": 3.298,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.891576313594663e-05,
+      "loss": 3.3089,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.890186266333056e-05,
+      "loss": 3.3401,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.888796219071449e-05,
+      "loss": 3.2666,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.887406171809842e-05,
+      "loss": 3.2675,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.886016124548235e-05,
+      "loss": 3.2672,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.884626077286628e-05,
+      "loss": 3.2524,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.883236030025021e-05,
+      "loss": 3.2468,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.881845982763414e-05,
+      "loss": 3.223,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.880455935501807e-05,
+      "loss": 3.2406,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.879065888240202e-05,
+      "loss": 3.1804,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.877675840978595e-05,
+      "loss": 3.2105,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.876285793716988e-05,
+      "loss": 3.2105,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.87489574645538e-05,
+      "loss": 3.1587,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.873505699193774e-05,
+      "loss": 3.1657,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.872115651932167e-05,
+      "loss": 3.1476,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.870725604670558e-05,
+      "loss": 3.1739,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.869335557408951e-05,
+      "loss": 3.1499,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.867945510147346e-05,
+      "loss": 3.1557,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.866555462885739e-05,
+      "loss": 3.1287,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.865165415624132e-05,
+      "loss": 3.146,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.863775368362525e-05,
+      "loss": 3.1196,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.862385321100918e-05,
+      "loss": 3.1288,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.86099527383931e-05,
+      "loss": 3.1102,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859605226577704e-05,
+      "loss": 3.11,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.858215179316097e-05,
+      "loss": 3.1094,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.85682513205449e-05,
+      "loss": 3.0835,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.855435084792884e-05,
+      "loss": 3.0989,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.854045037531277e-05,
+      "loss": 3.0969,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.85265499026967e-05,
+      "loss": 3.0879,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.851264943008063e-05,
+      "loss": 3.0584,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.849874895746456e-05,
+      "loss": 3.0783,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.848484848484849e-05,
+      "loss": 3.0509,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.847094801223242e-05,
+      "loss": 3.0639,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.845704753961635e-05,
+      "loss": 3.0746,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.844314706700028e-05,
+      "loss": 3.0454,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.842924659438421e-05,
+      "loss": 3.0485,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.841534612176815e-05,
+      "loss": 3.0427,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.840144564915208e-05,
+      "loss": 3.0454,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.838754517653601e-05,
+      "loss": 3.0375,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.837364470391994e-05,
+      "loss": 3.0491,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.835974423130387e-05,
+      "loss": 3.0316,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.83458437586878e-05,
+      "loss": 3.0412,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.833194328607172e-05,
+      "loss": 3.0218,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.831804281345565e-05,
+      "loss": 3.0309,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.83041423408396e-05,
+      "loss": 3.0286,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.829024186822352e-05,
+      "loss": 3.0351,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.827634139560745e-05,
+      "loss": 3.0113,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.826244092299138e-05,
+      "loss": 3.0332,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.824854045037531e-05,
+      "loss": 3.0351,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.823463997775924e-05,
+      "loss": 3.0347,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.822073950514317e-05,
+      "loss": 3.0452,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.82068390325271e-05,
+      "loss": 3.0252,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.819293855991104e-05,
+      "loss": 3.0255,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.817903808729498e-05,
+      "loss": 3.0349,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.816513761467891e-05,
+      "loss": 3.0192,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.815123714206284e-05,
+      "loss": 3.017,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.813733666944677e-05,
+      "loss": 3.0133,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.81234361968307e-05,
+      "loss": 3.0375,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.810953572421463e-05,
+      "loss": 3.0091,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.809563525159856e-05,
+      "loss": 3.0075,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.808173477898249e-05,
+      "loss": 3.0079,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.806783430636642e-05,
+      "loss": 3.0129,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.805393383375035e-05,
+      "loss": 3.0045,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.804003336113429e-05,
+      "loss": 3.0128,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.802613288851822e-05,
+      "loss": 3.0098,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.801223241590215e-05,
+      "loss": 3.0088,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.799833194328608e-05,
+      "loss": 2.999,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.798443147067001e-05,
+      "loss": 3.0072,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.797053099805393e-05,
+      "loss": 2.9876,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.795663052543786e-05,
+      "loss": 3.0147,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.794273005282179e-05,
+      "loss": 2.9873,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.792882958020573e-05,
+      "loss": 2.9863,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.791492910758966e-05,
+      "loss": 3.0006,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.79010286349736e-05,
+      "loss": 2.9903,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.788712816235752e-05,
+      "loss": 3.013,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.787322768974145e-05,
+      "loss": 2.9939,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.785932721712538e-05,
+      "loss": 3.0113,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.784542674450931e-05,
+      "loss": 3.0014,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.783152627189324e-05,
+      "loss": 2.9839,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.781762579927717e-05,
+      "loss": 2.9914,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.780372532666112e-05,
+      "loss": 2.9997,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.778982485404505e-05,
+      "loss": 3.0065,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.777592438142898e-05,
+      "loss": 2.9895,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.776202390881291e-05,
+      "loss": 3.01,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.774812343619684e-05,
+      "loss": 2.992,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.773422296358077e-05,
+      "loss": 3.0055,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.77203224909647e-05,
+      "loss": 2.9811,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.770642201834863e-05,
+      "loss": 2.983,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.769252154573256e-05,
+      "loss": 2.9912,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.767862107311649e-05,
+      "loss": 2.971,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.766472060050043e-05,
+      "loss": 2.9775,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.765082012788436e-05,
+      "loss": 3.011,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.763691965526829e-05,
+      "loss": 2.9865,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.762301918265222e-05,
+      "loss": 2.9976,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.760911871003614e-05,
+      "loss": 2.9951,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.759521823742007e-05,
+      "loss": 2.9778,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.7581317764804e-05,
+      "loss": 2.9588,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.756741729218793e-05,
+      "loss": 2.9928,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.755351681957187e-05,
+      "loss": 2.9753,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.75396163469558e-05,
+      "loss": 2.9734,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.752571587433973e-05,
+      "loss": 2.9742,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.751181540172366e-05,
+      "loss": 2.9817,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.749791492910759e-05,
+      "loss": 2.9667,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.748401445649152e-05,
+      "loss": 2.9646,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.747011398387545e-05,
+      "loss": 3.0162,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.745621351125938e-05,
+      "loss": 2.9624,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.744231303864331e-05,
+      "loss": 2.9985,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.742841256602724e-05,
+      "loss": 2.9847,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.741451209341119e-05,
+      "loss": 2.9749,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.740061162079512e-05,
+      "loss": 2.9813,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.738671114817905e-05,
+      "loss": 2.9873,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.737281067556298e-05,
+      "loss": 2.9715,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.73589102029469e-05,
+      "loss": 2.9737,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.734500973033084e-05,
+      "loss": 2.9575,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.733110925771477e-05,
+      "loss": 2.9436,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.73172087850987e-05,
+      "loss": 2.975,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.730330831248263e-05,
+      "loss": 2.9704,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.728940783986657e-05,
+      "loss": 2.9797,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.72755073672505e-05,
+      "loss": 2.9773,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.726160689463443e-05,
+      "loss": 3.0058,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.724770642201836e-05,
+      "loss": 2.9942,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.723380594940228e-05,
+      "loss": 2.9782,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.721990547678621e-05,
+      "loss": 2.9695,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.720600500417014e-05,
+      "loss": 2.9695,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.719210453155407e-05,
+      "loss": 2.9679,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.717820405893801e-05,
+      "loss": 2.9813,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.716430358632194e-05,
+      "loss": 2.9856,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.715040311370587e-05,
+      "loss": 2.9665,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.71365026410898e-05,
+      "loss": 2.9527,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.712260216847373e-05,
+      "loss": 2.9641,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.710870169585766e-05,
+      "loss": 2.9521,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.709480122324159e-05,
+      "loss": 2.9655,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.708090075062552e-05,
+      "loss": 2.9619,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.706700027800945e-05,
+      "loss": 2.9562,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.705309980539338e-05,
+      "loss": 2.9595,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.703919933277733e-05,
+      "loss": 2.9693,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.702529886016126e-05,
+      "loss": 2.9827,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.701139838754519e-05,
+      "loss": 2.9504,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.699749791492912e-05,
+      "loss": 2.9558,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.698359744231305e-05,
+      "loss": 2.9621,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.696969696969698e-05,
+      "loss": 2.9565,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.69557964970809e-05,
+      "loss": 2.9652,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.694189602446484e-05,
+      "loss": 2.957,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.692799555184877e-05,
+      "loss": 2.9544,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.691409507923271e-05,
+      "loss": 2.9699,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.690019460661664e-05,
+      "loss": 2.9645,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.688629413400057e-05,
+      "loss": 2.9701,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.687239366138449e-05,
+      "loss": 2.939,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.685849318876842e-05,
+      "loss": 2.9569,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.684459271615235e-05,
+      "loss": 2.9561,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.683069224353628e-05,
+      "loss": 2.9705,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.68167917709202e-05,
+      "loss": 2.9468,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.680289129830415e-05,
+      "loss": 2.965,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.678899082568808e-05,
+      "loss": 2.9585,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.677509035307201e-05,
+      "loss": 2.9581,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.676118988045594e-05,
+      "loss": 2.945,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.674728940783987e-05,
+      "loss": 2.9374,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.67333889352238e-05,
+      "loss": 2.9657,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.671948846260773e-05,
+      "loss": 2.9614,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.670558798999166e-05,
+      "loss": 2.9614,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.669168751737559e-05,
+      "loss": 2.9377,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.667778704475952e-05,
+      "loss": 2.9766,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.666388657214346e-05,
+      "loss": 2.95,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66499860995274e-05,
+      "loss": 2.9495,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.663608562691132e-05,
+      "loss": 2.9595,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.662218515429525e-05,
+      "loss": 2.9535,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.660828468167918e-05,
+      "loss": 2.9537,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.659438420906311e-05,
+      "loss": 2.9647,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.658048373644704e-05,
+      "loss": 2.9316,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.656658326383097e-05,
+      "loss": 2.9435,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.65526827912149e-05,
+      "loss": 2.9661,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.653878231859885e-05,
+      "loss": 2.9613,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.652488184598278e-05,
+      "loss": 2.9648,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.651098137336671e-05,
+      "loss": 2.9387,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.649708090075062e-05,
+      "loss": 2.9488,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.648318042813455e-05,
+      "loss": 2.9467,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.646927995551848e-05,
+      "loss": 2.9301,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.645537948290242e-05,
+      "loss": 2.9447,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.644147901028635e-05,
+      "loss": 2.9508,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.642757853767029e-05,
+      "loss": 2.9731,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.641367806505422e-05,
+      "loss": 2.9303,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.639977759243815e-05,
+      "loss": 2.9506,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.638587711982208e-05,
+      "loss": 2.9407,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.637197664720601e-05,
+      "loss": 2.9313,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.635807617458994e-05,
+      "loss": 2.9447,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.634417570197387e-05,
+      "loss": 2.9557,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.63302752293578e-05,
+      "loss": 2.9582,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.631637475674173e-05,
+      "loss": 2.9449,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.630247428412566e-05,
+      "loss": 2.9454,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.62885738115096e-05,
+      "loss": 2.9573,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.627467333889353e-05,
+      "loss": 2.9313,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.626077286627746e-05,
+      "loss": 2.9492,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.624687239366139e-05,
+      "loss": 2.9386,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.623297192104532e-05,
+      "loss": 2.9537,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.621907144842925e-05,
+      "loss": 2.9416,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.620517097581318e-05,
+      "loss": 2.9151,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.619127050319711e-05,
+      "loss": 2.912,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.617737003058104e-05,
+      "loss": 2.941,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.616346955796497e-05,
+      "loss": 2.917,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.614956908534892e-05,
+      "loss": 2.9265,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.613566861273283e-05,
+      "loss": 2.9116,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.612176814011676e-05,
+      "loss": 2.9275,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.61078676675007e-05,
+      "loss": 2.9214,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.609396719488462e-05,
+      "loss": 2.9478,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.608006672226855e-05,
+      "loss": 2.8861,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.606616624965248e-05,
+      "loss": 2.9081,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.605226577703643e-05,
+      "loss": 2.8971,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.603836530442036e-05,
+      "loss": 2.8711,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.602446483180429e-05,
+      "loss": 2.8798,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.601056435918822e-05,
+      "loss": 2.8862,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.599666388657215e-05,
+      "loss": 2.8556,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.598276341395608e-05,
+      "loss": 2.8785,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.596886294134001e-05,
+      "loss": 2.8598,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.595496246872394e-05,
+      "loss": 2.83,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.594106199610787e-05,
+      "loss": 2.8443,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.59271615234918e-05,
+      "loss": 2.8576,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.591326105087574e-05,
+      "loss": 2.8145,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.589936057825967e-05,
+      "loss": 2.7916,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.58854601056436e-05,
+      "loss": 2.8059,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.587155963302753e-05,
+      "loss": 2.7995,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.585765916041146e-05,
+      "loss": 2.785,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.584375868779539e-05,
+      "loss": 2.8079,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.582985821517932e-05,
+      "loss": 2.7798,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.581595774256325e-05,
+      "loss": 2.8037,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.580205726994718e-05,
+      "loss": 2.7855,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.578815679733111e-05,
+      "loss": 2.7803,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.577425632471506e-05,
+      "loss": 2.772,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.576035585209897e-05,
+      "loss": 2.785,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.57464553794829e-05,
+      "loss": 2.7739,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.573255490686683e-05,
+      "loss": 2.7464,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.571865443425076e-05,
+      "loss": 2.706,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.570475396163469e-05,
+      "loss": 2.762,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.569085348901862e-05,
+      "loss": 2.7325,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.567695301640255e-05,
+      "loss": 2.7476,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.56630525437865e-05,
+      "loss": 2.7191,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.564915207117043e-05,
+      "loss": 2.7389,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.563525159855436e-05,
+      "loss": 2.7069,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.562135112593829e-05,
+      "loss": 2.7269,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.560745065332222e-05,
+      "loss": 2.6702,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.559355018070615e-05,
+      "loss": 2.6889,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.557964970809008e-05,
+      "loss": 2.6745,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.556574923547401e-05,
+      "loss": 2.6567,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.555184876285794e-05,
+      "loss": 2.6856,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.553794829024188e-05,
+      "loss": 2.6563,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.552404781762581e-05,
+      "loss": 2.6599,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.551014734500974e-05,
+      "loss": 2.6488,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.549624687239367e-05,
+      "loss": 2.6145,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.54823463997776e-05,
+      "loss": 2.6383,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.546844592716153e-05,
+      "loss": 2.6171,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 2.5972,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.544064498192939e-05,
+      "loss": 2.5863,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.542674450931332e-05,
+      "loss": 2.6352,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.541284403669725e-05,
+      "loss": 2.6028,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.539894356408118e-05,
+      "loss": 2.5476,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.538504309146511e-05,
+      "loss": 2.5503,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.537114261884904e-05,
+      "loss": 2.5733,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.535724214623297e-05,
+      "loss": 2.5544,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.53433416736169e-05,
+      "loss": 2.5355,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.532944120100083e-05,
+      "loss": 2.5033,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.531554072838476e-05,
+      "loss": 2.5083,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.530164025576869e-05,
+      "loss": 2.502,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.528773978315264e-05,
+      "loss": 2.5,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.527383931053657e-05,
+      "loss": 2.4737,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.52599388379205e-05,
+      "loss": 2.446,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.524603836530443e-05,
+      "loss": 2.4596,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.523213789268836e-05,
+      "loss": 2.4772,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.521823742007229e-05,
+      "loss": 2.4443,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.520433694745622e-05,
+      "loss": 2.4341,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.519043647484015e-05,
+      "loss": 2.4157,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.517653600222408e-05,
+      "loss": 2.414,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.516263552960802e-05,
+      "loss": 2.3751,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.514873505699195e-05,
+      "loss": 2.4089,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.513483458437588e-05,
+      "loss": 2.3524,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.512093411175981e-05,
+      "loss": 2.3775,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.510703363914374e-05,
+      "loss": 2.3107,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.509313316652767e-05,
+      "loss": 2.3318,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.50792326939116e-05,
+      "loss": 2.3381,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.506533222129553e-05,
+      "loss": 2.3355,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.505143174867946e-05,
+      "loss": 2.3276,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.503753127606339e-05,
+      "loss": 2.2799,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.502363080344732e-05,
+      "loss": 2.3294,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.500973033083125e-05,
+      "loss": 2.2784,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.499582985821518e-05,
+      "loss": 2.2952,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.498192938559911e-05,
+      "loss": 2.2631,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.496802891298304e-05,
+      "loss": 2.2269,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.495412844036697e-05,
+      "loss": 2.2185,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.49402279677509e-05,
+      "loss": 2.2048,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.492632749513483e-05,
+      "loss": 2.1906,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.491242702251877e-05,
+      "loss": 2.1871,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.48985265499027e-05,
+      "loss": 2.146,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.488462607728663e-05,
+      "loss": 2.1369,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.487072560467056e-05,
+      "loss": 2.1342,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.48568251320545e-05,
+      "loss": 2.1313,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.484292465943842e-05,
+      "loss": 2.0866,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.482902418682235e-05,
+      "loss": 2.0832,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.481512371420628e-05,
+      "loss": 2.1002,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.480122324159021e-05,
+      "loss": 2.1132,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.478732276897416e-05,
+      "loss": 2.0869,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.477342229635809e-05,
+      "loss": 2.017,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.475952182374202e-05,
+      "loss": 2.0069,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.474562135112595e-05,
+      "loss": 2.024,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.473172087850988e-05,
+      "loss": 1.9961,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.471782040589381e-05,
+      "loss": 1.9885,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.470391993327774e-05,
+      "loss": 2.0208,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.469001946066167e-05,
+      "loss": 1.9888,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.46761189880456e-05,
+      "loss": 1.9213,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.466221851542953e-05,
+      "loss": 1.9607,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.464831804281346e-05,
+      "loss": 1.9367,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.463441757019739e-05,
+      "loss": 1.9778,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.462051709758132e-05,
+      "loss": 1.8459,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.460661662496525e-05,
+      "loss": 1.9006,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.459271615234918e-05,
+      "loss": 1.8715,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.457881567973311e-05,
+      "loss": 1.8744,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.456491520711704e-05,
+      "loss": 1.8847,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.455101473450097e-05,
+      "loss": 1.8692,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.453711426188491e-05,
+      "loss": 1.9059,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.452321378926884e-05,
+      "loss": 1.8419,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.450931331665277e-05,
+      "loss": 1.8273,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.44954128440367e-05,
+      "loss": 1.8489,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.448151237142063e-05,
+      "loss": 1.8057,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.446761189880456e-05,
+      "loss": 1.8147,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.44537114261885e-05,
+      "loss": 1.7893,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.443981095357242e-05,
+      "loss": 1.8406,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.442591048095635e-05,
+      "loss": 1.7981,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.44120100083403e-05,
+      "loss": 1.7153,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.439810953572423e-05,
+      "loss": 1.7951,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.438420906310816e-05,
+      "loss": 1.7283,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.437030859049209e-05,
+      "loss": 1.7482,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.435640811787602e-05,
+      "loss": 1.7334,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.434250764525995e-05,
+      "loss": 1.7083,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.432860717264388e-05,
+      "loss": 1.7127,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.431470670002781e-05,
+      "loss": 1.7315,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.430080622741174e-05,
+      "loss": 1.7451,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.428690575479567e-05,
+      "loss": 1.7062,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.42730052821796e-05,
+      "loss": 1.7401,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.425910480956353e-05,
+      "loss": 1.7238,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.424520433694746e-05,
+      "loss": 1.7234,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.423130386433139e-05,
+      "loss": 1.6829,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.421740339171532e-05,
+      "loss": 1.7105,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.420350291909925e-05,
+      "loss": 1.676,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.418960244648318e-05,
+      "loss": 1.6836,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.417570197386711e-05,
+      "loss": 1.6915,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.416180150125105e-05,
+      "loss": 1.7117,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.414790102863498e-05,
+      "loss": 1.6225,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.413400055601891e-05,
+      "loss": 1.6249,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.412010008340284e-05,
+      "loss": 1.6381,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.410619961078677e-05,
+      "loss": 1.6388,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.40922991381707e-05,
+      "loss": 1.6437,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.407839866555463e-05,
+      "loss": 1.5909,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.406449819293856e-05,
+      "loss": 1.5616,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.405059772032249e-05,
+      "loss": 1.5738,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.403669724770642e-05,
+      "loss": 1.6379,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.402279677509037e-05,
+      "loss": 1.5626,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.40088963024743e-05,
+      "loss": 1.6233,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.399499582985823e-05,
+      "loss": 1.6404,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.398109535724216e-05,
+      "loss": 1.5127,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.396719488462609e-05,
+      "loss": 1.5251,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.395329441201002e-05,
+      "loss": 1.5675,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.393939393939395e-05,
+      "loss": 1.4916,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.392549346677788e-05,
+      "loss": 1.5439,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.39115929941618e-05,
+      "loss": 1.5056,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.389769252154574e-05,
+      "loss": 1.5409,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.388379204892967e-05,
+      "loss": 1.572,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38698915763136e-05,
+      "loss": 1.5173,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.385599110369753e-05,
+      "loss": 1.5305,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.384209063108146e-05,
+      "loss": 1.5291,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.382819015846539e-05,
+      "loss": 1.5132,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.381428968584932e-05,
+      "loss": 1.5475,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.380038921323325e-05,
+      "loss": 1.4651,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.378648874061719e-05,
+      "loss": 1.5488,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.377258826800112e-05,
+      "loss": 1.5209,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.375868779538505e-05,
+      "loss": 1.5746,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.374478732276898e-05,
+      "loss": 1.4384,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.373088685015291e-05,
+      "loss": 1.513,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.371698637753684e-05,
+      "loss": 1.5119,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.370308590492077e-05,
+      "loss": 1.4522,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.36891854323047e-05,
+      "loss": 1.4947,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.367528495968863e-05,
+      "loss": 1.4903,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.366138448707256e-05,
+      "loss": 1.4839,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.36474840144565e-05,
+      "loss": 1.5111,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.363358354184043e-05,
+      "loss": 1.5329,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.361968306922436e-05,
+      "loss": 1.4653,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.36057825966083e-05,
+      "loss": 1.4176,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.359188212399223e-05,
+      "loss": 1.4303,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.357798165137616e-05,
+      "loss": 1.4215,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.356408117876007e-05,
+      "loss": 1.4203,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3550180706144e-05,
+      "loss": 1.4206,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.353628023352795e-05,
+      "loss": 1.4545,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.352237976091188e-05,
+      "loss": 1.5101,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.35084792882958e-05,
+      "loss": 1.4145,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.349457881567974e-05,
+      "loss": 1.3575,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.348067834306367e-05,
+      "loss": 1.4495,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.34667778704476e-05,
+      "loss": 1.3694,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.345287739783153e-05,
+      "loss": 1.3916,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.343897692521546e-05,
+      "loss": 1.3798,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.342507645259939e-05,
+      "loss": 1.3974,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.341117597998333e-05,
+      "loss": 1.4324,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.339727550736726e-05,
+      "loss": 1.3814,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.338337503475119e-05,
+      "loss": 1.3941,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.336947456213512e-05,
+      "loss": 1.3782,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.335557408951905e-05,
+      "loss": 1.3894,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.334167361690298e-05,
+      "loss": 1.3286,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.332777314428691e-05,
+      "loss": 1.3251,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.331387267167084e-05,
+      "loss": 1.4027,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.329997219905477e-05,
+      "loss": 1.4518,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.32860717264387e-05,
+      "loss": 1.4077,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.327217125382264e-05,
+      "loss": 1.3975,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.325827078120657e-05,
+      "loss": 1.3751,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.32443703085905e-05,
+      "loss": 1.3211,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.323046983597443e-05,
+      "loss": 1.3753,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.321656936335836e-05,
+      "loss": 1.3992,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.32026688907423e-05,
+      "loss": 1.2838,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.318876841812621e-05,
+      "loss": 1.339,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.317486794551014e-05,
+      "loss": 1.3805,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.316096747289408e-05,
+      "loss": 1.3718,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.314706700027801e-05,
+      "loss": 1.3841,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.313316652766194e-05,
+      "loss": 1.33,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.311926605504587e-05,
+      "loss": 1.2915,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.31053655824298e-05,
+      "loss": 1.2903,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.309146510981373e-05,
+      "loss": 1.2818,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.307756463719766e-05,
+      "loss": 1.283,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.30636641645816e-05,
+      "loss": 1.2949,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.304976369196552e-05,
+      "loss": 1.2891,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.303586321934947e-05,
+      "loss": 1.2617,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.30219627467334e-05,
+      "loss": 1.2524,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.300806227411733e-05,
+      "loss": 1.2445,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.299416180150126e-05,
+      "loss": 1.2796,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.298026132888519e-05,
+      "loss": 1.292,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.296636085626912e-05,
+      "loss": 1.2682,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.295246038365305e-05,
+      "loss": 1.2437,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.293855991103698e-05,
+      "loss": 1.2648,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.292465943842091e-05,
+      "loss": 1.1988,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.291075896580484e-05,
+      "loss": 1.1912,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.289685849318878e-05,
+      "loss": 1.2097,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.288295802057271e-05,
+      "loss": 1.2166,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.286905754795664e-05,
+      "loss": 1.2147,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.285515707534057e-05,
+      "loss": 1.25,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.28412566027245e-05,
+      "loss": 1.2327,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.282735613010842e-05,
+      "loss": 1.2249,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.281345565749235e-05,
+      "loss": 1.2475,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.279955518487628e-05,
+      "loss": 1.2433,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.278565471226022e-05,
+      "loss": 1.1933,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.277175423964415e-05,
+      "loss": 1.1309,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.275785376702808e-05,
+      "loss": 1.1808,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.274395329441201e-05,
+      "loss": 1.1227,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.273005282179594e-05,
+      "loss": 1.1504,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.271615234917987e-05,
+      "loss": 1.2867,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.27022518765638e-05,
+      "loss": 1.1577,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.268835140394773e-05,
+      "loss": 1.1227,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.267445093133166e-05,
+      "loss": 1.133,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.266055045871561e-05,
+      "loss": 1.1339,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.264664998609954e-05,
+      "loss": 1.1308,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.263274951348347e-05,
+      "loss": 1.0311,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.26188490408674e-05,
+      "loss": 1.1247,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.260494856825133e-05,
+      "loss": 1.0952,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.259104809563526e-05,
+      "loss": 1.1531,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.257714762301919e-05,
+      "loss": 1.0724,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.256324715040312e-05,
+      "loss": 1.0558,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.254934667778705e-05,
+      "loss": 1.037,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.253544620517098e-05,
+      "loss": 1.0187,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.252154573255492e-05,
+      "loss": 1.038,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.250764525993885e-05,
+      "loss": 1.0385,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.249374478732278e-05,
+      "loss": 1.0179,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.247984431470671e-05,
+      "loss": 1.0233,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.246594384209064e-05,
+      "loss": 1.079,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.245204336947456e-05,
+      "loss": 0.9888,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.243814289685849e-05,
+      "loss": 1.0613,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.242424242424242e-05,
+      "loss": 1.0866,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.241034195162636e-05,
+      "loss": 0.9817,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.239644147901029e-05,
+      "loss": 1.0398,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.238254100639422e-05,
+      "loss": 0.9585,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.236864053377815e-05,
+      "loss": 1.0183,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.235474006116208e-05,
+      "loss": 1.0132,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.234083958854601e-05,
+      "loss": 0.9914,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.232693911592994e-05,
+      "loss": 1.0009,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.231303864331387e-05,
+      "loss": 0.9609,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.22991381706978e-05,
+      "loss": 1.0656,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.228523769808173e-05,
+      "loss": 0.999,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.227133722546568e-05,
+      "loss": 0.9799,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.22574367528496e-05,
+      "loss": 0.9813,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.224353628023354e-05,
+      "loss": 0.9863,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.222963580761747e-05,
+      "loss": 0.9755,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.22157353350014e-05,
+      "loss": 0.9385,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.220183486238533e-05,
+      "loss": 0.9294,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.218793438976926e-05,
+      "loss": 0.8681,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.217403391715319e-05,
+      "loss": 0.9821,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.216013344453712e-05,
+      "loss": 0.8943,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.214623297192106e-05,
+      "loss": 0.9368,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.213233249930499e-05,
+      "loss": 0.9638,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.211843202668892e-05,
+      "loss": 0.9891,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.210453155407285e-05,
+      "loss": 0.9254,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.209063108145677e-05,
+      "loss": 0.9469,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.20767306088407e-05,
+      "loss": 0.9152,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.206283013622463e-05,
+      "loss": 0.9203,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.204892966360856e-05,
+      "loss": 0.9383,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.20350291909925e-05,
+      "loss": 0.8276,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.202112871837643e-05,
+      "loss": 0.8748,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.200722824576036e-05,
+      "loss": 0.9252,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.199332777314429e-05,
+      "loss": 0.9471,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.197942730052822e-05,
+      "loss": 0.9396,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.196552682791215e-05,
+      "loss": 0.8731,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.195162635529608e-05,
+      "loss": 0.9112,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.193772588268001e-05,
+      "loss": 0.8251,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.192382541006394e-05,
+      "loss": 0.8707,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.190992493744787e-05,
+      "loss": 0.8337,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.189602446483181e-05,
+      "loss": 0.8856,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.188212399221574e-05,
+      "loss": 0.8801,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.186822351959967e-05,
+      "loss": 0.9387,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.18543230469836e-05,
+      "loss": 0.8693,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.184042257436754e-05,
+      "loss": 0.8679,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.182652210175147e-05,
+      "loss": 0.8282,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.18126216291354e-05,
+      "loss": 0.8356,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.179872115651933e-05,
+      "loss": 0.8715,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.178482068390326e-05,
+      "loss": 0.8767,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.17709202112872e-05,
+      "loss": 0.8477,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.175701973867113e-05,
+      "loss": 0.8559,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.174311926605506e-05,
+      "loss": 0.7779,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.172921879343899e-05,
+      "loss": 0.874,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.17153183208229e-05,
+      "loss": 0.7699,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.170141784820684e-05,
+      "loss": 0.8325,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.168751737559077e-05,
+      "loss": 0.8471,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.16736169029747e-05,
+      "loss": 0.7941,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.165971643035864e-05,
+      "loss": 0.8067,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.164581595774257e-05,
+      "loss": 0.8635,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.16319154851265e-05,
+      "loss": 0.8206,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.161801501251043e-05,
+      "loss": 0.8695,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.160411453989436e-05,
+      "loss": 0.8736,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.159021406727829e-05,
+      "loss": 0.83,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.157631359466222e-05,
+      "loss": 0.8348,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.156241312204615e-05,
+      "loss": 0.8094,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.154851264943008e-05,
+      "loss": 0.7935,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.153461217681401e-05,
+      "loss": 0.7858,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.152071170419795e-05,
+      "loss": 0.7789,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.150681123158188e-05,
+      "loss": 0.7794,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.149291075896581e-05,
+      "loss": 0.8541,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.147901028634974e-05,
+      "loss": 0.7945,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.146510981373367e-05,
+      "loss": 0.8607,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.14512093411176e-05,
+      "loss": 0.7913,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.143730886850153e-05,
+      "loss": 0.7899,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.142340839588546e-05,
+      "loss": 0.7935,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.14095079232694e-05,
+      "loss": 0.8602,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.139560745065334e-05,
+      "loss": 0.7357,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.138170697803727e-05,
+      "loss": 0.7982,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.13678065054212e-05,
+      "loss": 0.7572,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.135390603280511e-05,
+      "loss": 0.7467,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.134000556018904e-05,
+      "loss": 0.8507,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.132610508757297e-05,
+      "loss": 0.8207,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.13122046149569e-05,
+      "loss": 0.7895,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.129830414234083e-05,
+      "loss": 0.7467,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.128440366972478e-05,
+      "loss": 0.7976,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.127050319710871e-05,
+      "loss": 0.7442,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.125660272449264e-05,
+      "loss": 0.8075,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.124270225187657e-05,
+      "loss": 0.7404,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.12288017792605e-05,
+      "loss": 0.8352,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.121490130664443e-05,
+      "loss": 0.7414,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.120100083402836e-05,
+      "loss": 0.7791,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.118710036141229e-05,
+      "loss": 0.7507,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.117319988879622e-05,
+      "loss": 0.7714,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.115929941618015e-05,
+      "loss": 0.791,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.114539894356409e-05,
+      "loss": 0.7887,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.113149847094802e-05,
+      "loss": 0.8124,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.111759799833195e-05,
+      "loss": 0.7045,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.110369752571588e-05,
+      "loss": 0.7611,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.108979705309981e-05,
+      "loss": 0.7476,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.107589658048374e-05,
+      "loss": 0.7766,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.106199610786767e-05,
+      "loss": 0.7484,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.10480956352516e-05,
+      "loss": 0.7936,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.103419516263553e-05,
+      "loss": 0.8016,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.102029469001946e-05,
+      "loss": 0.6969,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.10063942174034e-05,
+      "loss": 0.8283,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.099249374478732e-05,
+      "loss": 0.6821,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.097859327217125e-05,
+      "loss": 0.7783,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.096469279955518e-05,
+      "loss": 0.7948,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.095079232693911e-05,
+      "loss": 0.7511,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.093689185432304e-05,
+      "loss": 0.7064,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.092299138170697e-05,
+      "loss": 0.7351,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 0.7287,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.089519043647485e-05,
+      "loss": 0.7192,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.088128996385878e-05,
+      "loss": 0.7811,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.086738949124271e-05,
+      "loss": 0.7185,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.085348901862664e-05,
+      "loss": 0.7914,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.083958854601057e-05,
+      "loss": 0.8141,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.08256880733945e-05,
+      "loss": 0.7172,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.081178760077843e-05,
+      "loss": 0.7527,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.079788712816236e-05,
+      "loss": 0.7904,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.078398665554629e-05,
+      "loss": 0.7235,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.077008618293023e-05,
+      "loss": 0.766,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.075618571031416e-05,
+      "loss": 0.7049,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.074228523769809e-05,
+      "loss": 0.6949,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.072838476508202e-05,
+      "loss": 0.749,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.071448429246595e-05,
+      "loss": 0.7904,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.070058381984988e-05,
+      "loss": 0.7537,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.068668334723381e-05,
+      "loss": 0.7346,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.067278287461774e-05,
+      "loss": 0.7331,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.065888240200167e-05,
+      "loss": 0.7208,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.06449819293856e-05,
+      "loss": 0.7428,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.063108145676955e-05,
+      "loss": 0.7558,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.061718098415346e-05,
+      "loss": 0.8074,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.060328051153739e-05,
+      "loss": 0.7449,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.058938003892132e-05,
+      "loss": 0.7048,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.057547956630525e-05,
+      "loss": 0.7205,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.056157909368918e-05,
+      "loss": 0.7819,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.054767862107311e-05,
+      "loss": 0.7425,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.053377814845704e-05,
+      "loss": 0.7488,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.051987767584099e-05,
+      "loss": 0.7009,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.050597720322492e-05,
+      "loss": 0.6309,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.049207673060885e-05,
+      "loss": 0.6313,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.047817625799278e-05,
+      "loss": 0.752,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.04642757853767e-05,
+      "loss": 0.6643,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.045037531276064e-05,
+      "loss": 0.7602,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.043647484014457e-05,
+      "loss": 0.7414,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.04225743675285e-05,
+      "loss": 0.6433,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.040867389491243e-05,
+      "loss": 0.6963,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.039477342229637e-05,
+      "loss": 0.7424,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.03808729496803e-05,
+      "loss": 0.6748,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.036697247706423e-05,
+      "loss": 0.7125,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.035307200444816e-05,
+      "loss": 0.7657,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.033917153183209e-05,
+      "loss": 0.8028,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.032527105921602e-05,
+      "loss": 0.6739,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.031137058659995e-05,
+      "loss": 0.694,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.029747011398388e-05,
+      "loss": 0.7324,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.028356964136781e-05,
+      "loss": 0.6595,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.026966916875174e-05,
+      "loss": 0.7117,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.025576869613567e-05,
+      "loss": 0.6633,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.02418682235196e-05,
+      "loss": 0.675,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.022796775090353e-05,
+      "loss": 0.6274,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.021406727828746e-05,
+      "loss": 0.6591,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.020016680567139e-05,
+      "loss": 0.6924,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.018626633305532e-05,
+      "loss": 0.7227,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.017236586043925e-05,
+      "loss": 0.7074,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.015846538782318e-05,
+      "loss": 0.6539,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.014456491520712e-05,
+      "loss": 0.6674,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.013066444259105e-05,
+      "loss": 0.6664,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.011676396997498e-05,
+      "loss": 0.6913,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.010286349735892e-05,
+      "loss": 0.641,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.008896302474285e-05,
+      "loss": 0.695,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.007506255212678e-05,
+      "loss": 0.6228,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.00611620795107e-05,
+      "loss": 0.6735,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.004726160689464e-05,
+      "loss": 0.665,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.003336113427857e-05,
+      "loss": 0.6561,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.001946066166251e-05,
+      "loss": 0.7481,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.000556018904644e-05,
+      "loss": 0.6715,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.999165971643037e-05,
+      "loss": 0.6493,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.99777592438143e-05,
+      "loss": 0.7064,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.996385877119823e-05,
+      "loss": 0.6607,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.994995829858216e-05,
+      "loss": 0.6244,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.993605782596609e-05,
+      "loss": 0.6544,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.992215735335002e-05,
+      "loss": 0.6686,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.990825688073395e-05,
+      "loss": 0.6762,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.989435640811788e-05,
+      "loss": 0.6739,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.988045593550181e-05,
+      "loss": 0.6849,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.986655546288574e-05,
+      "loss": 0.7118,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.985265499026967e-05,
+      "loss": 0.6669,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.98387545176536e-05,
+      "loss": 0.6348,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.982485404503753e-05,
+      "loss": 0.6245,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.981095357242146e-05,
+      "loss": 0.631,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.979705309980539e-05,
+      "loss": 0.6701,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.978315262718932e-05,
+      "loss": 0.7233,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.976925215457326e-05,
+      "loss": 0.7383,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.97553516819572e-05,
+      "loss": 0.6801,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.974145120934112e-05,
+      "loss": 0.7506,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.972755073672505e-05,
+      "loss": 0.7012,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.971365026410898e-05,
+      "loss": 0.6471,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.969974979149291e-05,
+      "loss": 0.6822,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.968584931887684e-05,
+      "loss": 0.6494,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.967194884626077e-05,
+      "loss": 0.6644,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.96580483736447e-05,
+      "loss": 0.7519,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.964414790102865e-05,
+      "loss": 0.7206,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.963024742841258e-05,
+      "loss": 0.6469,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.961634695579651e-05,
+      "loss": 0.6605,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.960244648318044e-05,
+      "loss": 0.6829,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.958854601056437e-05,
+      "loss": 0.68,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.95746455379483e-05,
+      "loss": 0.6717,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.956074506533223e-05,
+      "loss": 0.6427,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.954684459271616e-05,
+      "loss": 0.6552,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.953294412010009e-05,
+      "loss": 0.5961,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.951904364748402e-05,
+      "loss": 0.7187,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.950514317486795e-05,
+      "loss": 0.6387,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.949124270225188e-05,
+      "loss": 0.6843,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.947734222963581e-05,
+      "loss": 0.6307,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.946344175701974e-05,
+      "loss": 0.6719,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.944954128440367e-05,
+      "loss": 0.6244,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.94356408117876e-05,
+      "loss": 0.6601,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.942174033917153e-05,
+      "loss": 0.7273,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.940783986655546e-05,
+      "loss": 0.6548,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.93939393939394e-05,
+      "loss": 0.6253,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.938003892132333e-05,
+      "loss": 0.6343,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.936613844870726e-05,
+      "loss": 0.6199,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.935223797609119e-05,
+      "loss": 0.709,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.933833750347512e-05,
+      "loss": 0.6393,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.932443703085905e-05,
+      "loss": 0.6687,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.931053655824298e-05,
+      "loss": 0.6673,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.929663608562691e-05,
+      "loss": 0.6817,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.928273561301084e-05,
+      "loss": 0.6239,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.926883514039477e-05,
+      "loss": 0.7012,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.925493466777872e-05,
+      "loss": 0.5898,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.924103419516265e-05,
+      "loss": 0.6481,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.922713372254658e-05,
+      "loss": 0.6226,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.921323324993051e-05,
+      "loss": 0.7248,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.919933277731444e-05,
+      "loss": 0.6369,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.918543230469837e-05,
+      "loss": 0.6113,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.91715318320823e-05,
+      "loss": 0.6789,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.915763135946623e-05,
+      "loss": 0.6169,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.914373088685016e-05,
+      "loss": 0.6689,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.912983041423409e-05,
+      "loss": 0.6204,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.911592994161802e-05,
+      "loss": 0.6465,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.910202946900195e-05,
+      "loss": 0.6211,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.908812899638588e-05,
+      "loss": 0.6679,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.907422852376981e-05,
+      "loss": 0.6409,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.906032805115374e-05,
+      "loss": 0.5891,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.904642757853767e-05,
+      "loss": 0.6075,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.90325271059216e-05,
+      "loss": 0.7299,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.901862663330554e-05,
+      "loss": 0.6408,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.900472616068947e-05,
+      "loss": 0.5937,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.89908256880734e-05,
+      "loss": 0.6346,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.897692521545733e-05,
+      "loss": 0.6132,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.896302474284126e-05,
+      "loss": 0.5891,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.894912427022519e-05,
+      "loss": 0.5715,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.893522379760912e-05,
+      "loss": 0.6573,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.892132332499305e-05,
+      "loss": 0.5747,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.890742285237698e-05,
+      "loss": 0.5944,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.889352237976091e-05,
+      "loss": 0.6551,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.887962190714486e-05,
+      "loss": 0.6685,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.886572143452879e-05,
+      "loss": 0.6644,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.885182096191272e-05,
+      "loss": 0.5938,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.883792048929665e-05,
+      "loss": 0.7087,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.882402001668058e-05,
+      "loss": 0.6179,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.88101195440645e-05,
+      "loss": 0.6329,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.879621907144844e-05,
+      "loss": 0.6313,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.878231859883235e-05,
+      "loss": 0.761,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.87684181262163e-05,
+      "loss": 0.6403,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.875451765360023e-05,
+      "loss": 0.5892,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.874061718098416e-05,
+      "loss": 0.6411,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.872671670836809e-05,
+      "loss": 0.6355,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.871281623575202e-05,
+      "loss": 0.6179,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.869891576313595e-05,
+      "loss": 0.6254,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.868501529051988e-05,
+      "loss": 0.6598,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.86711148179038e-05,
+      "loss": 0.6776,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.865721434528774e-05,
+      "loss": 0.666,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.864331387267168e-05,
+      "loss": 0.6525,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.862941340005561e-05,
+      "loss": 0.6313,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.861551292743954e-05,
+      "loss": 0.5956,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.860161245482347e-05,
+      "loss": 0.6534,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.85877119822074e-05,
+      "loss": 0.5895,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.857381150959133e-05,
+      "loss": 0.6198,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.855991103697526e-05,
+      "loss": 0.6213,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.854601056435919e-05,
+      "loss": 0.5633,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.853211009174312e-05,
+      "loss": 0.6225,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.851820961912705e-05,
+      "loss": 0.6102,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.8504309146511e-05,
+      "loss": 0.581,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.849040867389492e-05,
+      "loss": 0.5997,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.847650820127885e-05,
+      "loss": 0.6256,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.846260772866278e-05,
+      "loss": 0.6038,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.844870725604671e-05,
+      "loss": 0.5998,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.843480678343064e-05,
+      "loss": 0.5748,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.842090631081457e-05,
+      "loss": 0.6328,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.840700583819849e-05,
+      "loss": 0.5943,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.839310536558243e-05,
+      "loss": 0.5241,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.837920489296636e-05,
+      "loss": 0.5876,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.83653044203503e-05,
+      "loss": 0.5406,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.835140394773423e-05,
+      "loss": 0.5872,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.833750347511816e-05,
+      "loss": 0.6344,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.832360300250209e-05,
+      "loss": 0.5814,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.830970252988602e-05,
+      "loss": 0.5934,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.829580205726995e-05,
+      "loss": 0.6088,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.828190158465388e-05,
+      "loss": 0.6277,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.826800111203782e-05,
+      "loss": 0.6227,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.825410063942175e-05,
+      "loss": 0.6288,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.824020016680568e-05,
+      "loss": 0.59,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.822629969418961e-05,
+      "loss": 0.657,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.821239922157354e-05,
+      "loss": 0.6298,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.819849874895747e-05,
+      "loss": 0.5943,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.81845982763414e-05,
+      "loss": 0.6069,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.817069780372533e-05,
+      "loss": 0.5947,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.815679733110926e-05,
+      "loss": 0.6083,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.814289685849319e-05,
+      "loss": 0.6174,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.812899638587713e-05,
+      "loss": 0.6201,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.811509591326106e-05,
+      "loss": 0.6059,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.8101195440645e-05,
+      "loss": 0.5955,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.808729496802892e-05,
+      "loss": 0.6161,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.807339449541285e-05,
+      "loss": 0.658,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.805949402279678e-05,
+      "loss": 0.5905,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.80455935501807e-05,
+      "loss": 0.5885,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.803169307756463e-05,
+      "loss": 0.61,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.801779260494857e-05,
+      "loss": 0.5722,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.80038921323325e-05,
+      "loss": 0.5835,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.798999165971643e-05,
+      "loss": 0.6345,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.797609118710036e-05,
+      "loss": 0.5505,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.79621907144843e-05,
+      "loss": 0.6066,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.794829024186822e-05,
+      "loss": 0.5688,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.793438976925215e-05,
+      "loss": 0.5915,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.792048929663608e-05,
+      "loss": 0.603,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.790658882402001e-05,
+      "loss": 0.5747,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.789268835140396e-05,
+      "loss": 0.568,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.787878787878789e-05,
+      "loss": 0.5704,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.786488740617182e-05,
+      "loss": 0.5936,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.785098693355575e-05,
+      "loss": 0.5666,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.783708646093968e-05,
+      "loss": 0.5812,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.782318598832361e-05,
+      "loss": 0.571,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.780928551570754e-05,
+      "loss": 0.566,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.779538504309147e-05,
+      "loss": 0.5936,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.77814845704754e-05,
+      "loss": 0.6856,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.776758409785933e-05,
+      "loss": 0.5078,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.775368362524327e-05,
+      "loss": 0.5509,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.77397831526272e-05,
+      "loss": 0.5958,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.772588268001113e-05,
+      "loss": 0.5556,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.771198220739506e-05,
+      "loss": 0.5423,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.769808173477899e-05,
+      "loss": 0.5901,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.768418126216292e-05,
+      "loss": 0.5874,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.767028078954684e-05,
+      "loss": 0.6145,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.765638031693077e-05,
+      "loss": 0.5678,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.764247984431471e-05,
+      "loss": 0.62,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.762857937169864e-05,
+      "loss": 0.5891,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.761467889908257e-05,
+      "loss": 0.58,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.76007784264665e-05,
+      "loss": 0.5702,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.758687795385043e-05,
+      "loss": 0.5665,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.757297748123436e-05,
+      "loss": 0.5362,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.755907700861829e-05,
+      "loss": 0.5845,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.754517653600222e-05,
+      "loss": 0.5746,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.753127606338615e-05,
+      "loss": 0.6091,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.751737559077008e-05,
+      "loss": 0.5298,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.750347511815403e-05,
+      "loss": 0.6354,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.748957464553796e-05,
+      "loss": 0.6116,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.747567417292189e-05,
+      "loss": 0.5617,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.746177370030582e-05,
+      "loss": 0.5404,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.744787322768975e-05,
+      "loss": 0.5677,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.743397275507368e-05,
+      "loss": 0.5706,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.742007228245761e-05,
+      "loss": 0.6041,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.740617180984154e-05,
+      "loss": 0.5784,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.739227133722547e-05,
+      "loss": 0.6944,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.737837086460941e-05,
+      "loss": 0.6553,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.736447039199334e-05,
+      "loss": 0.5368,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.735056991937727e-05,
+      "loss": 0.579,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.73366694467612e-05,
+      "loss": 0.5358,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.732276897414513e-05,
+      "loss": 0.5094,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.730886850152905e-05,
+      "loss": 0.5971,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.729496802891298e-05,
+      "loss": 0.5895,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.728106755629691e-05,
+      "loss": 0.6305,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.726716708368085e-05,
+      "loss": 0.5904,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.725326661106478e-05,
+      "loss": 0.5592,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.723936613844871e-05,
+      "loss": 0.6345,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.722546566583264e-05,
+      "loss": 0.5623,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.721156519321657e-05,
+      "loss": 0.5785,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.71976647206005e-05,
+      "loss": 0.5973,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.718376424798443e-05,
+      "loss": 0.5851,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.716986377536836e-05,
+      "loss": 0.5494,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.715596330275229e-05,
+      "loss": 0.5614,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.714206283013622e-05,
+      "loss": 0.588,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.712816235752017e-05,
+      "loss": 0.5863,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.71142618849041e-05,
+      "loss": 0.5477,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.710036141228803e-05,
+      "loss": 0.6276,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.708646093967196e-05,
+      "loss": 0.5372,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.707256046705589e-05,
+      "loss": 0.6143,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.705865999443982e-05,
+      "loss": 0.5814,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.704475952182375e-05,
+      "loss": 0.5403,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.703085904920768e-05,
+      "loss": 0.5672,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.70169585765916e-05,
+      "loss": 0.5397,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.700305810397555e-05,
+      "loss": 0.529,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.698915763135948e-05,
+      "loss": 0.5245,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.697525715874341e-05,
+      "loss": 0.5317,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.696135668612734e-05,
+      "loss": 0.5505,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.694745621351126e-05,
+      "loss": 0.6173,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.693355574089519e-05,
+      "loss": 0.6034,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.691965526827912e-05,
+      "loss": 0.5742,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.690575479566305e-05,
+      "loss": 0.5732,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.689185432304699e-05,
+      "loss": 0.598,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.687795385043092e-05,
+      "loss": 0.5765,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.686405337781485e-05,
+      "loss": 0.5524,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.685015290519878e-05,
+      "loss": 0.5485,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.683625243258271e-05,
+      "loss": 0.5078,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.682235195996664e-05,
+      "loss": 0.5819,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.680845148735057e-05,
+      "loss": 0.5399,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.67945510147345e-05,
+      "loss": 0.5629,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.678065054211843e-05,
+      "loss": 0.5394,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.676675006950236e-05,
+      "loss": 0.5029,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.67528495968863e-05,
+      "loss": 0.5562,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.673894912427023e-05,
+      "loss": 0.6237,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.672504865165416e-05,
+      "loss": 0.5754,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.67111481790381e-05,
+      "loss": 0.4809,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.669724770642202e-05,
+      "loss": 0.5836,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.668334723380595e-05,
+      "loss": 0.5383,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.666944676118988e-05,
+      "loss": 0.4952,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.665554628857381e-05,
+      "loss": 0.5508,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.664164581595774e-05,
+      "loss": 0.578,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.662774534334169e-05,
+      "loss": 0.5965,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.661384487072562e-05,
+      "loss": 0.6146,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.659994439810955e-05,
+      "loss": 0.5959,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.658604392549348e-05,
+      "loss": 0.5539,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.65721434528774e-05,
+      "loss": 0.6393,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.655824298026133e-05,
+      "loss": 0.4721,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.654434250764526e-05,
+      "loss": 0.5027,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.653044203502919e-05,
+      "loss": 0.5766,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.651654156241313e-05,
+      "loss": 0.5494,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.650264108979706e-05,
+      "loss": 0.5313,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.648874061718099e-05,
+      "loss": 0.5936,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.647484014456492e-05,
+      "loss": 0.5462,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.646093967194885e-05,
+      "loss": 0.6192,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.644703919933278e-05,
+      "loss": 0.6535,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.643313872671671e-05,
+      "loss": 0.5153,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.641923825410064e-05,
+      "loss": 0.6091,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.640533778148457e-05,
+      "loss": 0.5041,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.63914373088685e-05,
+      "loss": 0.5014,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.637753683625244e-05,
+      "loss": 0.5538,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.636363636363637e-05,
+      "loss": 0.6158,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.63497358910203e-05,
+      "loss": 0.541,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.633583541840423e-05,
+      "loss": 0.5068,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.632193494578816e-05,
+      "loss": 0.6412,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.63080344731721e-05,
+      "loss": 0.5582,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.629413400055602e-05,
+      "loss": 0.5573,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.628023352793995e-05,
+      "loss": 0.5168,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.626633305532388e-05,
+      "loss": 0.5037,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.625243258270781e-05,
+      "loss": 0.5645,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.623853211009176e-05,
+      "loss": 0.501,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.622463163747569e-05,
+      "loss": 0.621,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.62107311648596e-05,
+      "loss": 0.5546,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.619683069224353e-05,
+      "loss": 0.5279,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.618293021962746e-05,
+      "loss": 0.5184,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.61690297470114e-05,
+      "loss": 0.5552,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.615512927439532e-05,
+      "loss": 0.5089,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.614122880177927e-05,
+      "loss": 0.5314,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.61273283291632e-05,
+      "loss": 0.5644,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.611342785654713e-05,
+      "loss": 0.5206,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.609952738393106e-05,
+      "loss": 0.5676,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.608562691131499e-05,
+      "loss": 0.5285,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.607172643869892e-05,
+      "loss": 0.5081,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.605782596608285e-05,
+      "loss": 0.4736,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.604392549346678e-05,
+      "loss": 0.4774,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.603002502085071e-05,
+      "loss": 0.5806,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.601612454823464e-05,
+      "loss": 0.5092,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.600222407561858e-05,
+      "loss": 0.5178,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.598832360300251e-05,
+      "loss": 0.511,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.597442313038644e-05,
+      "loss": 0.5359,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.596052265777037e-05,
+      "loss": 0.4775,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.59466221851543e-05,
+      "loss": 0.5034,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.593272171253823e-05,
+      "loss": 0.5362,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.591882123992216e-05,
+      "loss": 0.5144,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.590492076730609e-05,
+      "loss": 0.4934,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.589102029469002e-05,
+      "loss": 0.5501,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.587711982207395e-05,
+      "loss": 0.5128,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.58632193494579e-05,
+      "loss": 0.5234,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.584931887684183e-05,
+      "loss": 0.5047,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.583541840422574e-05,
+      "loss": 0.4753,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.582151793160967e-05,
+      "loss": 0.5236,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.58076174589936e-05,
+      "loss": 0.5714,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.579371698637753e-05,
+      "loss": 0.5359,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.577981651376146e-05,
+      "loss": 0.5549,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.576591604114539e-05,
+      "loss": 0.4759,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.575201556852934e-05,
+      "loss": 0.5272,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.573811509591327e-05,
+      "loss": 0.5089,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.57242146232972e-05,
+      "loss": 0.5838,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.571031415068113e-05,
+      "loss": 0.4928,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.569641367806506e-05,
+      "loss": 0.5404,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.568251320544899e-05,
+      "loss": 0.5215,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.566861273283292e-05,
+      "loss": 0.4795,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.565471226021685e-05,
+      "loss": 0.5417,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.564081178760078e-05,
+      "loss": 0.4968,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.562691131498472e-05,
+      "loss": 0.5315,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.561301084236865e-05,
+      "loss": 0.56,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.559911036975258e-05,
+      "loss": 0.5334,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.558520989713651e-05,
+      "loss": 0.538,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.557130942452044e-05,
+      "loss": 0.4954,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.555740895190437e-05,
+      "loss": 0.4819,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.55435084792883e-05,
+      "loss": 0.5178,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.552960800667223e-05,
+      "loss": 0.5239,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.551570753405616e-05,
+      "loss": 0.5279,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.550180706144009e-05,
+      "loss": 0.5357,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.548790658882404e-05,
+      "loss": 0.5068,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.547400611620795e-05,
+      "loss": 0.51,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.546010564359188e-05,
+      "loss": 0.5166,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.544620517097581e-05,
+      "loss": 0.4793,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.543230469835974e-05,
+      "loss": 0.5179,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.541840422574367e-05,
+      "loss": 0.5051,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.54045037531276e-05,
+      "loss": 0.5319,
+      "step": 1051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.539060328051153e-05,
+      "loss": 0.5055,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.537670280789548e-05,
+      "loss": 0.5511,
+      "step": 1053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.53628023352794e-05,
+      "loss": 0.5241,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.534890186266334e-05,
+      "loss": 0.5057,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.533500139004727e-05,
+      "loss": 0.5073,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.53211009174312e-05,
+      "loss": 0.5123,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.530720044481513e-05,
+      "loss": 0.5125,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.529329997219906e-05,
+      "loss": 0.5061,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.527939949958299e-05,
+      "loss": 0.5288,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.526549902696692e-05,
+      "loss": 0.5047,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.525159855435086e-05,
+      "loss": 0.5582,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.523769808173479e-05,
+      "loss": 0.5534,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.522379760911872e-05,
+      "loss": 0.5815,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.520989713650265e-05,
+      "loss": 0.5426,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.519599666388658e-05,
+      "loss": 0.5325,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.518209619127051e-05,
+      "loss": 0.5508,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.516819571865444e-05,
+      "loss": 0.5879,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.515429524603837e-05,
+      "loss": 0.4861,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.51403947734223e-05,
+      "loss": 0.5572,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.512649430080623e-05,
+      "loss": 0.5579,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.511259382819017e-05,
+      "loss": 0.5193,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.509869335557409e-05,
+      "loss": 0.4883,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.508479288295802e-05,
+      "loss": 0.5413,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.507089241034195e-05,
+      "loss": 0.489,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.505699193772588e-05,
+      "loss": 0.559,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.504309146510981e-05,
+      "loss": 0.4859,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.502919099249374e-05,
+      "loss": 0.6064,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.501529051987767e-05,
+      "loss": 0.5295,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.500139004726161e-05,
+      "loss": 0.5778,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.498748957464554e-05,
+      "loss": 0.5026,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.497358910202947e-05,
+      "loss": 0.6052,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.49596886294134e-05,
+      "loss": 0.5082,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.494578815679733e-05,
+      "loss": 0.4788,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.493188768418126e-05,
+      "loss": 0.4704,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.49179872115652e-05,
+      "loss": 0.5365,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.490408673894912e-05,
+      "loss": 0.5374,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.489018626633305e-05,
+      "loss": 0.5252,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.4876285793717e-05,
+      "loss": 0.5115,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.486238532110093e-05,
+      "loss": 0.5073,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.484848484848486e-05,
+      "loss": 0.4805,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.483458437586879e-05,
+      "loss": 0.5344,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.482068390325272e-05,
+      "loss": 0.4845,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.480678343063665e-05,
+      "loss": 0.4729,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.479288295802058e-05,
+      "loss": 0.4672,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.477898248540451e-05,
+      "loss": 0.5202,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.476508201278844e-05,
+      "loss": 0.4964,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.475118154017237e-05,
+      "loss": 0.4955,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.47372810675563e-05,
+      "loss": 0.5247,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.472338059494023e-05,
+      "loss": 0.5361,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.470948012232416e-05,
+      "loss": 0.5772,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.469557964970809e-05,
+      "loss": 0.5351,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.468167917709202e-05,
+      "loss": 0.5874,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.466777870447595e-05,
+      "loss": 0.4896,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.465387823185988e-05,
+      "loss": 0.5361,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.463997775924381e-05,
+      "loss": 0.4887,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.462607728662775e-05,
+      "loss": 0.4554,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.461217681401168e-05,
+      "loss": 0.5168,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.459827634139561e-05,
+      "loss": 0.4926,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.458437586877954e-05,
+      "loss": 0.4865,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.457047539616347e-05,
+      "loss": 0.5047,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.45565749235474e-05,
+      "loss": 0.5007,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.454267445093133e-05,
+      "loss": 0.5029,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.452877397831526e-05,
+      "loss": 0.5778,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.45148735056992e-05,
+      "loss": 0.5203,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.450097303308312e-05,
+      "loss": 0.4208,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.448707256046707e-05,
+      "loss": 0.4757,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.4473172087851e-05,
+      "loss": 0.4645,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.445927161523493e-05,
+      "loss": 0.5209,
+      "step": 1119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.444537114261886e-05,
+      "loss": 0.5521,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.443147067000279e-05,
+      "loss": 0.4611,
+      "step": 1121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.441757019738672e-05,
+      "loss": 0.4882,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.440366972477065e-05,
+      "loss": 0.5279,
+      "step": 1123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.438976925215458e-05,
+      "loss": 0.4658,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.437586877953851e-05,
+      "loss": 0.5464,
+      "step": 1125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.436196830692244e-05,
+      "loss": 0.5288,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.434806783430637e-05,
+      "loss": 0.5645,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.43341673616903e-05,
+      "loss": 0.4726,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.432026688907423e-05,
+      "loss": 0.5289,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.430636641645816e-05,
+      "loss": 0.5233,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.429246594384209e-05,
+      "loss": 0.5379,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.427856547122602e-05,
+      "loss": 0.5412,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.426466499860995e-05,
+      "loss": 0.5006,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.425076452599389e-05,
+      "loss": 0.4874,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.423686405337782e-05,
+      "loss": 0.4483,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.422296358076175e-05,
+      "loss": 0.5256,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.420906310814568e-05,
+      "loss": 0.4782,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.419516263552961e-05,
+      "loss": 0.4709,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.418126216291354e-05,
+      "loss": 0.4521,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.416736169029747e-05,
+      "loss": 0.5589,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.41534612176814e-05,
+      "loss": 0.4683,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.413956074506533e-05,
+      "loss": 0.5929,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.412566027244926e-05,
+      "loss": 0.5008,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.41117597998332e-05,
+      "loss": 0.4634,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.409785932721714e-05,
+      "loss": 0.4709,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.408395885460107e-05,
+      "loss": 0.5646,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.4070058381985e-05,
+      "loss": 0.5082,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.405615790936893e-05,
+      "loss": 0.4896,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.404225743675286e-05,
+      "loss": 0.4598,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.402835696413679e-05,
+      "loss": 0.4932,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.401445649152072e-05,
+      "loss": 0.5084,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.400055601890465e-05,
+      "loss": 0.4295,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.398665554628858e-05,
+      "loss": 0.5074,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.397275507367251e-05,
+      "loss": 0.5131,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.395885460105644e-05,
+      "loss": 0.4889,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.394495412844037e-05,
+      "loss": 0.4893,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.39310536558243e-05,
+      "loss": 0.4849,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.391715318320823e-05,
+      "loss": 0.5289,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.390325271059216e-05,
+      "loss": 0.533,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.388935223797609e-05,
+      "loss": 0.5856,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.387545176536003e-05,
+      "loss": 0.4922,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.386155129274396e-05,
+      "loss": 0.4588,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.384765082012789e-05,
+      "loss": 0.4988,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.383375034751182e-05,
+      "loss": 0.5634,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.381984987489575e-05,
+      "loss": 0.4878,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.380594940227968e-05,
+      "loss": 0.4764,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.379204892966361e-05,
+      "loss": 0.5012,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.377814845704754e-05,
+      "loss": 0.4942,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.376424798443147e-05,
+      "loss": 0.5322,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.37503475118154e-05,
+      "loss": 0.5496,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.373644703919935e-05,
+      "loss": 0.4935,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.372254656658328e-05,
+      "loss": 0.536,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.37086460939672e-05,
+      "loss": 0.4373,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.369474562135114e-05,
+      "loss": 0.4462,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.368084514873507e-05,
+      "loss": 0.4637,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.3666944676119e-05,
+      "loss": 0.5141,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.365304420350293e-05,
+      "loss": 0.5123,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.363914373088684e-05,
+      "loss": 0.5326,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.362524325827079e-05,
+      "loss": 0.5229,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.361134278565472e-05,
+      "loss": 0.4936,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.359744231303865e-05,
+      "loss": 0.4838,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.358354184042258e-05,
+      "loss": 0.4557,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.35696413678065e-05,
+      "loss": 0.4727,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.355574089519044e-05,
+      "loss": 0.5046,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.354184042257437e-05,
+      "loss": 0.5138,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.35279399499583e-05,
+      "loss": 0.5061,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.351403947734223e-05,
+      "loss": 0.525,
+      "step": 1187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.350013900472617e-05,
+      "loss": 0.5187,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.34862385321101e-05,
+      "loss": 0.4455,
+      "step": 1189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.347233805949403e-05,
+      "loss": 0.5053,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.345843758687796e-05,
+      "loss": 0.5473,
+      "step": 1191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.344453711426189e-05,
+      "loss": 0.5525,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.343063664164582e-05,
+      "loss": 0.4775,
+      "step": 1193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.341673616902975e-05,
+      "loss": 0.5394,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.340283569641368e-05,
+      "loss": 0.5284,
+      "step": 1195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.338893522379761e-05,
+      "loss": 0.508,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.337503475118154e-05,
+      "loss": 0.4569,
+      "step": 1197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.336113427856548e-05,
+      "loss": 0.5142,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.334723380594941e-05,
+      "loss": 0.5103,
+      "step": 1199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.5046,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.331943286071727e-05,
+      "loss": 0.5265,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.33055323881012e-05,
+      "loss": 0.5017,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.329163191548513e-05,
+      "loss": 0.5368,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.327773144286906e-05,
+      "loss": 0.5603,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.326383097025298e-05,
+      "loss": 0.4812,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.324993049763692e-05,
+      "loss": 0.5189,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.323603002502085e-05,
+      "loss": 0.4855,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.322212955240478e-05,
+      "loss": 0.5137,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.320822907978871e-05,
+      "loss": 0.4821,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.319432860717264e-05,
+      "loss": 0.5313,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.318042813455657e-05,
+      "loss": 0.4821,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.31665276619405e-05,
+      "loss": 0.4766,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.315262718932443e-05,
+      "loss": 0.517,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.313872671670836e-05,
+      "loss": 0.4575,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.312482624409231e-05,
+      "loss": 0.4165,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.311092577147624e-05,
+      "loss": 0.5099,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.309702529886017e-05,
+      "loss": 0.5059,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.30831248262441e-05,
+      "loss": 0.4716,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.306922435362803e-05,
+      "loss": 0.4726,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.305532388101196e-05,
+      "loss": 0.537,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.304142340839589e-05,
+      "loss": 0.465,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.302752293577982e-05,
+      "loss": 0.478,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.301362246316375e-05,
+      "loss": 0.4686,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.299972199054768e-05,
+      "loss": 0.5274,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.298582151793162e-05,
+      "loss": 0.4535,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.297192104531555e-05,
+      "loss": 0.5177,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.295802057269948e-05,
+      "loss": 0.4979,
+      "step": 1227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.294412010008341e-05,
+      "loss": 0.5367,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.293021962746734e-05,
+      "loss": 0.4827,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.291631915485127e-05,
+      "loss": 0.5082,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.290241868223519e-05,
+      "loss": 0.519,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.288851820961912e-05,
+      "loss": 0.5082,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.287461773700306e-05,
+      "loss": 0.5591,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.2860717264387e-05,
+      "loss": 0.4502,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.284681679177092e-05,
+      "loss": 0.5664,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.283291631915485e-05,
+      "loss": 0.4711,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.281901584653878e-05,
+      "loss": 0.5349,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.280511537392271e-05,
+      "loss": 0.4863,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.279121490130664e-05,
+      "loss": 0.4742,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.277731442869057e-05,
+      "loss": 0.4699,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.27634139560745e-05,
+      "loss": 0.4924,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.274951348345845e-05,
+      "loss": 0.445,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.273561301084238e-05,
+      "loss": 0.49,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.272171253822631e-05,
+      "loss": 0.5313,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.270781206561024e-05,
+      "loss": 0.4516,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.269391159299417e-05,
+      "loss": 0.5039,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.26800111203781e-05,
+      "loss": 0.5208,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.266611064776203e-05,
+      "loss": 0.4226,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.265221017514596e-05,
+      "loss": 0.4676,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.263830970252989e-05,
+      "loss": 0.5164,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.262440922991382e-05,
+      "loss": 0.534,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.261050875729776e-05,
+      "loss": 0.5381,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.259660828468169e-05,
+      "loss": 0.517,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.258270781206562e-05,
+      "loss": 0.5182,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.256880733944955e-05,
+      "loss": 0.4403,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.255490686683348e-05,
+      "loss": 0.4975,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.254100639421741e-05,
+      "loss": 0.4916,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.252710592160133e-05,
+      "loss": 0.4502,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.251320544898526e-05,
+      "loss": 0.4795,
+      "step": 1259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.24993049763692e-05,
+      "loss": 0.4876,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.248540450375313e-05,
+      "loss": 0.514,
+      "step": 1261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.247150403113706e-05,
+      "loss": 0.5105,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.245760355852099e-05,
+      "loss": 0.4655,
+      "step": 1263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.244370308590492e-05,
+      "loss": 0.4982,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.242980261328885e-05,
+      "loss": 0.502,
+      "step": 1265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.241590214067278e-05,
+      "loss": 0.4392,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.240200166805671e-05,
+      "loss": 0.4759,
+      "step": 1267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.238810119544064e-05,
+      "loss": 0.4951,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.237420072282457e-05,
+      "loss": 0.5922,
+      "step": 1269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.236030025020852e-05,
+      "loss": 0.5206,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.234639977759245e-05,
+      "loss": 0.5108,
+      "step": 1271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.233249930497638e-05,
+      "loss": 0.4871,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.23185988323603e-05,
+      "loss": 0.5205,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.230469835974424e-05,
+      "loss": 0.4693,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.229079788712817e-05,
+      "loss": 0.4786,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.22768974145121e-05,
+      "loss": 0.448,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.226299694189603e-05,
+      "loss": 0.4769,
+      "step": 1277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.224909646927996e-05,
+      "loss": 0.4681,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.22351959966639e-05,
+      "loss": 0.4342,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.222129552404783e-05,
+      "loss": 0.5302,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.220739505143176e-05,
+      "loss": 0.4536,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.219349457881569e-05,
+      "loss": 0.5036,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.217959410619962e-05,
+      "loss": 0.4233,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.216569363358354e-05,
+      "loss": 0.4645,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.215179316096747e-05,
+      "loss": 0.5261,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.21378926883514e-05,
+      "loss": 0.5003,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.212399221573534e-05,
+      "loss": 0.4912,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.211009174311927e-05,
+      "loss": 0.467,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.20961912705032e-05,
+      "loss": 0.4667,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.208229079788713e-05,
+      "loss": 0.495,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.206839032527106e-05,
+      "loss": 0.4905,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.205448985265499e-05,
+      "loss": 0.5054,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.204058938003892e-05,
+      "loss": 0.4632,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.202668890742285e-05,
+      "loss": 0.4622,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.201278843480678e-05,
+      "loss": 0.4724,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.199888796219071e-05,
+      "loss": 0.5132,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.198498748957466e-05,
+      "loss": 0.499,
+      "step": 1297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.197108701695859e-05,
+      "loss": 0.4664,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.195718654434252e-05,
+      "loss": 0.4533,
+      "step": 1299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.194328607172645e-05,
+      "loss": 0.5452,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.192938559911038e-05,
+      "loss": 0.4371,
+      "step": 1301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.19154851264943e-05,
+      "loss": 0.4473,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.190158465387824e-05,
+      "loss": 0.4691,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.188768418126217e-05,
+      "loss": 0.5011,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.18737837086461e-05,
+      "loss": 0.4988,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.185988323603004e-05,
+      "loss": 0.4553,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.184598276341397e-05,
+      "loss": 0.4535,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.18320822907979e-05,
+      "loss": 0.4813,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.181818181818183e-05,
+      "loss": 0.5288,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.180428134556576e-05,
+      "loss": 0.4088,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.179038087294968e-05,
+      "loss": 0.4818,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.17764804003336e-05,
+      "loss": 0.468,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.176257992771754e-05,
+      "loss": 0.465,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.174867945510148e-05,
+      "loss": 0.4363,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.173477898248541e-05,
+      "loss": 0.4937,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.172087850986934e-05,
+      "loss": 0.5161,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.170697803725327e-05,
+      "loss": 0.5035,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.16930775646372e-05,
+      "loss": 0.4319,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.167917709202113e-05,
+      "loss": 0.4992,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.166527661940506e-05,
+      "loss": 0.4241,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.165137614678899e-05,
+      "loss": 0.5062,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.163747567417292e-05,
+      "loss": 0.558,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.162357520155685e-05,
+      "loss": 0.5156,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.16096747289408e-05,
+      "loss": 0.4979,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.159577425632472e-05,
+      "loss": 0.498,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.158187378370865e-05,
+      "loss": 0.4404,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.156797331109258e-05,
+      "loss": 0.4751,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.155407283847651e-05,
+      "loss": 0.4495,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.154017236586044e-05,
+      "loss": 0.5517,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.152627189324437e-05,
+      "loss": 0.5115,
+      "step": 1330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.15123714206283e-05,
+      "loss": 0.4944,
+      "step": 1331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.149847094801223e-05,
+      "loss": 0.5195,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.148457047539618e-05,
+      "loss": 0.5279,
+      "step": 1333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.147067000278011e-05,
+      "loss": 0.4658,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.145676953016404e-05,
+      "loss": 0.475,
+      "step": 1335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.144286905754797e-05,
+      "loss": 0.5083,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.142896858493188e-05,
+      "loss": 0.5077,
+      "step": 1337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.141506811231581e-05,
+      "loss": 0.4809,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.140116763969974e-05,
+      "loss": 0.4971,
+      "step": 1339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.138726716708368e-05,
+      "loss": 0.4618,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.137336669446762e-05,
+      "loss": 0.5087,
+      "step": 1341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.135946622185155e-05,
+      "loss": 0.4396,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.134556574923548e-05,
+      "loss": 0.5156,
+      "step": 1343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.133166527661941e-05,
+      "loss": 0.5009,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.131776480400334e-05,
+      "loss": 0.4636,
+      "step": 1345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.130386433138727e-05,
+      "loss": 0.4457,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.12899638587712e-05,
+      "loss": 0.4157,
+      "step": 1347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.127606338615513e-05,
+      "loss": 0.4399,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.126216291353906e-05,
+      "loss": 0.4066,
+      "step": 1349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.124826244092299e-05,
+      "loss": 0.4738,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.123436196830693e-05,
+      "loss": 0.4606,
+      "step": 1351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.122046149569086e-05,
+      "loss": 0.444,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.120656102307479e-05,
+      "loss": 0.432,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.119266055045872e-05,
+      "loss": 0.4058,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.117876007784265e-05,
+      "loss": 0.4739,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.116485960522658e-05,
+      "loss": 0.4633,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.115095913261051e-05,
+      "loss": 0.4564,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.113705865999444e-05,
+      "loss": 0.558,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.112315818737837e-05,
+      "loss": 0.528,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.11092577147623e-05,
+      "loss": 0.4678,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.109535724214625e-05,
+      "loss": 0.4631,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.108145676953018e-05,
+      "loss": 0.4697,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.106755629691411e-05,
+      "loss": 0.4334,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.105365582429802e-05,
+      "loss": 0.5333,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.103975535168195e-05,
+      "loss": 0.5122,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.102585487906588e-05,
+      "loss": 0.4742,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.101195440644981e-05,
+      "loss": 0.4783,
+      "step": 1367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.099805393383376e-05,
+      "loss": 0.4616,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.098415346121769e-05,
+      "loss": 0.4496,
+      "step": 1369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.097025298860162e-05,
+      "loss": 0.5051,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.095635251598555e-05,
+      "loss": 0.5636,
+      "step": 1371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.094245204336948e-05,
+      "loss": 0.3814,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.092855157075341e-05,
+      "loss": 0.4578,
+      "step": 1373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.091465109813734e-05,
+      "loss": 0.496,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.090075062552127e-05,
+      "loss": 0.4337,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.08868501529052e-05,
+      "loss": 0.5196,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.087294968028913e-05,
+      "loss": 0.4138,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.085904920767307e-05,
+      "loss": 0.4584,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.0845148735057e-05,
+      "loss": 0.4373,
+      "step": 1379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.083124826244093e-05,
+      "loss": 0.4707,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.081734778982486e-05,
+      "loss": 0.5249,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.080344731720879e-05,
+      "loss": 0.492,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.078954684459272e-05,
+      "loss": 0.4457,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.077564637197665e-05,
+      "loss": 0.4812,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.076174589936058e-05,
+      "loss": 0.5163,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.074784542674451e-05,
+      "loss": 0.44,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.073394495412844e-05,
+      "loss": 0.4944,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.072004448151239e-05,
+      "loss": 0.4466,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.070614400889632e-05,
+      "loss": 0.4189,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.069224353628023e-05,
+      "loss": 0.4564,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.067834306366416e-05,
+      "loss": 0.4254,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.066444259104809e-05,
+      "loss": 0.5786,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.065054211843202e-05,
+      "loss": 0.4622,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.063664164581595e-05,
+      "loss": 0.5082,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.062274117319988e-05,
+      "loss": 0.3773,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.060884070058383e-05,
+      "loss": 0.4487,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.059494022796776e-05,
+      "loss": 0.453,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.058103975535169e-05,
+      "loss": 0.4537,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.056713928273562e-05,
+      "loss": 0.4308,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.055323881011955e-05,
+      "loss": 0.4169,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.053933833750348e-05,
+      "loss": 0.5418,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.05254378648874e-05,
+      "loss": 0.44,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.051153739227134e-05,
+      "loss": 0.5224,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.049763691965527e-05,
+      "loss": 0.557,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.048373644703921e-05,
+      "loss": 0.454,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.046983597442314e-05,
+      "loss": 0.4092,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.045593550180707e-05,
+      "loss": 0.4955,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0442035029191e-05,
+      "loss": 0.5118,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.042813455657493e-05,
+      "loss": 0.4543,
+      "step": 1409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.041423408395886e-05,
+      "loss": 0.459,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.040033361134279e-05,
+      "loss": 0.472,
+      "step": 1411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.038643313872672e-05,
+      "loss": 0.5219,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.037253266611065e-05,
+      "loss": 0.4703,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.035863219349458e-05,
+      "loss": 0.4166,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.034473172087852e-05,
+      "loss": 0.4171,
+      "step": 1415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.033083124826244e-05,
+      "loss": 0.4747,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.031693077564637e-05,
+      "loss": 0.4572,
+      "step": 1417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.03030303030303e-05,
+      "loss": 0.4682,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.028912983041423e-05,
+      "loss": 0.4406,
+      "step": 1419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.027522935779816e-05,
+      "loss": 0.4772,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.026132888518209e-05,
+      "loss": 0.4751,
+      "step": 1421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.024742841256602e-05,
+      "loss": 0.4242,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.023352793994997e-05,
+      "loss": 0.4616,
+      "step": 1423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.02196274673339e-05,
+      "loss": 0.4483,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.020572699471783e-05,
+      "loss": 0.3995,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.019182652210176e-05,
+      "loss": 0.4561,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.017792604948569e-05,
+      "loss": 0.476,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.016402557686962e-05,
+      "loss": 0.5045,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.015012510425355e-05,
+      "loss": 0.4216,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.013622463163748e-05,
+      "loss": 0.4573,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.01223241590214e-05,
+      "loss": 0.381,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.010842368640535e-05,
+      "loss": 0.453,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.009452321378928e-05,
+      "loss": 0.4776,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.008062274117321e-05,
+      "loss": 0.4382,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.006672226855714e-05,
+      "loss": 0.4126,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.005282179594107e-05,
+      "loss": 0.4816,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0038921323325e-05,
+      "loss": 0.4522,
+      "step": 1437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.002502085070893e-05,
+      "loss": 0.4429,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.001112037809286e-05,
+      "loss": 0.4258,
+      "step": 1439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.999721990547679e-05,
+      "loss": 0.4906,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.998331943286072e-05,
+      "loss": 0.4519,
+      "step": 1441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.996941896024466e-05,
+      "loss": 0.4615,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.995551848762858e-05,
+      "loss": 0.4693,
+      "step": 1443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.994161801501251e-05,
+      "loss": 0.4997,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.992771754239644e-05,
+      "loss": 0.5794,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.991381706978037e-05,
+      "loss": 0.523,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.98999165971643e-05,
+      "loss": 0.4208,
+      "step": 1447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.988601612454823e-05,
+      "loss": 0.5045,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.987211565193216e-05,
+      "loss": 0.5026,
+      "step": 1449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.98582151793161e-05,
+      "loss": 0.4844,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.984431470670003e-05,
+      "loss": 0.4611,
+      "step": 1451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.983041423408396e-05,
+      "loss": 0.4019,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.98165137614679e-05,
+      "loss": 0.4888,
+      "step": 1453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.980261328885182e-05,
+      "loss": 0.4015,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.978871281623575e-05,
+      "loss": 0.4191,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.977481234361968e-05,
+      "loss": 0.4387,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.976091187100361e-05,
+      "loss": 0.4709,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.974701139838754e-05,
+      "loss": 0.5733,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.973311092577149e-05,
+      "loss": 0.4245,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.971921045315542e-05,
+      "loss": 0.476,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.970530998053935e-05,
+      "loss": 0.4645,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.969140950792328e-05,
+      "loss": 0.5307,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.967750903530721e-05,
+      "loss": 0.4194,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.966360856269114e-05,
+      "loss": 0.4935,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.964970809007507e-05,
+      "loss": 0.4488,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.9635807617459e-05,
+      "loss": 0.4604,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.962190714484293e-05,
+      "loss": 0.4245,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.960800667222686e-05,
+      "loss": 0.4637,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.959410619961079e-05,
+      "loss": 0.4404,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.958020572699472e-05,
+      "loss": 0.4937,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.956630525437865e-05,
+      "loss": 0.4637,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.955240478176258e-05,
+      "loss": 0.4314,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.953850430914651e-05,
+      "loss": 0.5021,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.952460383653044e-05,
+      "loss": 0.47,
+      "step": 1474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.951070336391437e-05,
+      "loss": 0.4515,
+      "step": 1475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.94968028912983e-05,
+      "loss": 0.4567,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.948290241868224e-05,
+      "loss": 0.4904,
+      "step": 1477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.946900194606617e-05,
+      "loss": 0.4889,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.94551014734501e-05,
+      "loss": 0.4576,
+      "step": 1479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.944120100083403e-05,
+      "loss": 0.429,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.942730052821796e-05,
+      "loss": 0.4342,
+      "step": 1481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.941340005560189e-05,
+      "loss": 0.4937,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.939949958298582e-05,
+      "loss": 0.489,
+      "step": 1483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.938559911036975e-05,
+      "loss": 0.4495,
+      "step": 1484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.937169863775368e-05,
+      "loss": 0.4922,
+      "step": 1485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.935779816513761e-05,
+      "loss": 0.3954,
+      "step": 1486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.934389769252156e-05,
+      "loss": 0.4884,
+      "step": 1487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.932999721990549e-05,
+      "loss": 0.3996,
+      "step": 1488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.931609674728942e-05,
+      "loss": 0.4779,
+      "step": 1489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.930219627467335e-05,
+      "loss": 0.4547,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.928829580205728e-05,
+      "loss": 0.3813,
+      "step": 1491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.927439532944121e-05,
+      "loss": 0.4308,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.926049485682514e-05,
+      "loss": 0.4143,
+      "step": 1493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.924659438420907e-05,
+      "loss": 0.4799,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.9232693911593e-05,
+      "loss": 0.4908,
+      "step": 1495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.921879343897693e-05,
+      "loss": 0.4802,
+      "step": 1496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.920489296636086e-05,
+      "loss": 0.4433,
+      "step": 1497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.919099249374479e-05,
+      "loss": 0.4464,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.917709202112872e-05,
+      "loss": 0.4623,
+      "step": 1499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.916319154851265e-05,
+      "loss": 0.4381,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.914929107589658e-05,
+      "loss": 0.487,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.913539060328051e-05,
+      "loss": 0.4387,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.912149013066444e-05,
+      "loss": 0.4753,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.910758965804838e-05,
+      "loss": 0.4183,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.909368918543231e-05,
+      "loss": 0.4603,
+      "step": 1505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.907978871281624e-05,
+      "loss": 0.436,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.906588824020017e-05,
+      "loss": 0.4074,
+      "step": 1507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.90519877675841e-05,
+      "loss": 0.4724,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.903808729496803e-05,
+      "loss": 0.4024,
+      "step": 1509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.902418682235196e-05,
+      "loss": 0.5229,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.901028634973589e-05,
+      "loss": 0.4562,
+      "step": 1511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.899638587711982e-05,
+      "loss": 0.4674,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.898248540450375e-05,
+      "loss": 0.4366,
+      "step": 1513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.89685849318877e-05,
+      "loss": 0.4442,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.895468445927163e-05,
+      "loss": 0.4278,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.894078398665556e-05,
+      "loss": 0.4947,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.892688351403949e-05,
+      "loss": 0.464,
+      "step": 1517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.891298304142342e-05,
+      "loss": 0.5014,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.889908256880735e-05,
+      "loss": 0.4674,
+      "step": 1519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.888518209619128e-05,
+      "loss": 0.3667,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.88712816235752e-05,
+      "loss": 0.4664,
+      "step": 1521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.885738115095914e-05,
+      "loss": 0.4524,
+      "step": 1522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.884348067834307e-05,
+      "loss": 0.4225,
+      "step": 1523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.8829580205727e-05,
+      "loss": 0.4455,
+      "step": 1524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.881567973311093e-05,
+      "loss": 0.4214,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.880177926049486e-05,
+      "loss": 0.5278,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.878787878787879e-05,
+      "loss": 0.4638,
+      "step": 1527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.877397831526272e-05,
+      "loss": 0.4411,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.876007784264665e-05,
+      "loss": 0.3956,
+      "step": 1529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.874617737003058e-05,
+      "loss": 0.4694,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.873227689741452e-05,
+      "loss": 0.487,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.871837642479845e-05,
+      "loss": 0.44,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.870447595218238e-05,
+      "loss": 0.4214,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.869057547956631e-05,
+      "loss": 0.4769,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.867667500695024e-05,
+      "loss": 0.4007,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.866277453433417e-05,
+      "loss": 0.42,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.86488740617181e-05,
+      "loss": 0.4739,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.863497358910203e-05,
+      "loss": 0.4323,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.862107311648596e-05,
+      "loss": 0.4697,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.860717264386989e-05,
+      "loss": 0.4226,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.859327217125383e-05,
+      "loss": 0.4544,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.857937169863776e-05,
+      "loss": 0.5108,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.85654712260217e-05,
+      "loss": 0.5373,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.855157075340562e-05,
+      "loss": 0.4659,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.853767028078955e-05,
+      "loss": 0.4806,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.852376980817349e-05,
+      "loss": 0.4165,
+      "step": 1546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.850986933555742e-05,
+      "loss": 0.5095,
+      "step": 1547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.849596886294135e-05,
+      "loss": 0.444,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.848206839032528e-05,
+      "loss": 0.4364,
+      "step": 1549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.84681679177092e-05,
+      "loss": 0.4704,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.845426744509314e-05,
+      "loss": 0.4447,
+      "step": 1551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.844036697247707e-05,
+      "loss": 0.4678,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.8426466499861e-05,
+      "loss": 0.4594,
+      "step": 1553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.841256602724493e-05,
+      "loss": 0.4045,
+      "step": 1554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.839866555462886e-05,
+      "loss": 0.4294,
+      "step": 1555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.838476508201279e-05,
+      "loss": 0.4697,
+      "step": 1556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.837086460939672e-05,
+      "loss": 0.4467,
+      "step": 1557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.835696413678066e-05,
+      "loss": 0.4141,
+      "step": 1558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.834306366416459e-05,
+      "loss": 0.4437,
+      "step": 1559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.832916319154852e-05,
+      "loss": 0.4876,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.831526271893245e-05,
+      "loss": 0.3721,
+      "step": 1561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.830136224631638e-05,
+      "loss": 0.4202,
+      "step": 1562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.828746177370031e-05,
+      "loss": 0.4365,
+      "step": 1563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.827356130108424e-05,
+      "loss": 0.4511,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.825966082846817e-05,
+      "loss": 0.4419,
+      "step": 1565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.82457603558521e-05,
+      "loss": 0.4563,
+      "step": 1566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.823185988323603e-05,
+      "loss": 0.4405,
+      "step": 1567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.821795941061997e-05,
+      "loss": 0.4566,
+      "step": 1568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.82040589380039e-05,
+      "loss": 0.4221,
+      "step": 1569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.819015846538783e-05,
+      "loss": 0.4399,
+      "step": 1570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.817625799277176e-05,
+      "loss": 0.4517,
+      "step": 1571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.81623575201557e-05,
+      "loss": 0.435,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.814845704753962e-05,
+      "loss": 0.4827,
+      "step": 1573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.813455657492355e-05,
+      "loss": 0.424,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.812065610230747e-05,
+      "loss": 0.3926,
+      "step": 1575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.810675562969141e-05,
+      "loss": 0.4062,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.809285515707534e-05,
+      "loss": 0.4633,
+      "step": 1577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.809285515707534e-05,
+      "loss": 0.4381,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.807895468445927e-05,
+      "loss": 0.4384,
+      "step": 1579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.80650542118432e-05,
+      "loss": 0.4333,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.805115373922713e-05,
+      "loss": 0.4845,
+      "step": 1581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.803725326661106e-05,
+      "loss": 0.4235,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.8023352793995e-05,
+      "loss": 0.4251,
+      "step": 1583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.800945232137892e-05,
+      "loss": 0.4209,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.799555184876285e-05,
+      "loss": 0.4625,
+      "step": 1585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.79816513761468e-05,
+      "loss": 0.4823,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.796775090353073e-05,
+      "loss": 0.4191,
+      "step": 1587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.795385043091466e-05,
+      "loss": 0.4887,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.793994995829859e-05,
+      "loss": 0.4394,
+      "step": 1589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.792604948568252e-05,
+      "loss": 0.4115,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.791214901306645e-05,
+      "loss": 0.4458,
+      "step": 1591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.789824854045038e-05,
+      "loss": 0.4352,
+      "step": 1592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.788434806783431e-05,
+      "loss": 0.463,
+      "step": 1593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.787044759521824e-05,
+      "loss": 0.4618,
+      "step": 1594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.785654712260217e-05,
+      "loss": 0.4767,
+      "step": 1595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.784264664998611e-05,
+      "loss": 0.4362,
+      "step": 1596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.782874617737004e-05,
+      "loss": 0.4557,
+      "step": 1597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.781484570475397e-05,
+      "loss": 0.4189,
+      "step": 1598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.78009452321379e-05,
+      "loss": 0.4358,
+      "step": 1599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.778704475952183e-05,
+      "loss": 0.4885,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.777314428690576e-05,
+      "loss": 0.4764,
+      "step": 1601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.775924381428969e-05,
+      "loss": 0.448,
+      "step": 1602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.774534334167361e-05,
+      "loss": 0.4888,
+      "step": 1603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.773144286905755e-05,
+      "loss": 0.3935,
+      "step": 1604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.771754239644148e-05,
+      "loss": 0.4262,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.770364192382541e-05,
+      "loss": 0.4657,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.768974145120934e-05,
+      "loss": 0.4605,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.767584097859327e-05,
+      "loss": 0.465,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.76619405059772e-05,
+      "loss": 0.4693,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.764804003336113e-05,
+      "loss": 0.425,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.763413956074506e-05,
+      "loss": 0.4672,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.7620239088129e-05,
+      "loss": 0.4536,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.760633861551292e-05,
+      "loss": 0.4395,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.759243814289687e-05,
+      "loss": 0.5058,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.75785376702808e-05,
+      "loss": 0.4848,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.756463719766473e-05,
+      "loss": 0.5062,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.755073672504866e-05,
+      "loss": 0.4457,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.753683625243259e-05,
+      "loss": 0.4049,
+      "step": 1618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.752293577981652e-05,
+      "loss": 0.4023,
+      "step": 1619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.750903530720045e-05,
+      "loss": 0.4816,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.749513483458438e-05,
+      "loss": 0.3912,
+      "step": 1621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.748123436196831e-05,
+      "loss": 0.4762,
+      "step": 1622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.746733388935225e-05,
+      "loss": 0.4,
+      "step": 1623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.745343341673618e-05,
+      "loss": 0.4347,
+      "step": 1624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.743953294412011e-05,
+      "loss": 0.4072,
+      "step": 1625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.742563247150404e-05,
+      "loss": 0.4583,
+      "step": 1626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.741173199888797e-05,
+      "loss": 0.3699,
+      "step": 1627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.73978315262719e-05,
+      "loss": 0.4479,
+      "step": 1628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.738393105365582e-05,
+      "loss": 0.4737,
+      "step": 1629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.737003058103975e-05,
+      "loss": 0.4481,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.735613010842369e-05,
+      "loss": 0.3754,
+      "step": 1631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.734222963580762e-05,
+      "loss": 0.433,
+      "step": 1632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.732832916319155e-05,
+      "loss": 0.4224,
+      "step": 1633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.731442869057548e-05,
+      "loss": 0.4664,
+      "step": 1634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.730052821795941e-05,
+      "loss": 0.5136,
+      "step": 1635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.728662774534334e-05,
+      "loss": 0.4129,
+      "step": 1636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 0.5059,
+      "step": 1637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.72588268001112e-05,
+      "loss": 0.4393,
+      "step": 1638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.724492632749513e-05,
+      "loss": 0.4843,
+      "step": 1639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.723102585487906e-05,
+      "loss": 0.4812,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.7217125382263e-05,
+      "loss": 0.4372,
+      "step": 1641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.720322490964694e-05,
+      "loss": 0.42,
+      "step": 1642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.718932443703087e-05,
+      "loss": 0.4048,
+      "step": 1643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.71754239644148e-05,
+      "loss": 0.4413,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.716152349179873e-05,
+      "loss": 0.4536,
+      "step": 1645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.714762301918266e-05,
+      "loss": 0.4163,
+      "step": 1646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.713372254656659e-05,
+      "loss": 0.4845,
+      "step": 1647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.711982207395052e-05,
+      "loss": 0.4429,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.710592160133445e-05,
+      "loss": 0.4524,
+      "step": 1649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.709202112871839e-05,
+      "loss": 0.4161,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.707812065610232e-05,
+      "loss": 0.4436,
+      "step": 1651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.706422018348625e-05,
+      "loss": 0.4259,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.705031971087018e-05,
+      "loss": 0.4458,
+      "step": 1653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.703641923825411e-05,
+      "loss": 0.4457,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.702251876563804e-05,
+      "loss": 0.4157,
+      "step": 1655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.700861829302196e-05,
+      "loss": 0.4019,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.699471782040589e-05,
+      "loss": 0.4216,
+      "step": 1657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.698081734778983e-05,
+      "loss": 0.4103,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.696691687517376e-05,
+      "loss": 0.4868,
+      "step": 1659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.695301640255769e-05,
+      "loss": 0.4216,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.693911592994162e-05,
+      "loss": 0.486,
+      "step": 1661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.692521545732555e-05,
+      "loss": 0.4809,
+      "step": 1662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.691131498470948e-05,
+      "loss": 0.4315,
+      "step": 1663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.689741451209341e-05,
+      "loss": 0.4336,
+      "step": 1664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.688351403947734e-05,
+      "loss": 0.4016,
+      "step": 1665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.686961356686127e-05,
+      "loss": 0.5045,
+      "step": 1666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.68557130942452e-05,
+      "loss": 0.4362,
+      "step": 1667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.684181262162914e-05,
+      "loss": 0.4498,
+      "step": 1668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.682791214901307e-05,
+      "loss": 0.4243,
+      "step": 1669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.6814011676397e-05,
+      "loss": 0.4535,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.680011120378093e-05,
+      "loss": 0.4618,
+      "step": 1671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.678621073116487e-05,
+      "loss": 0.3567,
+      "step": 1672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.67723102585488e-05,
+      "loss": 0.4467,
+      "step": 1673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.675840978593273e-05,
+      "loss": 0.3688,
+      "step": 1674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.674450931331666e-05,
+      "loss": 0.4496,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.673060884070059e-05,
+      "loss": 0.4408,
+      "step": 1676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.671670836808453e-05,
+      "loss": 0.4024,
+      "step": 1677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.670280789546846e-05,
+      "loss": 0.3982,
+      "step": 1678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.668890742285239e-05,
+      "loss": 0.3761,
+      "step": 1679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.667500695023632e-05,
+      "loss": 0.4691,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.666110647762025e-05,
+      "loss": 0.4534,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.664720600500417e-05,
+      "loss": 0.4655,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.66333055323881e-05,
+      "loss": 0.462,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.661940505977203e-05,
+      "loss": 0.4537,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.660550458715597e-05,
+      "loss": 0.4325,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.65916041145399e-05,
+      "loss": 0.5036,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.657770364192383e-05,
+      "loss": 0.444,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.656380316930776e-05,
+      "loss": 0.4762,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.654990269669169e-05,
+      "loss": 0.4931,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.653600222407562e-05,
+      "loss": 0.4291,
+      "step": 1690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.652210175145955e-05,
+      "loss": 0.422,
+      "step": 1691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.650820127884348e-05,
+      "loss": 0.4472,
+      "step": 1692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.649430080622741e-05,
+      "loss": 0.4739,
+      "step": 1693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.648040033361134e-05,
+      "loss": 0.5114,
+      "step": 1694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.646649986099528e-05,
+      "loss": 0.5013,
+      "step": 1695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.645259938837921e-05,
+      "loss": 0.4158,
+      "step": 1696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.643869891576314e-05,
+      "loss": 0.3636,
+      "step": 1697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.642479844314707e-05,
+      "loss": 0.4585,
+      "step": 1698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.6410897970531e-05,
+      "loss": 0.4724,
+      "step": 1699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.639699749791493e-05,
+      "loss": 0.5097,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.638309702529886e-05,
+      "loss": 0.4371,
+      "step": 1701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.63691965526828e-05,
+      "loss": 0.4573,
+      "step": 1702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.635529608006672e-05,
+      "loss": 0.4739,
+      "step": 1703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.634139560745065e-05,
+      "loss": 0.4054,
+      "step": 1704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.63274951348346e-05,
+      "loss": 0.398,
+      "step": 1705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.631359466221853e-05,
+      "loss": 0.48,
+      "step": 1706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.629969418960246e-05,
+      "loss": 0.5308,
+      "step": 1707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.628579371698637e-05,
+      "loss": 0.4813,
+      "step": 1708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.62718932443703e-05,
+      "loss": 0.4391,
+      "step": 1709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.625799277175423e-05,
+      "loss": 0.3949,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.624409229913816e-05,
+      "loss": 0.4271,
+      "step": 1711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.623019182652211e-05,
+      "loss": 0.5167,
+      "step": 1712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.621629135390604e-05,
+      "loss": 0.4384,
+      "step": 1713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.620239088128997e-05,
+      "loss": 0.4438,
+      "step": 1714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.61884904086739e-05,
+      "loss": 0.5717,
+      "step": 1715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.617458993605783e-05,
+      "loss": 0.419,
+      "step": 1716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.616068946344176e-05,
+      "loss": 0.4434,
+      "step": 1717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.614678899082569e-05,
+      "loss": 0.4559,
+      "step": 1718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.613288851820962e-05,
+      "loss": 0.4345,
+      "step": 1719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.611898804559355e-05,
+      "loss": 0.4007,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.610508757297748e-05,
+      "loss": 0.4254,
+      "step": 1721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.609118710036142e-05,
+      "loss": 0.4082,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.607728662774535e-05,
+      "loss": 0.4374,
+      "step": 1723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.606338615512928e-05,
+      "loss": 0.4584,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.604948568251321e-05,
+      "loss": 0.4322,
+      "step": 1725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.603558520989714e-05,
+      "loss": 0.4645,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.602168473728107e-05,
+      "loss": 0.4984,
+      "step": 1727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.6007784264665e-05,
+      "loss": 0.3839,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.599388379204893e-05,
+      "loss": 0.3644,
+      "step": 1729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.597998331943286e-05,
+      "loss": 0.4349,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.596608284681679e-05,
+      "loss": 0.442,
+      "step": 1731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.595218237420074e-05,
+      "loss": 0.4313,
+      "step": 1732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.593828190158467e-05,
+      "loss": 0.4227,
+      "step": 1733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.59243814289686e-05,
+      "loss": 0.4765,
+      "step": 1734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.591048095635251e-05,
+      "loss": 0.4631,
+      "step": 1735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.589658048373644e-05,
+      "loss": 0.4586,
+      "step": 1736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.588268001112037e-05,
+      "loss": 0.4694,
+      "step": 1737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.58687795385043e-05,
+      "loss": 0.4084,
+      "step": 1738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.585487906588823e-05,
+      "loss": 0.4144,
+      "step": 1739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.584097859327218e-05,
+      "loss": 0.4626,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.582707812065611e-05,
+      "loss": 0.4587,
+      "step": 1741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.581317764804004e-05,
+      "loss": 0.4551,
+      "step": 1742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.579927717542397e-05,
+      "loss": 0.4256,
+      "step": 1743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.57853767028079e-05,
+      "loss": 0.4344,
+      "step": 1744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.577147623019183e-05,
+      "loss": 0.4496,
+      "step": 1745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 0.4679,
+      "step": 1746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.574367528495969e-05,
+      "loss": 0.4585,
+      "step": 1747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.572977481234362e-05,
+      "loss": 0.4254,
+      "step": 1748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.571587433972756e-05,
+      "loss": 0.4497,
+      "step": 1749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.570197386711149e-05,
+      "loss": 0.4254,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.568807339449542e-05,
+      "loss": 0.4938,
+      "step": 1751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.567417292187935e-05,
+      "loss": 0.4076,
+      "step": 1752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.566027244926328e-05,
+      "loss": 0.441,
+      "step": 1753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.564637197664721e-05,
+      "loss": 0.4898,
+      "step": 1754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.563247150403114e-05,
+      "loss": 0.4949,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.561857103141507e-05,
+      "loss": 0.4628,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.5604670558799e-05,
+      "loss": 0.43,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.559077008618293e-05,
+      "loss": 0.4253,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.557686961356688e-05,
+      "loss": 0.4332,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.55629691409508e-05,
+      "loss": 0.4762,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.554906866833472e-05,
+      "loss": 0.41,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.553516819571865e-05,
+      "loss": 0.4392,
+      "step": 1762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.552126772310258e-05,
+      "loss": 0.4573,
+      "step": 1763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.550736725048651e-05,
+      "loss": 0.5107,
+      "step": 1764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.549346677787044e-05,
+      "loss": 0.4085,
+      "step": 1765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.547956630525437e-05,
+      "loss": 0.4458,
+      "step": 1766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.546566583263832e-05,
+      "loss": 0.4532,
+      "step": 1767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.545176536002225e-05,
+      "loss": 0.4591,
+      "step": 1768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.543786488740618e-05,
+      "loss": 0.4366,
+      "step": 1769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.54239644147901e-05,
+      "loss": 0.4372,
+      "step": 1770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.541006394217404e-05,
+      "loss": 0.3958,
+      "step": 1771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.539616346955797e-05,
+      "loss": 0.3719,
+      "step": 1772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.53822629969419e-05,
+      "loss": 0.473,
+      "step": 1773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.536836252432583e-05,
+      "loss": 0.5036,
+      "step": 1774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.535446205170976e-05,
+      "loss": 0.3995,
+      "step": 1775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.53405615790937e-05,
+      "loss": 0.4513,
+      "step": 1776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.532666110647763e-05,
+      "loss": 0.4517,
+      "step": 1777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.531276063386156e-05,
+      "loss": 0.4545,
+      "step": 1778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.529886016124549e-05,
+      "loss": 0.5039,
+      "step": 1779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.528495968862942e-05,
+      "loss": 0.4943,
+      "step": 1780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.527105921601335e-05,
+      "loss": 0.4715,
+      "step": 1781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.525715874339728e-05,
+      "loss": 0.3802,
+      "step": 1782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.524325827078121e-05,
+      "loss": 0.488,
+      "step": 1783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.522935779816514e-05,
+      "loss": 0.399,
+      "step": 1784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.521545732554907e-05,
+      "loss": 0.5334,
+      "step": 1785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.520155685293301e-05,
+      "loss": 0.4856,
+      "step": 1786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.518765638031694e-05,
+      "loss": 0.4497,
+      "step": 1787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.517375590770086e-05,
+      "loss": 0.4246,
+      "step": 1788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.515985543508479e-05,
+      "loss": 0.492,
+      "step": 1789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.514595496246872e-05,
+      "loss": 0.4368,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.513205448985265e-05,
+      "loss": 0.4608,
+      "step": 1791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.511815401723658e-05,
+      "loss": 0.4996,
+      "step": 1792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.510425354462051e-05,
+      "loss": 0.3864,
+      "step": 1793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.509035307200445e-05,
+      "loss": 0.4586,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.507645259938838e-05,
+      "loss": 0.4036,
+      "step": 1795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.506255212677231e-05,
+      "loss": 0.394,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.504865165415624e-05,
+      "loss": 0.4134,
+      "step": 1797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.503475118154018e-05,
+      "loss": 0.484,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.50208507089241e-05,
+      "loss": 0.4417,
+      "step": 1799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.500695023630804e-05,
+      "loss": 0.3812,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.499304976369197e-05,
+      "loss": 0.4517,
+      "step": 1801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.49791492910759e-05,
+      "loss": 0.4216,
+      "step": 1802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.496524881845984e-05,
+      "loss": 0.4864,
+      "step": 1803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.495134834584377e-05,
+      "loss": 0.4679,
+      "step": 1804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.49374478732277e-05,
+      "loss": 0.4247,
+      "step": 1805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.492354740061163e-05,
+      "loss": 0.3943,
+      "step": 1806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.490964692799556e-05,
+      "loss": 0.4525,
+      "step": 1807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.489574645537949e-05,
+      "loss": 0.4735,
+      "step": 1808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.488184598276342e-05,
+      "loss": 0.4461,
+      "step": 1809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.486794551014735e-05,
+      "loss": 0.3985,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.485404503753128e-05,
+      "loss": 0.4233,
+      "step": 1811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.484014456491521e-05,
+      "loss": 0.4657,
+      "step": 1812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.482624409229915e-05,
+      "loss": 0.4284,
+      "step": 1813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.481234361968307e-05,
+      "loss": 0.4464,
+      "step": 1814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.4798443147067e-05,
+      "loss": 0.4264,
+      "step": 1815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.478454267445093e-05,
+      "loss": 0.402,
+      "step": 1816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.477064220183486e-05,
+      "loss": 0.449,
+      "step": 1817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.475674172921879e-05,
+      "loss": 0.4258,
+      "step": 1818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.474284125660272e-05,
+      "loss": 0.4044,
+      "step": 1819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.472894078398665e-05,
+      "loss": 0.4273,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.47150403113706e-05,
+      "loss": 0.4307,
+      "step": 1821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.470113983875452e-05,
+      "loss": 0.448,
+      "step": 1822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.468723936613845e-05,
+      "loss": 0.401,
+      "step": 1823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.467333889352238e-05,
+      "loss": 0.3937,
+      "step": 1824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.465943842090631e-05,
+      "loss": 0.41,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.464553794829024e-05,
+      "loss": 0.5274,
+      "step": 1826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.463163747567417e-05,
+      "loss": 0.4087,
+      "step": 1827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.46177370030581e-05,
+      "loss": 0.391,
+      "step": 1828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.460383653044203e-05,
+      "loss": 0.4547,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.458993605782596e-05,
+      "loss": 0.4365,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.458993605782596e-05,
+      "loss": 0.4915,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.457603558520991e-05,
+      "loss": 0.434,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.456213511259384e-05,
+      "loss": 0.4034,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.454823463997777e-05,
+      "loss": 0.4248,
+      "step": 1834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.45343341673617e-05,
+      "loss": 0.4513,
+      "step": 1835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.452043369474563e-05,
+      "loss": 0.4872,
+      "step": 1836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.450653322212956e-05,
+      "loss": 0.4281,
+      "step": 1837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.449263274951349e-05,
+      "loss": 0.47,
+      "step": 1838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.447873227689742e-05,
+      "loss": 0.4153,
+      "step": 1839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.446483180428135e-05,
+      "loss": 0.3878,
+      "step": 1840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.445093133166529e-05,
+      "loss": 0.4282,
+      "step": 1841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.443703085904921e-05,
+      "loss": 0.4201,
+      "step": 1842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.442313038643314e-05,
+      "loss": 0.4432,
+      "step": 1843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440922991381707e-05,
+      "loss": 0.3583,
+      "step": 1844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.4395329441201e-05,
+      "loss": 0.402,
+      "step": 1845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.438142896858493e-05,
+      "loss": 0.3704,
+      "step": 1846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.436752849596886e-05,
+      "loss": 0.4729,
+      "step": 1847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.435362802335279e-05,
+      "loss": 0.4252,
+      "step": 1848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.433972755073673e-05,
+      "loss": 0.4095,
+      "step": 1849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.432582707812066e-05,
+      "loss": 0.4409,
+      "step": 1850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.431192660550459e-05,
+      "loss": 0.472,
+      "step": 1851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.429802613288852e-05,
+      "loss": 0.4446,
+      "step": 1852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.428412566027245e-05,
+      "loss": 0.447,
+      "step": 1853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.427022518765638e-05,
+      "loss": 0.4134,
+      "step": 1854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.425632471504031e-05,
+      "loss": 0.412,
+      "step": 1855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.424242424242424e-05,
+      "loss": 0.4586,
+      "step": 1856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.422852376980817e-05,
+      "loss": 0.4044,
+      "step": 1857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.42146232971921e-05,
+      "loss": 0.4289,
+      "step": 1858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.420072282457605e-05,
+      "loss": 0.482,
+      "step": 1859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.418682235195998e-05,
+      "loss": 0.4188,
+      "step": 1860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.41729218793439e-05,
+      "loss": 0.4108,
+      "step": 1861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.415902140672784e-05,
+      "loss": 0.4081,
+      "step": 1862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.414512093411177e-05,
+      "loss": 0.4305,
+      "step": 1863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.41312204614957e-05,
+      "loss": 0.4096,
+      "step": 1864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.411731998887963e-05,
+      "loss": 0.4338,
+      "step": 1865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.410341951626356e-05,
+      "loss": 0.4581,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.408951904364749e-05,
+      "loss": 0.4371,
+      "step": 1867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.407561857103142e-05,
+      "loss": 0.3997,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.406171809841535e-05,
+      "loss": 0.451,
+      "step": 1869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.404781762579928e-05,
+      "loss": 0.3558,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.403391715318321e-05,
+      "loss": 0.376,
+      "step": 1871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.402001668056714e-05,
+      "loss": 0.4449,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.400611620795107e-05,
+      "loss": 0.453,
+      "step": 1873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.3992215735335e-05,
+      "loss": 0.3931,
+      "step": 1874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.397831526271893e-05,
+      "loss": 0.4058,
+      "step": 1875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.396441479010287e-05,
+      "loss": 0.4522,
+      "step": 1876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.39505143174868e-05,
+      "loss": 0.4267,
+      "step": 1877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.393661384487073e-05,
+      "loss": 0.4111,
+      "step": 1878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.392271337225466e-05,
+      "loss": 0.402,
+      "step": 1879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.390881289963859e-05,
+      "loss": 0.4346,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.389491242702252e-05,
+      "loss": 0.4221,
+      "step": 1881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.388101195440645e-05,
+      "loss": 0.4443,
+      "step": 1882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.386711148179038e-05,
+      "loss": 0.4888,
+      "step": 1883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.385321100917431e-05,
+      "loss": 0.4437,
+      "step": 1884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.383931053655824e-05,
+      "loss": 0.4779,
+      "step": 1885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.382541006394219e-05,
+      "loss": 0.4067,
+      "step": 1886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.381150959132612e-05,
+      "loss": 0.388,
+      "step": 1887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.379760911871005e-05,
+      "loss": 0.3987,
+      "step": 1888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.378370864609398e-05,
+      "loss": 0.4351,
+      "step": 1889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.37698081734779e-05,
+      "loss": 0.4405,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.375590770086184e-05,
+      "loss": 0.4132,
+      "step": 1891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.374200722824577e-05,
+      "loss": 0.4784,
+      "step": 1892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.37281067556297e-05,
+      "loss": 0.4158,
+      "step": 1893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.371420628301363e-05,
+      "loss": 0.409,
+      "step": 1894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.370030581039756e-05,
+      "loss": 0.3948,
+      "step": 1895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.368640533778149e-05,
+      "loss": 0.3754,
+      "step": 1896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.367250486516542e-05,
+      "loss": 0.3689,
+      "step": 1897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.365860439254935e-05,
+      "loss": 0.3998,
+      "step": 1898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.364470391993328e-05,
+      "loss": 0.3999,
+      "step": 1899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.36308034473172e-05,
+      "loss": 0.3798,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.361690297470114e-05,
+      "loss": 0.4259,
+      "step": 1901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.360300250208507e-05,
+      "loss": 0.4553,
+      "step": 1902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.358910202946901e-05,
+      "loss": 0.4473,
+      "step": 1903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.357520155685294e-05,
+      "loss": 0.4347,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.356130108423687e-05,
+      "loss": 0.3799,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.35474006116208e-05,
+      "loss": 0.4436,
+      "step": 1906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.353350013900473e-05,
+      "loss": 0.5312,
+      "step": 1907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.351959966638866e-05,
+      "loss": 0.4322,
+      "step": 1908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.350569919377259e-05,
+      "loss": 0.4343,
+      "step": 1909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.349179872115652e-05,
+      "loss": 0.4576,
+      "step": 1910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.347789824854045e-05,
+      "loss": 0.4754,
+      "step": 1911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.346399777592438e-05,
+      "loss": 0.4491,
+      "step": 1912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.345009730330832e-05,
+      "loss": 0.449,
+      "step": 1913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.343619683069225e-05,
+      "loss": 0.4546,
+      "step": 1914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.342229635807618e-05,
+      "loss": 0.4165,
+      "step": 1915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.340839588546011e-05,
+      "loss": 0.4123,
+      "step": 1916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.339449541284404e-05,
+      "loss": 0.4027,
+      "step": 1917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.338059494022797e-05,
+      "loss": 0.4457,
+      "step": 1918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.33666944676119e-05,
+      "loss": 0.3794,
+      "step": 1919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.335279399499583e-05,
+      "loss": 0.4166,
+      "step": 1920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.333889352237976e-05,
+      "loss": 0.4445,
+      "step": 1921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.33249930497637e-05,
+      "loss": 0.392,
+      "step": 1922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.331109257714762e-05,
+      "loss": 0.4041,
+      "step": 1923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.329719210453156e-05,
+      "loss": 0.3637,
+      "step": 1924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.328329163191549e-05,
+      "loss": 0.433,
+      "step": 1925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.326939115929942e-05,
+      "loss": 0.4196,
+      "step": 1926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.325549068668335e-05,
+      "loss": 0.4457,
+      "step": 1927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.324159021406728e-05,
+      "loss": 0.4514,
+      "step": 1928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.32276897414512e-05,
+      "loss": 0.4971,
+      "step": 1929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.321378926883515e-05,
+      "loss": 0.4623,
+      "step": 1930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.319988879621908e-05,
+      "loss": 0.4061,
+      "step": 1931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.318598832360301e-05,
+      "loss": 0.4308,
+      "step": 1932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.317208785098694e-05,
+      "loss": 0.4435,
+      "step": 1933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.315818737837087e-05,
+      "loss": 0.4438,
+      "step": 1934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.31442869057548e-05,
+      "loss": 0.4204,
+      "step": 1935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.313038643313873e-05,
+      "loss": 0.4387,
+      "step": 1936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.311648596052266e-05,
+      "loss": 0.3651,
+      "step": 1937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.310258548790659e-05,
+      "loss": 0.419,
+      "step": 1938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.308868501529052e-05,
+      "loss": 0.4451,
+      "step": 1939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.307478454267446e-05,
+      "loss": 0.4775,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.306088407005839e-05,
+      "loss": 0.4106,
+      "step": 1941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.304698359744232e-05,
+      "loss": 0.4306,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.303308312482625e-05,
+      "loss": 0.4445,
+      "step": 1943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.301918265221018e-05,
+      "loss": 0.4733,
+      "step": 1944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.300528217959411e-05,
+      "loss": 0.4034,
+      "step": 1945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.299138170697804e-05,
+      "loss": 0.3986,
+      "step": 1946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.297748123436196e-05,
+      "loss": 0.4065,
+      "step": 1947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.29635807617459e-05,
+      "loss": 0.4505,
+      "step": 1948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.294968028912983e-05,
+      "loss": 0.4337,
+      "step": 1949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.293577981651376e-05,
+      "loss": 0.4461,
+      "step": 1950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.29218793438977e-05,
+      "loss": 0.4073,
+      "step": 1951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.290797887128162e-05,
+      "loss": 0.4716,
+      "step": 1952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.289407839866555e-05,
+      "loss": 0.4569,
+      "step": 1953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.288017792604948e-05,
+      "loss": 0.4202,
+      "step": 1954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.286627745343341e-05,
+      "loss": 0.4012,
+      "step": 1955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.285237698081734e-05,
+      "loss": 0.4178,
+      "step": 1956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.283847650820127e-05,
+      "loss": 0.4193,
+      "step": 1957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.282457603558522e-05,
+      "loss": 0.4787,
+      "step": 1958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.281067556296915e-05,
+      "loss": 0.4402,
+      "step": 1959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.279677509035308e-05,
+      "loss": 0.4217,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.278287461773701e-05,
+      "loss": 0.3897,
+      "step": 1961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.276897414512094e-05,
+      "loss": 0.4603,
+      "step": 1962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.275507367250487e-05,
+      "loss": 0.4634,
+      "step": 1963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.27411731998888e-05,
+      "loss": 0.4748,
+      "step": 1964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.5262,
+      "step": 1965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.271337225465666e-05,
+      "loss": 0.461,
+      "step": 1966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.26994717820406e-05,
+      "loss": 0.399,
+      "step": 1967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.268557130942453e-05,
+      "loss": 0.4382,
+      "step": 1968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.267167083680846e-05,
+      "loss": 0.4188,
+      "step": 1969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.265777036419239e-05,
+      "loss": 0.426,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.264386989157632e-05,
+      "loss": 0.3676,
+      "step": 1971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.262996941896025e-05,
+      "loss": 0.4786,
+      "step": 1972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.261606894634418e-05,
+      "loss": 0.4377,
+      "step": 1973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.26021684737281e-05,
+      "loss": 0.4108,
+      "step": 1974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.258826800111204e-05,
+      "loss": 0.3885,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.257436752849597e-05,
+      "loss": 0.4033,
+      "step": 1976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.25604670558799e-05,
+      "loss": 0.4705,
+      "step": 1977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.254656658326383e-05,
+      "loss": 0.5029,
+      "step": 1978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.253266611064776e-05,
+      "loss": 0.4228,
+      "step": 1979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.251876563803169e-05,
+      "loss": 0.3968,
+      "step": 1980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.250486516541562e-05,
+      "loss": 0.4471,
+      "step": 1981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.249096469279955e-05,
+      "loss": 0.4176,
+      "step": 1982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.247706422018348e-05,
+      "loss": 0.4648,
+      "step": 1983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.246316374756741e-05,
+      "loss": 0.453,
+      "step": 1984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.244926327495136e-05,
+      "loss": 0.3989,
+      "step": 1985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.243536280233529e-05,
+      "loss": 0.3961,
+      "step": 1986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.242146232971922e-05,
+      "loss": 0.4352,
+      "step": 1987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.240756185710315e-05,
+      "loss": 0.4203,
+      "step": 1988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.239366138448708e-05,
+      "loss": 0.4664,
+      "step": 1989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.237976091187101e-05,
+      "loss": 0.4685,
+      "step": 1990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.236586043925494e-05,
+      "loss": 0.3601,
+      "step": 1991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.235195996663887e-05,
+      "loss": 0.3019,
+      "step": 1992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.23380594940228e-05,
+      "loss": 0.4134,
+      "step": 1993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.232415902140674e-05,
+      "loss": 0.3981,
+      "step": 1994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.231025854879067e-05,
+      "loss": 0.4281,
+      "step": 1995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.22963580761746e-05,
+      "loss": 0.482,
+      "step": 1996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.228245760355853e-05,
+      "loss": 0.401,
+      "step": 1997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.226855713094246e-05,
+      "loss": 0.4012,
+      "step": 1998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.225465665832639e-05,
+      "loss": 0.4841,
+      "step": 1999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.224075618571031e-05,
+      "loss": 0.4309,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.222685571309424e-05,
+      "loss": 0.4241,
+      "step": 2001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.221295524047818e-05,
+      "loss": 0.3752,
+      "step": 2002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.219905476786211e-05,
+      "loss": 0.3711,
+      "step": 2003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.218515429524604e-05,
+      "loss": 0.416,
+      "step": 2004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.217125382262997e-05,
+      "loss": 0.3207,
+      "step": 2005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.21573533500139e-05,
+      "loss": 0.4281,
+      "step": 2006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.214345287739783e-05,
+      "loss": 0.4058,
+      "step": 2007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.212955240478176e-05,
+      "loss": 0.4199,
+      "step": 2008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.211565193216569e-05,
+      "loss": 0.4026,
+      "step": 2009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.210175145954962e-05,
+      "loss": 0.3683,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.208785098693355e-05,
+      "loss": 0.3942,
+      "step": 2011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.20739505143175e-05,
+      "loss": 0.4068,
+      "step": 2012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.206005004170143e-05,
+      "loss": 0.4514,
+      "step": 2013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.204614956908536e-05,
+      "loss": 0.3896,
+      "step": 2014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.203224909646929e-05,
+      "loss": 0.396,
+      "step": 2015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.201834862385322e-05,
+      "loss": 0.4204,
+      "step": 2016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.200444815123715e-05,
+      "loss": 0.3808,
+      "step": 2017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.199054767862108e-05,
+      "loss": 0.4694,
+      "step": 2018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.1976647206005e-05,
+      "loss": 0.4391,
+      "step": 2019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.196274673338894e-05,
+      "loss": 0.4457,
+      "step": 2020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.194884626077288e-05,
+      "loss": 0.4165,
+      "step": 2021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.193494578815681e-05,
+      "loss": 0.3945,
+      "step": 2022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.192104531554074e-05,
+      "loss": 0.4775,
+      "step": 2023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.190714484292467e-05,
+      "loss": 0.4045,
+      "step": 2024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.18932443703086e-05,
+      "loss": 0.4282,
+      "step": 2025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.187934389769253e-05,
+      "loss": 0.3908,
+      "step": 2026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.186544342507645e-05,
+      "loss": 0.4229,
+      "step": 2027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.185154295246038e-05,
+      "loss": 0.4141,
+      "step": 2028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.183764247984432e-05,
+      "loss": 0.4404,
+      "step": 2029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.182374200722825e-05,
+      "loss": 0.4613,
+      "step": 2030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.180984153461218e-05,
+      "loss": 0.4098,
+      "step": 2031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.179594106199611e-05,
+      "loss": 0.4543,
+      "step": 2032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.178204058938004e-05,
+      "loss": 0.4525,
+      "step": 2033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.176814011676397e-05,
+      "loss": 0.5294,
+      "step": 2034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.17542396441479e-05,
+      "loss": 0.4206,
+      "step": 2035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.174033917153183e-05,
+      "loss": 0.4328,
+      "step": 2036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.172643869891576e-05,
+      "loss": 0.4343,
+      "step": 2037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.171253822629969e-05,
+      "loss": 0.4018,
+      "step": 2038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.169863775368363e-05,
+      "loss": 0.4522,
+      "step": 2039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.168473728106756e-05,
+      "loss": 0.4117,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.16708368084515e-05,
+      "loss": 0.4183,
+      "step": 2041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.165693633583542e-05,
+      "loss": 0.4314,
+      "step": 2042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.164303586321935e-05,
+      "loss": 0.4289,
+      "step": 2043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.162913539060328e-05,
+      "loss": 0.3717,
+      "step": 2044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.161523491798721e-05,
+      "loss": 0.4508,
+      "step": 2045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.160133444537114e-05,
+      "loss": 0.4068,
+      "step": 2046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.158743397275507e-05,
+      "loss": 0.4855,
+      "step": 2047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.157353350013902e-05,
+      "loss": 0.3946,
+      "step": 2048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.155963302752295e-05,
+      "loss": 0.3911,
+      "step": 2049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.154573255490688e-05,
+      "loss": 0.35,
+      "step": 2050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.153183208229081e-05,
+      "loss": 0.4389,
+      "step": 2051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.151793160967474e-05,
+      "loss": 0.4314,
+      "step": 2052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.150403113705866e-05,
+      "loss": 0.3672,
+      "step": 2053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.149013066444259e-05,
+      "loss": 0.4299,
+      "step": 2054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.147623019182652e-05,
+      "loss": 0.4126,
+      "step": 2055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.146232971921046e-05,
+      "loss": 0.4389,
+      "step": 2056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.144842924659439e-05,
+      "loss": 0.4122,
+      "step": 2057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.143452877397832e-05,
+      "loss": 0.3657,
+      "step": 2058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.142062830136225e-05,
+      "loss": 0.4265,
+      "step": 2059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.140672782874618e-05,
+      "loss": 0.4366,
+      "step": 2060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.139282735613011e-05,
+      "loss": 0.4222,
+      "step": 2061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.137892688351404e-05,
+      "loss": 0.3766,
+      "step": 2062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.136502641089797e-05,
+      "loss": 0.4421,
+      "step": 2063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.13511259382819e-05,
+      "loss": 0.4069,
+      "step": 2064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.133722546566583e-05,
+      "loss": 0.5208,
+      "step": 2065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.132332499304977e-05,
+      "loss": 0.4164,
+      "step": 2066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.13094245204337e-05,
+      "loss": 0.4112,
+      "step": 2067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.129552404781763e-05,
+      "loss": 0.4601,
+      "step": 2068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.128162357520156e-05,
+      "loss": 0.4236,
+      "step": 2069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.12677231025855e-05,
+      "loss": 0.3863,
+      "step": 2070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.125382262996942e-05,
+      "loss": 0.3696,
+      "step": 2071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.123992215735335e-05,
+      "loss": 0.4228,
+      "step": 2072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.122602168473728e-05,
+      "loss": 0.4102,
+      "step": 2073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.121212121212121e-05,
+      "loss": 0.4058,
+      "step": 2074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.119822073950514e-05,
+      "loss": 0.3519,
+      "step": 2075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.118432026688909e-05,
+      "loss": 0.3954,
+      "step": 2076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.117041979427302e-05,
+      "loss": 0.3744,
+      "step": 2077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.115651932165695e-05,
+      "loss": 0.479,
+      "step": 2078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.114261884904088e-05,
+      "loss": 0.4425,
+      "step": 2079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.11287183764248e-05,
+      "loss": 0.3902,
+      "step": 2080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.111481790380872e-05,
+      "loss": 0.4845,
+      "step": 2081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.110091743119265e-05,
+      "loss": 0.3904,
+      "step": 2082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.10870169585766e-05,
+      "loss": 0.3928,
+      "step": 2083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.107311648596053e-05,
+      "loss": 0.3771,
+      "step": 2084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.105921601334446e-05,
+      "loss": 0.4446,
+      "step": 2085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.104531554072839e-05,
+      "loss": 0.4631,
+      "step": 2086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.103141506811232e-05,
+      "loss": 0.449,
+      "step": 2087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.101751459549625e-05,
+      "loss": 0.4498,
+      "step": 2088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.100361412288018e-05,
+      "loss": 0.4325,
+      "step": 2089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.098971365026411e-05,
+      "loss": 0.3925,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.097581317764804e-05,
+      "loss": 0.3468,
+      "step": 2091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.096191270503197e-05,
+      "loss": 0.4689,
+      "step": 2092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.094801223241591e-05,
+      "loss": 0.4182,
+      "step": 2093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.093411175979984e-05,
+      "loss": 0.4614,
+      "step": 2094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.092021128718377e-05,
+      "loss": 0.4343,
+      "step": 2095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.09063108145677e-05,
+      "loss": 0.4106,
+      "step": 2096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.089241034195163e-05,
+      "loss": 0.4368,
+      "step": 2097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.087850986933556e-05,
+      "loss": 0.3633,
+      "step": 2098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.086460939671949e-05,
+      "loss": 0.4025,
+      "step": 2099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.085070892410342e-05,
+      "loss": 0.4232,
+      "step": 2100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.083680845148735e-05,
+      "loss": 0.4275,
+      "step": 2101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.082290797887128e-05,
+      "loss": 0.3947,
+      "step": 2102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.080900750625523e-05,
+      "loss": 0.4132,
+      "step": 2103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.079510703363916e-05,
+      "loss": 0.4657,
+      "step": 2104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.078120656102309e-05,
+      "loss": 0.4531,
+      "step": 2105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.0767306088407e-05,
+      "loss": 0.4209,
+      "step": 2106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.075340561579093e-05,
+      "loss": 0.4331,
+      "step": 2107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.073950514317486e-05,
+      "loss": 0.354,
+      "step": 2108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.072560467055879e-05,
+      "loss": 0.4038,
+      "step": 2109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.071170419794272e-05,
+      "loss": 0.406,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.069780372532667e-05,
+      "loss": 0.4313,
+      "step": 2111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.06839032527106e-05,
+      "loss": 0.4053,
+      "step": 2112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.067000278009453e-05,
+      "loss": 0.4271,
+      "step": 2113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.065610230747846e-05,
+      "loss": 0.3912,
+      "step": 2114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.064220183486239e-05,
+      "loss": 0.52,
+      "step": 2115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.062830136224632e-05,
+      "loss": 0.4045,
+      "step": 2116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.061440088963025e-05,
+      "loss": 0.3879,
+      "step": 2117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.060050041701418e-05,
+      "loss": 0.4301,
+      "step": 2118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.058659994439811e-05,
+      "loss": 0.428,
+      "step": 2119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.057269947178205e-05,
+      "loss": 0.4202,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.055879899916598e-05,
+      "loss": 0.3667,
+      "step": 2121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.054489852654991e-05,
+      "loss": 0.4184,
+      "step": 2122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.053099805393384e-05,
+      "loss": 0.381,
+      "step": 2123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.051709758131777e-05,
+      "loss": 0.311,
+      "step": 2124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.05031971087017e-05,
+      "loss": 0.4628,
+      "step": 2125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.048929663608563e-05,
+      "loss": 0.4462,
+      "step": 2126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.047539616346956e-05,
+      "loss": 0.4137,
+      "step": 2127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.046149569085349e-05,
+      "loss": 0.442,
+      "step": 2128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.044759521823742e-05,
+      "loss": 0.4488,
+      "step": 2129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.043369474562137e-05,
+      "loss": 0.4083,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.04197942730053e-05,
+      "loss": 0.4478,
+      "step": 2131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.040589380038923e-05,
+      "loss": 0.3979,
+      "step": 2132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.039199332777314e-05,
+      "loss": 0.4023,
+      "step": 2133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.037809285515707e-05,
+      "loss": 0.416,
+      "step": 2134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.0364192382541e-05,
+      "loss": 0.4283,
+      "step": 2135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.035029190992493e-05,
+      "loss": 0.3668,
+      "step": 2136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.033639143730886e-05,
+      "loss": 0.3733,
+      "step": 2137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.03224909646928e-05,
+      "loss": 0.3849,
+      "step": 2138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.030859049207674e-05,
+      "loss": 0.4452,
+      "step": 2139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.029469001946067e-05,
+      "loss": 0.4137,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.02807895468446e-05,
+      "loss": 0.4514,
+      "step": 2141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.026688907422853e-05,
+      "loss": 0.4319,
+      "step": 2142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.025298860161246e-05,
+      "loss": 0.3923,
+      "step": 2143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.023908812899639e-05,
+      "loss": 0.4525,
+      "step": 2144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.022518765638032e-05,
+      "loss": 0.3596,
+      "step": 2145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.021128718376425e-05,
+      "loss": 0.4066,
+      "step": 2146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.019738671114819e-05,
+      "loss": 0.4209,
+      "step": 2147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.018348623853212e-05,
+      "loss": 0.4206,
+      "step": 2148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.016958576591605e-05,
+      "loss": 0.4242,
+      "step": 2149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.015568529329998e-05,
+      "loss": 0.4829,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.014178482068391e-05,
+      "loss": 0.4438,
+      "step": 2151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.012788434806784e-05,
+      "loss": 0.4479,
+      "step": 2152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.011398387545177e-05,
+      "loss": 0.3983,
+      "step": 2153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.01000834028357e-05,
+      "loss": 0.4,
+      "step": 2154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.008618293021963e-05,
+      "loss": 0.4301,
+      "step": 2155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.007228245760356e-05,
+      "loss": 0.4489,
+      "step": 2156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.00583819849875e-05,
+      "loss": 0.3667,
+      "step": 2157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.004448151237143e-05,
+      "loss": 0.3881,
+      "step": 2158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.003058103975535e-05,
+      "loss": 0.3519,
+      "step": 2159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.001668056713928e-05,
+      "loss": 0.3733,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.000278009452321e-05,
+      "loss": 0.4357,
+      "step": 2161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.998887962190714e-05,
+      "loss": 0.3942,
+      "step": 2162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.997497914929107e-05,
+      "loss": 0.477,
+      "step": 2163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9961078676675e-05,
+      "loss": 0.3979,
+      "step": 2164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.994717820405894e-05,
+      "loss": 0.4824,
+      "step": 2165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.993327773144287e-05,
+      "loss": 0.4111,
+      "step": 2166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.99193772588268e-05,
+      "loss": 0.4165,
+      "step": 2167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.990547678621073e-05,
+      "loss": 0.3994,
+      "step": 2168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.989157631359466e-05,
+      "loss": 0.3681,
+      "step": 2169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.98776758409786e-05,
+      "loss": 0.4115,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.986377536836252e-05,
+      "loss": 0.4308,
+      "step": 2171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.984987489574645e-05,
+      "loss": 0.4316,
+      "step": 2172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.983597442313038e-05,
+      "loss": 0.4599,
+      "step": 2173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.982207395051433e-05,
+      "loss": 0.4113,
+      "step": 2174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.980817347789826e-05,
+      "loss": 0.4116,
+      "step": 2175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.979427300528219e-05,
+      "loss": 0.4315,
+      "step": 2176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.978037253266612e-05,
+      "loss": 0.4537,
+      "step": 2177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.976647206005005e-05,
+      "loss": 0.4137,
+      "step": 2178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.975257158743398e-05,
+      "loss": 0.462,
+      "step": 2179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.973867111481791e-05,
+      "loss": 0.402,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.972477064220184e-05,
+      "loss": 0.4365,
+      "step": 2181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.971087016958577e-05,
+      "loss": 0.3661,
+      "step": 2182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.96969696969697e-05,
+      "loss": 0.3704,
+      "step": 2183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.968306922435364e-05,
+      "loss": 0.4141,
+      "step": 2184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.966916875173757e-05,
+      "loss": 0.4507,
+      "step": 2185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.965526827912149e-05,
+      "loss": 0.4091,
+      "step": 2186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.964136780650542e-05,
+      "loss": 0.436,
+      "step": 2187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.962746733388935e-05,
+      "loss": 0.4329,
+      "step": 2188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.961356686127328e-05,
+      "loss": 0.3665,
+      "step": 2189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.959966638865721e-05,
+      "loss": 0.4249,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.958576591604114e-05,
+      "loss": 0.3887,
+      "step": 2191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.957186544342508e-05,
+      "loss": 0.417,
+      "step": 2192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.955796497080901e-05,
+      "loss": 0.4536,
+      "step": 2193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.954406449819294e-05,
+      "loss": 0.4469,
+      "step": 2194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.953016402557687e-05,
+      "loss": 0.3977,
+      "step": 2195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.95162635529608e-05,
+      "loss": 0.4572,
+      "step": 2196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.950236308034473e-05,
+      "loss": 0.5035,
+      "step": 2197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.948846260772866e-05,
+      "loss": 0.4348,
+      "step": 2198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.94745621351126e-05,
+      "loss": 0.3918,
+      "step": 2199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.946066166249652e-05,
+      "loss": 0.4185,
+      "step": 2200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.944676118988045e-05,
+      "loss": 0.3874,
+      "step": 2201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.94328607172644e-05,
+      "loss": 0.4116,
+      "step": 2202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.941896024464833e-05,
+      "loss": 0.3802,
+      "step": 2203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.940505977203226e-05,
+      "loss": 0.4214,
+      "step": 2204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.939115929941619e-05,
+      "loss": 0.4473,
+      "step": 2205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.937725882680012e-05,
+      "loss": 0.4007,
+      "step": 2206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.936335835418405e-05,
+      "loss": 0.4309,
+      "step": 2207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.934945788156798e-05,
+      "loss": 0.3707,
+      "step": 2208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.933555740895191e-05,
+      "loss": 0.3642,
+      "step": 2209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.932165693633584e-05,
+      "loss": 0.3683,
+      "step": 2210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.930775646371978e-05,
+      "loss": 0.3943,
+      "step": 2211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.92938559911037e-05,
+      "loss": 0.4161,
+      "step": 2212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.927995551848763e-05,
+      "loss": 0.3993,
+      "step": 2213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.926605504587156e-05,
+      "loss": 0.4171,
+      "step": 2214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.925215457325549e-05,
+      "loss": 0.4508,
+      "step": 2215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.923825410063942e-05,
+      "loss": 0.4436,
+      "step": 2216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.922435362802335e-05,
+      "loss": 0.4323,
+      "step": 2217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.921045315540728e-05,
+      "loss": 0.3975,
+      "step": 2218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.919655268279122e-05,
+      "loss": 0.4294,
+      "step": 2219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.918265221017515e-05,
+      "loss": 0.381,
+      "step": 2220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.916875173755908e-05,
+      "loss": 0.4333,
+      "step": 2221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.915485126494301e-05,
+      "loss": 0.4363,
+      "step": 2222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.914095079232694e-05,
+      "loss": 0.3765,
+      "step": 2223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.912705031971087e-05,
+      "loss": 0.463,
+      "step": 2224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.91131498470948e-05,
+      "loss": 0.3984,
+      "step": 2225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.909924937447873e-05,
+      "loss": 0.4155,
+      "step": 2226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.908534890186266e-05,
+      "loss": 0.4201,
+      "step": 2227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.907144842924659e-05,
+      "loss": 0.3933,
+      "step": 2228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.905754795663054e-05,
+      "loss": 0.3869,
+      "step": 2229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.904364748401447e-05,
+      "loss": 0.3871,
+      "step": 2230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.90297470113984e-05,
+      "loss": 0.4637,
+      "step": 2231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.901584653878233e-05,
+      "loss": 0.4212,
+      "step": 2232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.900194606616626e-05,
+      "loss": 0.3921,
+      "step": 2233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.898804559355019e-05,
+      "loss": 0.424,
+      "step": 2234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.897414512093412e-05,
+      "loss": 0.4083,
+      "step": 2235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.896024464831805e-05,
+      "loss": 0.3551,
+      "step": 2236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.894634417570198e-05,
+      "loss": 0.4019,
+      "step": 2237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.893244370308591e-05,
+      "loss": 0.4571,
+      "step": 2238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.891854323046984e-05,
+      "loss": 0.3761,
+      "step": 2239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.890464275785377e-05,
+      "loss": 0.4649,
+      "step": 2240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.88907422852377e-05,
+      "loss": 0.469,
+      "step": 2241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.887684181262163e-05,
+      "loss": 0.3439,
+      "step": 2242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.886294134000556e-05,
+      "loss": 0.449,
+      "step": 2243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.884904086738949e-05,
+      "loss": 0.377,
+      "step": 2244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.883514039477342e-05,
+      "loss": 0.414,
+      "step": 2245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.882123992215736e-05,
+      "loss": 0.4294,
+      "step": 2246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.880733944954129e-05,
+      "loss": 0.3935,
+      "step": 2247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.879343897692522e-05,
+      "loss": 0.4175,
+      "step": 2248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.877953850430915e-05,
+      "loss": 0.4121,
+      "step": 2249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.876563803169308e-05,
+      "loss": 0.4629,
+      "step": 2250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.875173755907701e-05,
+      "loss": 0.3925,
+      "step": 2251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.873783708646094e-05,
+      "loss": 0.4488,
+      "step": 2252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.872393661384487e-05,
+      "loss": 0.3718,
+      "step": 2253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.87100361412288e-05,
+      "loss": 0.4056,
+      "step": 2254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.869613566861273e-05,
+      "loss": 0.4268,
+      "step": 2255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.868223519599668e-05,
+      "loss": 0.5505,
+      "step": 2256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.86683347233806e-05,
+      "loss": 0.4051,
+      "step": 2257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.865443425076454e-05,
+      "loss": 0.3921,
+      "step": 2258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.864053377814847e-05,
+      "loss": 0.3855,
+      "step": 2259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.86266333055324e-05,
+      "loss": 0.3711,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.861273283291633e-05,
+      "loss": 0.398,
+      "step": 2261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.859883236030026e-05,
+      "loss": 0.3764,
+      "step": 2262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.858493188768419e-05,
+      "loss": 0.3786,
+      "step": 2263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.857103141506812e-05,
+      "loss": 0.521,
+      "step": 2264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.855713094245205e-05,
+      "loss": 0.4205,
+      "step": 2265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.854323046983598e-05,
+      "loss": 0.3933,
+      "step": 2266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.85293299972199e-05,
+      "loss": 0.4637,
+      "step": 2267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.851542952460384e-05,
+      "loss": 0.4,
+      "step": 2268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.850152905198777e-05,
+      "loss": 0.4145,
+      "step": 2269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.84876285793717e-05,
+      "loss": 0.3982,
+      "step": 2270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.847372810675563e-05,
+      "loss": 0.4075,
+      "step": 2271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.845982763413956e-05,
+      "loss": 0.4378,
+      "step": 2272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.84459271615235e-05,
+      "loss": 0.4292,
+      "step": 2273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.843202668890743e-05,
+      "loss": 0.4217,
+      "step": 2274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.841812621629136e-05,
+      "loss": 0.4128,
+      "step": 2275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.840422574367529e-05,
+      "loss": 0.3885,
+      "step": 2276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.839032527105922e-05,
+      "loss": 0.4053,
+      "step": 2277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.837642479844315e-05,
+      "loss": 0.4066,
+      "step": 2278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.836252432582708e-05,
+      "loss": 0.4186,
+      "step": 2279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.834862385321101e-05,
+      "loss": 0.4266,
+      "step": 2280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.833472338059494e-05,
+      "loss": 0.4229,
+      "step": 2281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.832082290797887e-05,
+      "loss": 0.3736,
+      "step": 2282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.830692243536281e-05,
+      "loss": 0.3806,
+      "step": 2283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.829302196274674e-05,
+      "loss": 0.36,
+      "step": 2284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.827912149013067e-05,
+      "loss": 0.4125,
+      "step": 2285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.82652210175146e-05,
+      "loss": 0.4132,
+      "step": 2286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.825132054489853e-05,
+      "loss": 0.4606,
+      "step": 2287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.823742007228246e-05,
+      "loss": 0.4268,
+      "step": 2288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.82235195996664e-05,
+      "loss": 0.4175,
+      "step": 2289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.820961912705032e-05,
+      "loss": 0.4036,
+      "step": 2290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.819571865443425e-05,
+      "loss": 0.4134,
+      "step": 2291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.818181818181818e-05,
+      "loss": 0.3459,
+      "step": 2292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.816791770920211e-05,
+      "loss": 0.4272,
+      "step": 2293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.815401723658604e-05,
+      "loss": 0.4572,
+      "step": 2294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.814011676396997e-05,
+      "loss": 0.4247,
+      "step": 2295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.81262162913539e-05,
+      "loss": 0.3777,
+      "step": 2296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.811231581873783e-05,
+      "loss": 0.3737,
+      "step": 2297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.809841534612176e-05,
+      "loss": 0.3719,
+      "step": 2298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.80845148735057e-05,
+      "loss": 0.3738,
+      "step": 2299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.807061440088964e-05,
+      "loss": 0.399,
+      "step": 2300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.805671392827357e-05,
+      "loss": 0.4076,
+      "step": 2301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.80428134556575e-05,
+      "loss": 0.3783,
+      "step": 2302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.802891298304143e-05,
+      "loss": 0.4077,
+      "step": 2303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.801501251042536e-05,
+      "loss": 0.4147,
+      "step": 2304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.800111203780929e-05,
+      "loss": 0.3986,
+      "step": 2305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.798721156519322e-05,
+      "loss": 0.427,
+      "step": 2306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.797331109257715e-05,
+      "loss": 0.4286,
+      "step": 2307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.795941061996108e-05,
+      "loss": 0.3312,
+      "step": 2308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.794551014734501e-05,
+      "loss": 0.3974,
+      "step": 2309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.793160967472895e-05,
+      "loss": 0.4197,
+      "step": 2310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.791770920211288e-05,
+      "loss": 0.4111,
+      "step": 2311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.790380872949681e-05,
+      "loss": 0.4156,
+      "step": 2312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.788990825688074e-05,
+      "loss": 0.4198,
+      "step": 2313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.787600778426467e-05,
+      "loss": 0.4182,
+      "step": 2314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.78621073116486e-05,
+      "loss": 0.4321,
+      "step": 2315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.784820683903253e-05,
+      "loss": 0.4343,
+      "step": 2316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.783430636641646e-05,
+      "loss": 0.3907,
+      "step": 2317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.78204058938004e-05,
+      "loss": 0.387,
+      "step": 2318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.780650542118432e-05,
+      "loss": 0.4135,
+      "step": 2319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.779260494856825e-05,
+      "loss": 0.4225,
+      "step": 2320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.777870447595218e-05,
+      "loss": 0.3779,
+      "step": 2321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.776480400333611e-05,
+      "loss": 0.3835,
+      "step": 2322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.775090353072004e-05,
+      "loss": 0.3662,
+      "step": 2323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.773700305810397e-05,
+      "loss": 0.4792,
+      "step": 2324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.77231025854879e-05,
+      "loss": 0.4201,
+      "step": 2325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.770920211287183e-05,
+      "loss": 0.4286,
+      "step": 2326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.769530164025576e-05,
+      "loss": 0.432,
+      "step": 2327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.768140116763971e-05,
+      "loss": 0.398,
+      "step": 2328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.766750069502364e-05,
+      "loss": 0.3925,
+      "step": 2329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.765360022240757e-05,
+      "loss": 0.3975,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.76396997497915e-05,
+      "loss": 0.4014,
+      "step": 2331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.762579927717543e-05,
+      "loss": 0.4051,
+      "step": 2332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.761189880455936e-05,
+      "loss": 0.47,
+      "step": 2333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.759799833194329e-05,
+      "loss": 0.4148,
+      "step": 2334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.758409785932722e-05,
+      "loss": 0.3815,
+      "step": 2335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.757019738671115e-05,
+      "loss": 0.4345,
+      "step": 2336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.755629691409509e-05,
+      "loss": 0.3887,
+      "step": 2337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.754239644147902e-05,
+      "loss": 0.4097,
+      "step": 2338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.752849596886295e-05,
+      "loss": 0.3825,
+      "step": 2339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.751459549624688e-05,
+      "loss": 0.3608,
+      "step": 2340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.750069502363081e-05,
+      "loss": 0.4311,
+      "step": 2341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.748679455101474e-05,
+      "loss": 0.4214,
+      "step": 2342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.747289407839867e-05,
+      "loss": 0.4159,
+      "step": 2343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.745899360578259e-05,
+      "loss": 0.4324,
+      "step": 2344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.744509313316653e-05,
+      "loss": 0.4068,
+      "step": 2345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.743119266055046e-05,
+      "loss": 0.4451,
+      "step": 2346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.741729218793439e-05,
+      "loss": 0.4445,
+      "step": 2347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.740339171531832e-05,
+      "loss": 0.3813,
+      "step": 2348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.738949124270225e-05,
+      "loss": 0.3892,
+      "step": 2349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.737559077008618e-05,
+      "loss": 0.4267,
+      "step": 2350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.736169029747011e-05,
+      "loss": 0.3578,
+      "step": 2351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.734778982485404e-05,
+      "loss": 0.366,
+      "step": 2352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.733388935223797e-05,
+      "loss": 0.4389,
+      "step": 2353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.73199888796219e-05,
+      "loss": 0.4135,
+      "step": 2354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.730608840700585e-05,
+      "loss": 0.4883,
+      "step": 2355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.729218793438978e-05,
+      "loss": 0.3772,
+      "step": 2356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.72782874617737e-05,
+      "loss": 0.3858,
+      "step": 2357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.726438698915764e-05,
+      "loss": 0.4251,
+      "step": 2358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.725048651654157e-05,
+      "loss": 0.3359,
+      "step": 2359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.72365860439255e-05,
+      "loss": 0.4476,
+      "step": 2360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.722268557130943e-05,
+      "loss": 0.5018,
+      "step": 2361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.720878509869336e-05,
+      "loss": 0.3452,
+      "step": 2362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.719488462607729e-05,
+      "loss": 0.425,
+      "step": 2363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.718098415346123e-05,
+      "loss": 0.3602,
+      "step": 2364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.716708368084516e-05,
+      "loss": 0.3695,
+      "step": 2365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.715318320822909e-05,
+      "loss": 0.4263,
+      "step": 2366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.713928273561302e-05,
+      "loss": 0.4257,
+      "step": 2367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.712538226299695e-05,
+      "loss": 0.4366,
+      "step": 2368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.711148179038088e-05,
+      "loss": 0.4643,
+      "step": 2369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.709758131776481e-05,
+      "loss": 0.4021,
+      "step": 2370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.708368084514873e-05,
+      "loss": 0.3645,
+      "step": 2371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.706978037253267e-05,
+      "loss": 0.3672,
+      "step": 2372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.70558798999166e-05,
+      "loss": 0.375,
+      "step": 2373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.704197942730053e-05,
+      "loss": 0.4392,
+      "step": 2374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.702807895468446e-05,
+      "loss": 0.3813,
+      "step": 2375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.701417848206839e-05,
+      "loss": 0.3766,
+      "step": 2376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.700027800945232e-05,
+      "loss": 0.4352,
+      "step": 2377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.698637753683625e-05,
+      "loss": 0.4564,
+      "step": 2378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.697247706422018e-05,
+      "loss": 0.3749,
+      "step": 2379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.695857659160411e-05,
+      "loss": 0.3747,
+      "step": 2380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.694467611898804e-05,
+      "loss": 0.4066,
+      "step": 2381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.693077564637199e-05,
+      "loss": 0.3767,
+      "step": 2382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.691687517375592e-05,
+      "loss": 0.4462,
+      "step": 2383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.690297470113985e-05,
+      "loss": 0.407,
+      "step": 2384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.688907422852378e-05,
+      "loss": 0.4424,
+      "step": 2385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.68751737559077e-05,
+      "loss": 0.4092,
+      "step": 2386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.686127328329164e-05,
+      "loss": 0.4427,
+      "step": 2387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.684737281067557e-05,
+      "loss": 0.3994,
+      "step": 2388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.68334723380595e-05,
+      "loss": 0.4689,
+      "step": 2389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.681957186544343e-05,
+      "loss": 0.4091,
+      "step": 2390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.680567139282737e-05,
+      "loss": 0.4215,
+      "step": 2391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.67917709202113e-05,
+      "loss": 0.3902,
+      "step": 2392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.677787044759523e-05,
+      "loss": 0.4053,
+      "step": 2393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.676396997497916e-05,
+      "loss": 0.3651,
+      "step": 2394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.675006950236309e-05,
+      "loss": 0.4083,
+      "step": 2395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.673616902974702e-05,
+      "loss": 0.3881,
+      "step": 2396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.672226855713094e-05,
+      "loss": 0.4124,
+      "step": 2397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.670836808451487e-05,
+      "loss": 0.3647,
+      "step": 2398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.669446761189881e-05,
+      "loss": 0.4502,
+      "step": 2399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.668056713928274e-05,
+      "loss": 0.4117,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.3806,
+      "step": 2401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.66527661940506e-05,
+      "loss": 0.3973,
+      "step": 2402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.663886572143453e-05,
+      "loss": 0.375,
+      "step": 2403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.662496524881846e-05,
+      "loss": 0.3405,
+      "step": 2404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.661106477620239e-05,
+      "loss": 0.4142,
+      "step": 2405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.659716430358632e-05,
+      "loss": 0.3646,
+      "step": 2406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.658326383097025e-05,
+      "loss": 0.3915,
+      "step": 2407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.656936335835418e-05,
+      "loss": 0.4133,
+      "step": 2408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.655546288573812e-05,
+      "loss": 0.4203,
+      "step": 2409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.654156241312205e-05,
+      "loss": 0.4384,
+      "step": 2410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.652766194050598e-05,
+      "loss": 0.3472,
+      "step": 2411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.651376146788991e-05,
+      "loss": 0.4035,
+      "step": 2412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.649986099527384e-05,
+      "loss": 0.3523,
+      "step": 2413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.648596052265777e-05,
+      "loss": 0.454,
+      "step": 2414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.64720600500417e-05,
+      "loss": 0.3942,
+      "step": 2415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.645815957742563e-05,
+      "loss": 0.3988,
+      "step": 2416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.644425910480956e-05,
+      "loss": 0.3775,
+      "step": 2417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.64303586321935e-05,
+      "loss": 0.4346,
+      "step": 2418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.641645815957744e-05,
+      "loss": 0.4456,
+      "step": 2419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.640255768696137e-05,
+      "loss": 0.4044,
+      "step": 2420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.63886572143453e-05,
+      "loss": 0.4447,
+      "step": 2421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.637475674172923e-05,
+      "loss": 0.4056,
+      "step": 2422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.636085626911316e-05,
+      "loss": 0.4553,
+      "step": 2423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.634695579649707e-05,
+      "loss": 0.388,
+      "step": 2424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.6333055323881e-05,
+      "loss": 0.3199,
+      "step": 2425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.631915485126495e-05,
+      "loss": 0.4108,
+      "step": 2426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.630525437864888e-05,
+      "loss": 0.3811,
+      "step": 2427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.629135390603281e-05,
+      "loss": 0.4029,
+      "step": 2428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.627745343341674e-05,
+      "loss": 0.3586,
+      "step": 2429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.626355296080067e-05,
+      "loss": 0.4198,
+      "step": 2430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.62496524881846e-05,
+      "loss": 0.4055,
+      "step": 2431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.623575201556853e-05,
+      "loss": 0.4485,
+      "step": 2432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.622185154295246e-05,
+      "loss": 0.3704,
+      "step": 2433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.620795107033639e-05,
+      "loss": 0.4221,
+      "step": 2434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.619405059772032e-05,
+      "loss": 0.4133,
+      "step": 2435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.618015012510426e-05,
+      "loss": 0.4178,
+      "step": 2436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.616624965248819e-05,
+      "loss": 0.4234,
+      "step": 2437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.615234917987212e-05,
+      "loss": 0.3775,
+      "step": 2438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.613844870725605e-05,
+      "loss": 0.4308,
+      "step": 2439
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.612454823463998e-05,
+      "loss": 0.4568,
+      "step": 2440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.611064776202391e-05,
+      "loss": 0.4479,
+      "step": 2441
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.609674728940784e-05,
+      "loss": 0.357,
+      "step": 2442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.608284681679177e-05,
+      "loss": 0.3691,
+      "step": 2443
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.60689463441757e-05,
+      "loss": 0.3808,
+      "step": 2444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.605504587155963e-05,
+      "loss": 0.378,
+      "step": 2445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.604114539894358e-05,
+      "loss": 0.4253,
+      "step": 2446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.602724492632751e-05,
+      "loss": 0.4467,
+      "step": 2447
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.601334445371144e-05,
+      "loss": 0.3471,
+      "step": 2448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.599944398109537e-05,
+      "loss": 0.4784,
+      "step": 2449
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.598554350847928e-05,
+      "loss": 0.3972,
+      "step": 2450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.597164303586321e-05,
+      "loss": 0.4319,
+      "step": 2451
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.595774256324714e-05,
+      "loss": 0.4036,
+      "step": 2452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.594384209063107e-05,
+      "loss": 0.3847,
+      "step": 2453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.592994161801502e-05,
+      "loss": 0.4039,
+      "step": 2454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.591604114539895e-05,
+      "loss": 0.4156,
+      "step": 2455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.590214067278288e-05,
+      "loss": 0.4022,
+      "step": 2456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.588824020016681e-05,
+      "loss": 0.3782,
+      "step": 2457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.587433972755074e-05,
+      "loss": 0.3642,
+      "step": 2458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.586043925493467e-05,
+      "loss": 0.4001,
+      "step": 2459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.58465387823186e-05,
+      "loss": 0.3979,
+      "step": 2460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.583263830970253e-05,
+      "loss": 0.4227,
+      "step": 2461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.581873783708646e-05,
+      "loss": 0.4103,
+      "step": 2462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.58048373644704e-05,
+      "loss": 0.4473,
+      "step": 2463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.579093689185433e-05,
+      "loss": 0.336,
+      "step": 2464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.577703641923826e-05,
+      "loss": 0.4047,
+      "step": 2465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.576313594662219e-05,
+      "loss": 0.3876,
+      "step": 2466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.574923547400612e-05,
+      "loss": 0.3945,
+      "step": 2467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.573533500139005e-05,
+      "loss": 0.3962,
+      "step": 2468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.572143452877398e-05,
+      "loss": 0.3775,
+      "step": 2469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.570753405615791e-05,
+      "loss": 0.3705,
+      "step": 2470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.569363358354184e-05,
+      "loss": 0.374,
+      "step": 2471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.567973311092577e-05,
+      "loss": 0.4281,
+      "step": 2472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.566583263830972e-05,
+      "loss": 0.3434,
+      "step": 2473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.565193216569365e-05,
+      "loss": 0.3964,
+      "step": 2474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.563803169307758e-05,
+      "loss": 0.4322,
+      "step": 2475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.562413122046149e-05,
+      "loss": 0.412,
+      "step": 2476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.561023074784542e-05,
+      "loss": 0.4054,
+      "step": 2477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.559633027522935e-05,
+      "loss": 0.4205,
+      "step": 2478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.558242980261328e-05,
+      "loss": 0.431,
+      "step": 2479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.556852932999721e-05,
+      "loss": 0.3965,
+      "step": 2480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.555462885738116e-05,
+      "loss": 0.367,
+      "step": 2481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.554072838476509e-05,
+      "loss": 0.417,
+      "step": 2482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.552682791214902e-05,
+      "loss": 0.3627,
+      "step": 2483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.551292743953295e-05,
+      "loss": 0.3839,
+      "step": 2484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.549902696691688e-05,
+      "loss": 0.4096,
+      "step": 2485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.54851264943008e-05,
+      "loss": 0.4297,
+      "step": 2486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.547122602168474e-05,
+      "loss": 0.4164,
+      "step": 2487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.545732554906867e-05,
+      "loss": 0.4031,
+      "step": 2488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.54434250764526e-05,
+      "loss": 0.4474,
+      "step": 2489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.542952460383654e-05,
+      "loss": 0.4159,
+      "step": 2490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.541562413122047e-05,
+      "loss": 0.3665,
+      "step": 2491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.54017236586044e-05,
+      "loss": 0.3781,
+      "step": 2492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.538782318598833e-05,
+      "loss": 0.3605,
+      "step": 2493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.537392271337226e-05,
+      "loss": 0.3793,
+      "step": 2494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.536002224075619e-05,
+      "loss": 0.4418,
+      "step": 2495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.534612176814012e-05,
+      "loss": 0.3931,
+      "step": 2496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.533222129552405e-05,
+      "loss": 0.4534,
+      "step": 2497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.531832082290798e-05,
+      "loss": 0.3569,
+      "step": 2498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.530442035029191e-05,
+      "loss": 0.4313,
+      "step": 2499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.529051987767585e-05,
+      "loss": 0.4054,
+      "step": 2500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.527661940505978e-05,
+      "loss": 0.3944,
+      "step": 2501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.526271893244371e-05,
+      "loss": 0.4099,
+      "step": 2502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.524881845982763e-05,
+      "loss": 0.3667,
+      "step": 2503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.523491798721156e-05,
+      "loss": 0.3265,
+      "step": 2504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.522101751459549e-05,
+      "loss": 0.4389,
+      "step": 2505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.520711704197942e-05,
+      "loss": 0.3938,
+      "step": 2506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.519321656936335e-05,
+      "loss": 0.4086,
+      "step": 2507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.51793160967473e-05,
+      "loss": 0.3406,
+      "step": 2508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.516541562413123e-05,
+      "loss": 0.3559,
+      "step": 2509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.515151515151516e-05,
+      "loss": 0.4639,
+      "step": 2510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.513761467889909e-05,
+      "loss": 0.328,
+      "step": 2511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.512371420628302e-05,
+      "loss": 0.3792,
+      "step": 2512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.510981373366695e-05,
+      "loss": 0.3834,
+      "step": 2513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.509591326105088e-05,
+      "loss": 0.3975,
+      "step": 2514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.50820127884348e-05,
+      "loss": 0.3662,
+      "step": 2515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.506811231581874e-05,
+      "loss": 0.348,
+      "step": 2516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.505421184320268e-05,
+      "loss": 0.338,
+      "step": 2517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.504031137058661e-05,
+      "loss": 0.4126,
+      "step": 2518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.502641089797054e-05,
+      "loss": 0.4339,
+      "step": 2519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.501251042535447e-05,
+      "loss": 0.3312,
+      "step": 2520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.49986099527384e-05,
+      "loss": 0.4029,
+      "step": 2521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.498470948012233e-05,
+      "loss": 0.3779,
+      "step": 2522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.497080900750626e-05,
+      "loss": 0.4235,
+      "step": 2523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.495690853489019e-05,
+      "loss": 0.4135,
+      "step": 2524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.494300806227412e-05,
+      "loss": 0.3942,
+      "step": 2525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.492910758965805e-05,
+      "loss": 0.4263,
+      "step": 2526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.4915207117042e-05,
+      "loss": 0.3622,
+      "step": 2527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.490130664442592e-05,
+      "loss": 0.4283,
+      "step": 2528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.488740617180984e-05,
+      "loss": 0.4302,
+      "step": 2529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.487350569919377e-05,
+      "loss": 0.3607,
+      "step": 2530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.48596052265777e-05,
+      "loss": 0.4343,
+      "step": 2531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.484570475396163e-05,
+      "loss": 0.3578,
+      "step": 2532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.483180428134556e-05,
+      "loss": 0.3996,
+      "step": 2533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.481790380872949e-05,
+      "loss": 0.3861,
+      "step": 2534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.480400333611343e-05,
+      "loss": 0.4012,
+      "step": 2535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.479010286349736e-05,
+      "loss": 0.4323,
+      "step": 2536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.47762023908813e-05,
+      "loss": 0.3943,
+      "step": 2537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.476230191826522e-05,
+      "loss": 0.4673,
+      "step": 2538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.474840144564915e-05,
+      "loss": 0.4253,
+      "step": 2539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.473450097303308e-05,
+      "loss": 0.3992,
+      "step": 2540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.472060050041701e-05,
+      "loss": 0.44,
+      "step": 2541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.470670002780094e-05,
+      "loss": 0.3642,
+      "step": 2542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.469279955518487e-05,
+      "loss": 0.3828,
+      "step": 2543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.46788990825688e-05,
+      "loss": 0.4014,
+      "step": 2544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.466499860995275e-05,
+      "loss": 0.3811,
+      "step": 2545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.465109813733668e-05,
+      "loss": 0.379,
+      "step": 2546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.463719766472061e-05,
+      "loss": 0.4415,
+      "step": 2547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.462329719210454e-05,
+      "loss": 0.4023,
+      "step": 2548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.460939671948847e-05,
+      "loss": 0.3373,
+      "step": 2549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.45954962468724e-05,
+      "loss": 0.3885,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.458159577425633e-05,
+      "loss": 0.3906,
+      "step": 2551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.456769530164026e-05,
+      "loss": 0.3942,
+      "step": 2552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.455379482902419e-05,
+      "loss": 0.4054,
+      "step": 2553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.453989435640813e-05,
+      "loss": 0.4437,
+      "step": 2554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.452599388379206e-05,
+      "loss": 0.4556,
+      "step": 2555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.451209341117598e-05,
+      "loss": 0.3593,
+      "step": 2556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.449819293855991e-05,
+      "loss": 0.383,
+      "step": 2557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.448429246594384e-05,
+      "loss": 0.4156,
+      "step": 2558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.447039199332777e-05,
+      "loss": 0.3605,
+      "step": 2559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.44564915207117e-05,
+      "loss": 0.3223,
+      "step": 2560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.444259104809563e-05,
+      "loss": 0.3979,
+      "step": 2561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.442869057547957e-05,
+      "loss": 0.3396,
+      "step": 2562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.44147901028635e-05,
+      "loss": 0.3644,
+      "step": 2563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.440088963024743e-05,
+      "loss": 0.3868,
+      "step": 2564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.438698915763136e-05,
+      "loss": 0.3671,
+      "step": 2565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.437308868501529e-05,
+      "loss": 0.4126,
+      "step": 2566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.435918821239922e-05,
+      "loss": 0.4575,
+      "step": 2567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.434528773978315e-05,
+      "loss": 0.3633,
+      "step": 2568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.433138726716708e-05,
+      "loss": 0.4507,
+      "step": 2569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.431748679455101e-05,
+      "loss": 0.3653,
+      "step": 2570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.430358632193494e-05,
+      "loss": 0.3781,
+      "step": 2571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.428968584931889e-05,
+      "loss": 0.3618,
+      "step": 2572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.427578537670282e-05,
+      "loss": 0.4192,
+      "step": 2573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.426188490408675e-05,
+      "loss": 0.3871,
+      "step": 2574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.424798443147068e-05,
+      "loss": 0.3854,
+      "step": 2575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.423408395885461e-05,
+      "loss": 0.4223,
+      "step": 2576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.422018348623854e-05,
+      "loss": 0.3873,
+      "step": 2577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.420628301362247e-05,
+      "loss": 0.3774,
+      "step": 2578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.41923825410064e-05,
+      "loss": 0.4507,
+      "step": 2579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.417848206839033e-05,
+      "loss": 0.3651,
+      "step": 2580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.416458159577427e-05,
+      "loss": 0.4198,
+      "step": 2581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.415068112315819e-05,
+      "loss": 0.4068,
+      "step": 2582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.413678065054212e-05,
+      "loss": 0.3446,
+      "step": 2583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.412288017792605e-05,
+      "loss": 0.3765,
+      "step": 2584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.410897970530998e-05,
+      "loss": 0.3292,
+      "step": 2585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.409507923269391e-05,
+      "loss": 0.363,
+      "step": 2586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.408117876007784e-05,
+      "loss": 0.3861,
+      "step": 2587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.406727828746177e-05,
+      "loss": 0.3696,
+      "step": 2588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.405337781484571e-05,
+      "loss": 0.3959,
+      "step": 2589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.403947734222964e-05,
+      "loss": 0.4105,
+      "step": 2590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.402557686961357e-05,
+      "loss": 0.3263,
+      "step": 2591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.40116763969975e-05,
+      "loss": 0.3827,
+      "step": 2592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.399777592438143e-05,
+      "loss": 0.4303,
+      "step": 2593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.398387545176536e-05,
+      "loss": 0.3951,
+      "step": 2594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.396997497914929e-05,
+      "loss": 0.3892,
+      "step": 2595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.395607450653322e-05,
+      "loss": 0.4319,
+      "step": 2596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.394217403391715e-05,
+      "loss": 0.3774,
+      "step": 2597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.392827356130108e-05,
+      "loss": 0.3854,
+      "step": 2598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.391437308868503e-05,
+      "loss": 0.3932,
+      "step": 2599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.390047261606896e-05,
+      "loss": 0.3922,
+      "step": 2600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.388657214345289e-05,
+      "loss": 0.4472,
+      "step": 2601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.387267167083682e-05,
+      "loss": 0.3692,
+      "step": 2602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.385877119822075e-05,
+      "loss": 0.3556,
+      "step": 2603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.384487072560468e-05,
+      "loss": 0.4201,
+      "step": 2604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.38309702529886e-05,
+      "loss": 0.403,
+      "step": 2605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.381706978037254e-05,
+      "loss": 0.399,
+      "step": 2606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.380316930775647e-05,
+      "loss": 0.3946,
+      "step": 2607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.378926883514041e-05,
+      "loss": 0.3651,
+      "step": 2608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.377536836252433e-05,
+      "loss": 0.3838,
+      "step": 2609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.376146788990826e-05,
+      "loss": 0.4036,
+      "step": 2610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.374756741729219e-05,
+      "loss": 0.4226,
+      "step": 2611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.373366694467612e-05,
+      "loss": 0.367,
+      "step": 2612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.371976647206005e-05,
+      "loss": 0.4056,
+      "step": 2613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.370586599944398e-05,
+      "loss": 0.3711,
+      "step": 2614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.369196552682791e-05,
+      "loss": 0.4546,
+      "step": 2615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.367806505421185e-05,
+      "loss": 0.3908,
+      "step": 2616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.366416458159578e-05,
+      "loss": 0.363,
+      "step": 2617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.365026410897971e-05,
+      "loss": 0.3849,
+      "step": 2618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.363636363636364e-05,
+      "loss": 0.4916,
+      "step": 2619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.362246316374757e-05,
+      "loss": 0.3625,
+      "step": 2620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.36085626911315e-05,
+      "loss": 0.4183,
+      "step": 2621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.359466221851543e-05,
+      "loss": 0.4161,
+      "step": 2622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.358076174589936e-05,
+      "loss": 0.4185,
+      "step": 2623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.356686127328329e-05,
+      "loss": 0.4407,
+      "step": 2624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.355296080066722e-05,
+      "loss": 0.4376,
+      "step": 2625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.353906032805116e-05,
+      "loss": 0.4156,
+      "step": 2626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.35251598554351e-05,
+      "loss": 0.364,
+      "step": 2627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.351125938281902e-05,
+      "loss": 0.417,
+      "step": 2628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.349735891020295e-05,
+      "loss": 0.3364,
+      "step": 2629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.348345843758688e-05,
+      "loss": 0.4008,
+      "step": 2630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.346955796497081e-05,
+      "loss": 0.3113,
+      "step": 2631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.345565749235475e-05,
+      "loss": 0.3672,
+      "step": 2632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.344175701973868e-05,
+      "loss": 0.3752,
+      "step": 2633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.34278565471226e-05,
+      "loss": 0.4617,
+      "step": 2634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.341395607450654e-05,
+      "loss": 0.3772,
+      "step": 2635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.340005560189047e-05,
+      "loss": 0.3722,
+      "step": 2636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.33861551292744e-05,
+      "loss": 0.3829,
+      "step": 2637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.337225465665833e-05,
+      "loss": 0.4362,
+      "step": 2638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.335835418404226e-05,
+      "loss": 0.3341,
+      "step": 2639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.334445371142619e-05,
+      "loss": 0.3878,
+      "step": 2640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.333055323881012e-05,
+      "loss": 0.368,
+      "step": 2641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.331665276619405e-05,
+      "loss": 0.3192,
+      "step": 2642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.330275229357799e-05,
+      "loss": 0.3724,
+      "step": 2643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.328885182096192e-05,
+      "loss": 0.4353,
+      "step": 2644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.327495134834585e-05,
+      "loss": 0.4122,
+      "step": 2645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.326105087572978e-05,
+      "loss": 0.3875,
+      "step": 2646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.324715040311371e-05,
+      "loss": 0.4103,
+      "step": 2647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.323324993049764e-05,
+      "loss": 0.3866,
+      "step": 2648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.321934945788157e-05,
+      "loss": 0.4106,
+      "step": 2649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.32054489852655e-05,
+      "loss": 0.3921,
+      "step": 2650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.319154851264943e-05,
+      "loss": 0.3832,
+      "step": 2651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.317764804003336e-05,
+      "loss": 0.4252,
+      "step": 2652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.31637475674173e-05,
+      "loss": 0.3749,
+      "step": 2653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.314984709480123e-05,
+      "loss": 0.4234,
+      "step": 2654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.313594662218516e-05,
+      "loss": 0.379,
+      "step": 2655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.31220461495691e-05,
+      "loss": 0.4439,
+      "step": 2656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.310814567695302e-05,
+      "loss": 0.3784,
+      "step": 2657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.309424520433695e-05,
+      "loss": 0.3491,
+      "step": 2658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.308034473172088e-05,
+      "loss": 0.3547,
+      "step": 2659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.306644425910481e-05,
+      "loss": 0.4248,
+      "step": 2660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.305254378648874e-05,
+      "loss": 0.4323,
+      "step": 2661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.303864331387267e-05,
+      "loss": 0.4146,
+      "step": 2662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.30247428412566e-05,
+      "loss": 0.4251,
+      "step": 2663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.301084236864053e-05,
+      "loss": 0.3866,
+      "step": 2664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.299694189602446e-05,
+      "loss": 0.3943,
+      "step": 2665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.29830414234084e-05,
+      "loss": 0.3895,
+      "step": 2666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.296914095079232e-05,
+      "loss": 0.3496,
+      "step": 2667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.295524047817625e-05,
+      "loss": 0.4354,
+      "step": 2668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.294134000556018e-05,
+      "loss": 0.3298,
+      "step": 2669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.292743953294411e-05,
+      "loss": 0.4044,
+      "step": 2670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.291353906032806e-05,
+      "loss": 0.3717,
+      "step": 2671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.289963858771199e-05,
+      "loss": 0.3487,
+      "step": 2672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.288573811509592e-05,
+      "loss": 0.388,
+      "step": 2673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.287183764247985e-05,
+      "loss": 0.3311,
+      "step": 2674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.285793716986378e-05,
+      "loss": 0.3813,
+      "step": 2675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.284403669724771e-05,
+      "loss": 0.4283,
+      "step": 2676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.283013622463164e-05,
+      "loss": 0.4491,
+      "step": 2677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.281623575201557e-05,
+      "loss": 0.3421,
+      "step": 2678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.28023352793995e-05,
+      "loss": 0.3535,
+      "step": 2679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.278843480678344e-05,
+      "loss": 0.4097,
+      "step": 2680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.277453433416737e-05,
+      "loss": 0.3906,
+      "step": 2681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.27606338615513e-05,
+      "loss": 0.3239,
+      "step": 2682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.274673338893523e-05,
+      "loss": 0.3847,
+      "step": 2683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.273283291631916e-05,
+      "loss": 0.3375,
+      "step": 2684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.271893244370309e-05,
+      "loss": 0.4049,
+      "step": 2685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.270503197108702e-05,
+      "loss": 0.3821,
+      "step": 2686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.269113149847095e-05,
+      "loss": 0.4188,
+      "step": 2687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.267723102585488e-05,
+      "loss": 0.3996,
+      "step": 2688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.266333055323881e-05,
+      "loss": 0.4082,
+      "step": 2689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.264943008062274e-05,
+      "loss": 0.4109,
+      "step": 2690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.263552960800667e-05,
+      "loss": 0.3826,
+      "step": 2691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.26216291353906e-05,
+      "loss": 0.431,
+      "step": 2692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.260772866277453e-05,
+      "loss": 0.3515,
+      "step": 2693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.259382819015846e-05,
+      "loss": 0.3923,
+      "step": 2694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.25799277175424e-05,
+      "loss": 0.4478,
+      "step": 2695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.256602724492632e-05,
+      "loss": 0.3375,
+      "step": 2696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.255212677231025e-05,
+      "loss": 0.3922,
+      "step": 2697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.25382262996942e-05,
+      "loss": 0.3579,
+      "step": 2698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.252432582707813e-05,
+      "loss": 0.3696,
+      "step": 2699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.251042535446206e-05,
+      "loss": 0.3869,
+      "step": 2700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.249652488184599e-05,
+      "loss": 0.4201,
+      "step": 2701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.248262440922992e-05,
+      "loss": 0.3488,
+      "step": 2702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.246872393661385e-05,
+      "loss": 0.4052,
+      "step": 2703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.245482346399778e-05,
+      "loss": 0.4035,
+      "step": 2704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.244092299138171e-05,
+      "loss": 0.3973,
+      "step": 2705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.242702251876564e-05,
+      "loss": 0.3986,
+      "step": 2706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.241312204614958e-05,
+      "loss": 0.4014,
+      "step": 2707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.239922157353351e-05,
+      "loss": 0.3984,
+      "step": 2708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.238532110091744e-05,
+      "loss": 0.3466,
+      "step": 2709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.237142062830137e-05,
+      "loss": 0.3587,
+      "step": 2710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.23575201556853e-05,
+      "loss": 0.3911,
+      "step": 2711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.234361968306923e-05,
+      "loss": 0.5022,
+      "step": 2712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.232971921045316e-05,
+      "loss": 0.4145,
+      "step": 2713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.231581873783708e-05,
+      "loss": 0.3753,
+      "step": 2714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.230191826522102e-05,
+      "loss": 0.382,
+      "step": 2715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.228801779260495e-05,
+      "loss": 0.3638,
+      "step": 2716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.227411731998888e-05,
+      "loss": 0.3527,
+      "step": 2717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.226021684737281e-05,
+      "loss": 0.36,
+      "step": 2718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.224631637475674e-05,
+      "loss": 0.4457,
+      "step": 2719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.223241590214067e-05,
+      "loss": 0.3307,
+      "step": 2720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.22185154295246e-05,
+      "loss": 0.3299,
+      "step": 2721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.220461495690853e-05,
+      "loss": 0.3801,
+      "step": 2722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.219071448429246e-05,
+      "loss": 0.3293,
+      "step": 2723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.217681401167639e-05,
+      "loss": 0.4606,
+      "step": 2724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.216291353906034e-05,
+      "loss": 0.3387,
+      "step": 2725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.214901306644427e-05,
+      "loss": 0.4215,
+      "step": 2726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.21351125938282e-05,
+      "loss": 0.3835,
+      "step": 2727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.212121212121213e-05,
+      "loss": 0.4016,
+      "step": 2728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.210731164859606e-05,
+      "loss": 0.452,
+      "step": 2729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.209341117597999e-05,
+      "loss": 0.3446,
+      "step": 2730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.207951070336392e-05,
+      "loss": 0.3857,
+      "step": 2731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.206561023074785e-05,
+      "loss": 0.4148,
+      "step": 2732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.205170975813178e-05,
+      "loss": 0.3562,
+      "step": 2733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.203780928551572e-05,
+      "loss": 0.4164,
+      "step": 2734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.202390881289965e-05,
+      "loss": 0.3574,
+      "step": 2735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.201000834028358e-05,
+      "loss": 0.3917,
+      "step": 2736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.199610786766751e-05,
+      "loss": 0.3762,
+      "step": 2737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.198220739505144e-05,
+      "loss": 0.4427,
+      "step": 2738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.196830692243537e-05,
+      "loss": 0.4685,
+      "step": 2739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.19544064498193e-05,
+      "loss": 0.3968,
+      "step": 2740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.194050597720322e-05,
+      "loss": 0.4245,
+      "step": 2741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.192660550458716e-05,
+      "loss": 0.3767,
+      "step": 2742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.191270503197109e-05,
+      "loss": 0.3856,
+      "step": 2743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.189880455935502e-05,
+      "loss": 0.3548,
+      "step": 2744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.188490408673895e-05,
+      "loss": 0.359,
+      "step": 2745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.187100361412288e-05,
+      "loss": 0.3564,
+      "step": 2746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.185710314150681e-05,
+      "loss": 0.4022,
+      "step": 2747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.184320266889074e-05,
+      "loss": 0.3424,
+      "step": 2748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.182930219627467e-05,
+      "loss": 0.4264,
+      "step": 2749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.18154017236586e-05,
+      "loss": 0.4171,
+      "step": 2750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.180150125104253e-05,
+      "loss": 0.2932,
+      "step": 2751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.178760077842647e-05,
+      "loss": 0.3599,
+      "step": 2752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.17737003058104e-05,
+      "loss": 0.3567,
+      "step": 2753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.175979983319433e-05,
+      "loss": 0.34,
+      "step": 2754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.174589936057826e-05,
+      "loss": 0.3138,
+      "step": 2755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.17319988879622e-05,
+      "loss": 0.3602,
+      "step": 2756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.171809841534612e-05,
+      "loss": 0.374,
+      "step": 2757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.170419794273006e-05,
+      "loss": 0.4142,
+      "step": 2758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.169029747011399e-05,
+      "loss": 0.4571,
+      "step": 2759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.167639699749792e-05,
+      "loss": 0.3808,
+      "step": 2760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.166249652488185e-05,
+      "loss": 0.435,
+      "step": 2761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.164859605226579e-05,
+      "loss": 0.4094,
+      "step": 2762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.163469557964972e-05,
+      "loss": 0.3818,
+      "step": 2763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.162079510703365e-05,
+      "loss": 0.3726,
+      "step": 2764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.160689463441758e-05,
+      "loss": 0.3663,
+      "step": 2765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.159299416180151e-05,
+      "loss": 0.4384,
+      "step": 2766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.157909368918543e-05,
+      "loss": 0.3491,
+      "step": 2767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.156519321656936e-05,
+      "loss": 0.3186,
+      "step": 2768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.15512927439533e-05,
+      "loss": 0.4031,
+      "step": 2769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.153739227133723e-05,
+      "loss": 0.3786,
+      "step": 2770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.152349179872116e-05,
+      "loss": 0.3788,
+      "step": 2771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.150959132610509e-05,
+      "loss": 0.4177,
+      "step": 2772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.149569085348902e-05,
+      "loss": 0.4267,
+      "step": 2773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.148179038087295e-05,
+      "loss": 0.3376,
+      "step": 2774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.146788990825688e-05,
+      "loss": 0.4339,
+      "step": 2775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.145398943564081e-05,
+      "loss": 0.337,
+      "step": 2776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.144008896302474e-05,
+      "loss": 0.3724,
+      "step": 2777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.142618849040867e-05,
+      "loss": 0.359,
+      "step": 2778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.141228801779261e-05,
+      "loss": 0.3548,
+      "step": 2779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.139838754517654e-05,
+      "loss": 0.4365,
+      "step": 2780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.138448707256047e-05,
+      "loss": 0.4022,
+      "step": 2781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.13705865999444e-05,
+      "loss": 0.4285,
+      "step": 2782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.135668612732833e-05,
+      "loss": 0.3553,
+      "step": 2783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.134278565471226e-05,
+      "loss": 0.3704,
+      "step": 2784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.13288851820962e-05,
+      "loss": 0.3537,
+      "step": 2785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.131498470948012e-05,
+      "loss": 0.3574,
+      "step": 2786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.130108423686405e-05,
+      "loss": 0.3496,
+      "step": 2787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.128718376424798e-05,
+      "loss": 0.3594,
+      "step": 2788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.127328329163193e-05,
+      "loss": 0.3628,
+      "step": 2789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.125938281901586e-05,
+      "loss": 0.4184,
+      "step": 2790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.124548234639979e-05,
+      "loss": 0.3864,
+      "step": 2791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.123158187378372e-05,
+      "loss": 0.364,
+      "step": 2792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.121768140116765e-05,
+      "loss": 0.3907,
+      "step": 2793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.120378092855156e-05,
+      "loss": 0.4457,
+      "step": 2794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.11898804559355e-05,
+      "loss": 0.4495,
+      "step": 2795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.117597998331942e-05,
+      "loss": 0.3879,
+      "step": 2796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.116207951070337e-05,
+      "loss": 0.3521,
+      "step": 2797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.11481790380873e-05,
+      "loss": 0.3713,
+      "step": 2798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.113427856547123e-05,
+      "loss": 0.4273,
+      "step": 2799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.112037809285516e-05,
+      "loss": 0.4173,
+      "step": 2800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.110647762023909e-05,
+      "loss": 0.4003,
+      "step": 2801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.109257714762302e-05,
+      "loss": 0.383,
+      "step": 2802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.107867667500695e-05,
+      "loss": 0.4229,
+      "step": 2803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.106477620239088e-05,
+      "loss": 0.3962,
+      "step": 2804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.105087572977481e-05,
+      "loss": 0.3687,
+      "step": 2805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.103697525715875e-05,
+      "loss": 0.333,
+      "step": 2806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1023074784542676e-05,
+      "loss": 0.3392,
+      "step": 2807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.1009174311926606e-05,
+      "loss": 0.329,
+      "step": 2808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.099527383931054e-05,
+      "loss": 0.3729,
+      "step": 2809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.098137336669447e-05,
+      "loss": 0.4476,
+      "step": 2810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.09674728940784e-05,
+      "loss": 0.323,
+      "step": 2811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.095357242146233e-05,
+      "loss": 0.3582,
+      "step": 2812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.093967194884627e-05,
+      "loss": 0.3958,
+      "step": 2813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.09257714762302e-05,
+      "loss": 0.3231,
+      "step": 2814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.091187100361413e-05,
+      "loss": 0.3563,
+      "step": 2815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.089797053099806e-05,
+      "loss": 0.4061,
+      "step": 2816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.088407005838199e-05,
+      "loss": 0.4902,
+      "step": 2817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.087016958576593e-05,
+      "loss": 0.4143,
+      "step": 2818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.085626911314986e-05,
+      "loss": 0.3562,
+      "step": 2819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.084236864053377e-05,
+      "loss": 0.4102,
+      "step": 2820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.082846816791771e-05,
+      "loss": 0.3493,
+      "step": 2821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.081456769530164e-05,
+      "loss": 0.3464,
+      "step": 2822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.080066722268557e-05,
+      "loss": 0.3789,
+      "step": 2823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.07867667500695e-05,
+      "loss": 0.3507,
+      "step": 2824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.077286627745343e-05,
+      "loss": 0.3503,
+      "step": 2825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.075896580483737e-05,
+      "loss": 0.3616,
+      "step": 2826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.07450653322213e-05,
+      "loss": 0.3331,
+      "step": 2827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.073116485960523e-05,
+      "loss": 0.4032,
+      "step": 2828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.071726438698916e-05,
+      "loss": 0.3937,
+      "step": 2829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.070336391437309e-05,
+      "loss": 0.3538,
+      "step": 2830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0689463441757024e-05,
+      "loss": 0.4228,
+      "step": 2831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0675562969140954e-05,
+      "loss": 0.4286,
+      "step": 2832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0661662496524884e-05,
+      "loss": 0.389,
+      "step": 2833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0647762023908814e-05,
+      "loss": 0.3733,
+      "step": 2834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0633861551292745e-05,
+      "loss": 0.3573,
+      "step": 2835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.061996107867668e-05,
+      "loss": 0.3611,
+      "step": 2836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 0.3907,
+      "step": 2837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.059216013344454e-05,
+      "loss": 0.4147,
+      "step": 2838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.057825966082847e-05,
+      "loss": 0.4065,
+      "step": 2839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.056435918821241e-05,
+      "loss": 0.4418,
+      "step": 2840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.055045871559634e-05,
+      "loss": 0.3724,
+      "step": 2841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.053655824298027e-05,
+      "loss": 0.4196,
+      "step": 2842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.05226577703642e-05,
+      "loss": 0.32,
+      "step": 2843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.050875729774813e-05,
+      "loss": 0.39,
+      "step": 2844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0494856825132065e-05,
+      "loss": 0.3453,
+      "step": 2845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0480956352515996e-05,
+      "loss": 0.3843,
+      "step": 2846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.046705587989991e-05,
+      "loss": 0.3435,
+      "step": 2847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.045315540728385e-05,
+      "loss": 0.3925,
+      "step": 2848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.043925493466778e-05,
+      "loss": 0.4439,
+      "step": 2849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.042535446205171e-05,
+      "loss": 0.369,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.041145398943564e-05,
+      "loss": 0.3734,
+      "step": 2851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.039755351681957e-05,
+      "loss": 0.365,
+      "step": 2852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0383653044203506e-05,
+      "loss": 0.3705,
+      "step": 2853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0369752571587436e-05,
+      "loss": 0.3601,
+      "step": 2854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0355852098971366e-05,
+      "loss": 0.4264,
+      "step": 2855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0341951626355296e-05,
+      "loss": 0.3894,
+      "step": 2856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0328051153739226e-05,
+      "loss": 0.3604,
+      "step": 2857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.031415068112316e-05,
+      "loss": 0.3907,
+      "step": 2858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.030025020850709e-05,
+      "loss": 0.4189,
+      "step": 2859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.028634973589102e-05,
+      "loss": 0.4379,
+      "step": 2860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.027244926327495e-05,
+      "loss": 0.3581,
+      "step": 2861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.025854879065888e-05,
+      "loss": 0.3956,
+      "step": 2862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.024464831804282e-05,
+      "loss": 0.4153,
+      "step": 2863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.023074784542675e-05,
+      "loss": 0.3789,
+      "step": 2864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.021684737281068e-05,
+      "loss": 0.3747,
+      "step": 2865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.020294690019461e-05,
+      "loss": 0.3528,
+      "step": 2866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.018904642757854e-05,
+      "loss": 0.4394,
+      "step": 2867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.017514595496248e-05,
+      "loss": 0.3805,
+      "step": 2868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.016124548234641e-05,
+      "loss": 0.3648,
+      "step": 2869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.014734500973034e-05,
+      "loss": 0.3424,
+      "step": 2870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.013344453711427e-05,
+      "loss": 0.3237,
+      "step": 2871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0119544064498204e-05,
+      "loss": 0.3585,
+      "step": 2872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.010564359188212e-05,
+      "loss": 0.3517,
+      "step": 2873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.009174311926605e-05,
+      "loss": 0.3809,
+      "step": 2874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.007784264664999e-05,
+      "loss": 0.4112,
+      "step": 2875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.006394217403392e-05,
+      "loss": 0.3596,
+      "step": 2876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.005004170141785e-05,
+      "loss": 0.4237,
+      "step": 2877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.003614122880178e-05,
+      "loss": 0.4139,
+      "step": 2878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.002224075618571e-05,
+      "loss": 0.388,
+      "step": 2879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0008340283569645e-05,
+      "loss": 0.3699,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9994439810953575e-05,
+      "loss": 0.4172,
+      "step": 2881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9980539338337505e-05,
+      "loss": 0.3516,
+      "step": 2882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9966638865721435e-05,
+      "loss": 0.419,
+      "step": 2883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9952738393105365e-05,
+      "loss": 0.3641,
+      "step": 2884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.99388379204893e-05,
+      "loss": 0.3567,
+      "step": 2885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.992493744787323e-05,
+      "loss": 0.4402,
+      "step": 2886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.991103697525716e-05,
+      "loss": 0.3607,
+      "step": 2887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.989713650264109e-05,
+      "loss": 0.3716,
+      "step": 2888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.988323603002502e-05,
+      "loss": 0.4141,
+      "step": 2889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.986933555740896e-05,
+      "loss": 0.4022,
+      "step": 2890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.985543508479289e-05,
+      "loss": 0.3787,
+      "step": 2891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.984153461217682e-05,
+      "loss": 0.3994,
+      "step": 2892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.982763413956075e-05,
+      "loss": 0.3443,
+      "step": 2893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.981373366694468e-05,
+      "loss": 0.3415,
+      "step": 2894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9799833194328616e-05,
+      "loss": 0.3739,
+      "step": 2895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9785932721712546e-05,
+      "loss": 0.4204,
+      "step": 2896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9772032249096476e-05,
+      "loss": 0.4017,
+      "step": 2897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9758131776480406e-05,
+      "loss": 0.3214,
+      "step": 2898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9744231303864336e-05,
+      "loss": 0.4154,
+      "step": 2899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.973033083124826e-05,
+      "loss": 0.3611,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.971643035863219e-05,
+      "loss": 0.3898,
+      "step": 2901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.970252988601612e-05,
+      "loss": 0.432,
+      "step": 2902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.968862941340006e-05,
+      "loss": 0.3273,
+      "step": 2903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.967472894078399e-05,
+      "loss": 0.3883,
+      "step": 2904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.966082846816792e-05,
+      "loss": 0.3888,
+      "step": 2905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.964692799555185e-05,
+      "loss": 0.3951,
+      "step": 2906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9633027522935784e-05,
+      "loss": 0.3892,
+      "step": 2907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9619127050319714e-05,
+      "loss": 0.4413,
+      "step": 2908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9605226577703644e-05,
+      "loss": 0.4603,
+      "step": 2909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9591326105087574e-05,
+      "loss": 0.361,
+      "step": 2910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9577425632471504e-05,
+      "loss": 0.3733,
+      "step": 2911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.956352515985544e-05,
+      "loss": 0.4019,
+      "step": 2912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.954962468723937e-05,
+      "loss": 0.4242,
+      "step": 2913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.95357242146233e-05,
+      "loss": 0.3467,
+      "step": 2914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.952182374200723e-05,
+      "loss": 0.42,
+      "step": 2915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950792326939116e-05,
+      "loss": 0.3911,
+      "step": 2916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.94940227967751e-05,
+      "loss": 0.3639,
+      "step": 2917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.948012232415903e-05,
+      "loss": 0.3628,
+      "step": 2918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.946622185154296e-05,
+      "loss": 0.3879,
+      "step": 2919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.945232137892689e-05,
+      "loss": 0.392,
+      "step": 2920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.943842090631082e-05,
+      "loss": 0.3621,
+      "step": 2921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9424520433694755e-05,
+      "loss": 0.3771,
+      "step": 2922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9410619961078685e-05,
+      "loss": 0.307,
+      "step": 2923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9396719488462615e-05,
+      "loss": 0.3697,
+      "step": 2924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9382819015846545e-05,
+      "loss": 0.3692,
+      "step": 2925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.936891854323047e-05,
+      "loss": 0.3585,
+      "step": 2926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.93550180706144e-05,
+      "loss": 0.3975,
+      "step": 2927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.934111759799833e-05,
+      "loss": 0.3748,
+      "step": 2928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.932721712538226e-05,
+      "loss": 0.3964,
+      "step": 2929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9313316652766195e-05,
+      "loss": 0.4024,
+      "step": 2930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9299416180150126e-05,
+      "loss": 0.4042,
+      "step": 2931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9285515707534056e-05,
+      "loss": 0.4207,
+      "step": 2932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9271615234917986e-05,
+      "loss": 0.4097,
+      "step": 2933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9257714762301916e-05,
+      "loss": 0.3848,
+      "step": 2934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.924381428968585e-05,
+      "loss": 0.3771,
+      "step": 2935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.922991381706978e-05,
+      "loss": 0.3534,
+      "step": 2936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.921601334445371e-05,
+      "loss": 0.4245,
+      "step": 2937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.920211287183764e-05,
+      "loss": 0.372,
+      "step": 2938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.918821239922158e-05,
+      "loss": 0.3809,
+      "step": 2939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.917431192660551e-05,
+      "loss": 0.4105,
+      "step": 2940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.916041145398944e-05,
+      "loss": 0.3708,
+      "step": 2941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.914651098137337e-05,
+      "loss": 0.4038,
+      "step": 2942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.91326105087573e-05,
+      "loss": 0.4393,
+      "step": 2943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.911871003614124e-05,
+      "loss": 0.4094,
+      "step": 2944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.910480956352517e-05,
+      "loss": 0.4472,
+      "step": 2945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.90909090909091e-05,
+      "loss": 0.3686,
+      "step": 2946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.907700861829303e-05,
+      "loss": 0.3928,
+      "step": 2947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.906310814567696e-05,
+      "loss": 0.3685,
+      "step": 2948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9049207673060894e-05,
+      "loss": 0.3659,
+      "step": 2949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9035307200444824e-05,
+      "loss": 0.4127,
+      "step": 2950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9021406727828754e-05,
+      "loss": 0.4051,
+      "step": 2951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9007506255212684e-05,
+      "loss": 0.395,
+      "step": 2952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.899360578259661e-05,
+      "loss": 0.3779,
+      "step": 2953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.897970530998054e-05,
+      "loss": 0.4277,
+      "step": 2954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.896580483736447e-05,
+      "loss": 0.4044,
+      "step": 2955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.89519043647484e-05,
+      "loss": 0.4366,
+      "step": 2956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8938003892132334e-05,
+      "loss": 0.3929,
+      "step": 2957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8924103419516264e-05,
+      "loss": 0.3671,
+      "step": 2958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8910202946900194e-05,
+      "loss": 0.3676,
+      "step": 2959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8896302474284124e-05,
+      "loss": 0.3635,
+      "step": 2960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8882402001668055e-05,
+      "loss": 0.3662,
+      "step": 2961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.886850152905199e-05,
+      "loss": 0.4121,
+      "step": 2962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.885460105643592e-05,
+      "loss": 0.358,
+      "step": 2963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.884070058381985e-05,
+      "loss": 0.4114,
+      "step": 2964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.882680011120378e-05,
+      "loss": 0.4396,
+      "step": 2965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.881289963858772e-05,
+      "loss": 0.4218,
+      "step": 2966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.879899916597165e-05,
+      "loss": 0.3586,
+      "step": 2967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.878509869335558e-05,
+      "loss": 0.3629,
+      "step": 2968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.877119822073951e-05,
+      "loss": 0.4144,
+      "step": 2969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.875729774812344e-05,
+      "loss": 0.367,
+      "step": 2970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8743397275507376e-05,
+      "loss": 0.3919,
+      "step": 2971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8729496802891306e-05,
+      "loss": 0.3858,
+      "step": 2972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8715596330275236e-05,
+      "loss": 0.335,
+      "step": 2973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8701695857659166e-05,
+      "loss": 0.43,
+      "step": 2974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8687795385043096e-05,
+      "loss": 0.4068,
+      "step": 2975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.867389491242703e-05,
+      "loss": 0.344,
+      "step": 2976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.865999443981096e-05,
+      "loss": 0.3742,
+      "step": 2977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.864609396719489e-05,
+      "loss": 0.4096,
+      "step": 2978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8632193494578816e-05,
+      "loss": 0.3735,
+      "step": 2979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8618293021962746e-05,
+      "loss": 0.3536,
+      "step": 2980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8604392549346676e-05,
+      "loss": 0.386,
+      "step": 2981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8590492076730606e-05,
+      "loss": 0.3304,
+      "step": 2982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8576591604114536e-05,
+      "loss": 0.419,
+      "step": 2983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.856269113149847e-05,
+      "loss": 0.368,
+      "step": 2984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.85487906588824e-05,
+      "loss": 0.3527,
+      "step": 2985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.853489018626633e-05,
+      "loss": 0.3839,
+      "step": 2986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.852098971365026e-05,
+      "loss": 0.3506,
+      "step": 2987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8507089241034193e-05,
+      "loss": 0.396,
+      "step": 2988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.849318876841813e-05,
+      "loss": 0.3422,
+      "step": 2989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.847928829580206e-05,
+      "loss": 0.3389,
+      "step": 2990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.846538782318599e-05,
+      "loss": 0.3853,
+      "step": 2991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.845148735056992e-05,
+      "loss": 0.4118,
+      "step": 2992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.843758687795385e-05,
+      "loss": 0.3891,
+      "step": 2993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842368640533779e-05,
+      "loss": 0.3845,
+      "step": 2994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840978593272172e-05,
+      "loss": 0.3724,
+      "step": 2995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839588546010565e-05,
+      "loss": 0.3888,
+      "step": 2996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838198498748958e-05,
+      "loss": 0.3924,
+      "step": 2997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8368084514873514e-05,
+      "loss": 0.4343,
+      "step": 2998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8354184042257444e-05,
+      "loss": 0.3921,
+      "step": 2999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8340283569641374e-05,
+      "loss": 0.4345,
+      "step": 3000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8326383097025305e-05,
+      "loss": 0.3784,
+      "step": 3001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8312482624409235e-05,
+      "loss": 0.3677,
+      "step": 3002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.829858215179317e-05,
+      "loss": 0.3969,
+      "step": 3003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.82846816791771e-05,
+      "loss": 0.3672,
+      "step": 3004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.827078120656102e-05,
+      "loss": 0.3558,
+      "step": 3005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8256880733944955e-05,
+      "loss": 0.3853,
+      "step": 3006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8242980261328885e-05,
+      "loss": 0.3757,
+      "step": 3007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8229079788712815e-05,
+      "loss": 0.4138,
+      "step": 3008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8215179316096745e-05,
+      "loss": 0.3747,
+      "step": 3009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8201278843480675e-05,
+      "loss": 0.2939,
+      "step": 3010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.818737837086461e-05,
+      "loss": 0.4071,
+      "step": 3011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.817347789824854e-05,
+      "loss": 0.3497,
+      "step": 3012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.815957742563247e-05,
+      "loss": 0.3098,
+      "step": 3013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.81456769530164e-05,
+      "loss": 0.363,
+      "step": 3014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.813177648040033e-05,
+      "loss": 0.3872,
+      "step": 3015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.811787600778427e-05,
+      "loss": 0.3567,
+      "step": 3016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.81039755351682e-05,
+      "loss": 0.3559,
+      "step": 3017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.809007506255213e-05,
+      "loss": 0.4481,
+      "step": 3018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.807617458993606e-05,
+      "loss": 0.3338,
+      "step": 3019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.806227411731999e-05,
+      "loss": 0.3856,
+      "step": 3020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8048373644703926e-05,
+      "loss": 0.3762,
+      "step": 3021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8034473172087856e-05,
+      "loss": 0.3937,
+      "step": 3022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8020572699471786e-05,
+      "loss": 0.5045,
+      "step": 3023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8006672226855716e-05,
+      "loss": 0.3484,
+      "step": 3024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.799277175423965e-05,
+      "loss": 0.3888,
+      "step": 3025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.797887128162358e-05,
+      "loss": 0.3862,
+      "step": 3026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.796497080900751e-05,
+      "loss": 0.3546,
+      "step": 3027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.795107033639144e-05,
+      "loss": 0.3708,
+      "step": 3028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7937169863775373e-05,
+      "loss": 0.3692,
+      "step": 3029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.792326939115931e-05,
+      "loss": 0.364,
+      "step": 3030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.790936891854324e-05,
+      "loss": 0.3352,
+      "step": 3031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.789546844592716e-05,
+      "loss": 0.3239,
+      "step": 3032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7881567973311094e-05,
+      "loss": 0.3529,
+      "step": 3033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7867667500695024e-05,
+      "loss": 0.3832,
+      "step": 3034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7853767028078954e-05,
+      "loss": 0.4201,
+      "step": 3035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7839866555462884e-05,
+      "loss": 0.3844,
+      "step": 3036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7825966082846814e-05,
+      "loss": 0.3892,
+      "step": 3037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.781206561023075e-05,
+      "loss": 0.3621,
+      "step": 3038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.779816513761468e-05,
+      "loss": 0.3482,
+      "step": 3039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.778426466499861e-05,
+      "loss": 0.409,
+      "step": 3040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.777036419238254e-05,
+      "loss": 0.3844,
+      "step": 3041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.775646371976647e-05,
+      "loss": 0.3884,
+      "step": 3042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.774256324715041e-05,
+      "loss": 0.4161,
+      "step": 3043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.772866277453434e-05,
+      "loss": 0.4208,
+      "step": 3044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.771476230191827e-05,
+      "loss": 0.3882,
+      "step": 3045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.77008618293022e-05,
+      "loss": 0.3972,
+      "step": 3046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.768696135668613e-05,
+      "loss": 0.3842,
+      "step": 3047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7673060884070065e-05,
+      "loss": 0.3765,
+      "step": 3048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7659160411453995e-05,
+      "loss": 0.3689,
+      "step": 3049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7645259938837925e-05,
+      "loss": 0.4248,
+      "step": 3050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7631359466221855e-05,
+      "loss": 0.4258,
+      "step": 3051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7617458993605785e-05,
+      "loss": 0.3637,
+      "step": 3052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.760355852098972e-05,
+      "loss": 0.3737,
+      "step": 3053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.758965804837365e-05,
+      "loss": 0.3534,
+      "step": 3054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.757575757575758e-05,
+      "loss": 0.3689,
+      "step": 3055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.756185710314151e-05,
+      "loss": 0.3241,
+      "step": 3056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.754795663052545e-05,
+      "loss": 0.3458,
+      "step": 3057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7534056157909366e-05,
+      "loss": 0.3835,
+      "step": 3058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7520155685293296e-05,
+      "loss": 0.44,
+      "step": 3059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.750625521267723e-05,
+      "loss": 0.4113,
+      "step": 3060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.749235474006116e-05,
+      "loss": 0.3983,
+      "step": 3061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.747845426744509e-05,
+      "loss": 0.4394,
+      "step": 3062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.746455379482902e-05,
+      "loss": 0.3866,
+      "step": 3063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.745065332221295e-05,
+      "loss": 0.397,
+      "step": 3064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.743675284959689e-05,
+      "loss": 0.3581,
+      "step": 3065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.742285237698082e-05,
+      "loss": 0.387,
+      "step": 3066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.740895190436475e-05,
+      "loss": 0.3722,
+      "step": 3067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.739505143174868e-05,
+      "loss": 0.3238,
+      "step": 3068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.738115095913261e-05,
+      "loss": 0.4526,
+      "step": 3069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.736725048651655e-05,
+      "loss": 0.3887,
+      "step": 3070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.735335001390048e-05,
+      "loss": 0.3991,
+      "step": 3071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.733944954128441e-05,
+      "loss": 0.416,
+      "step": 3072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.732554906866834e-05,
+      "loss": 0.3702,
+      "step": 3073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.731164859605227e-05,
+      "loss": 0.348,
+      "step": 3074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7297748123436204e-05,
+      "loss": 0.386,
+      "step": 3075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7283847650820134e-05,
+      "loss": 0.3973,
+      "step": 3076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7269947178204064e-05,
+      "loss": 0.3645,
+      "step": 3077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7256046705587994e-05,
+      "loss": 0.3873,
+      "step": 3078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7242146232971924e-05,
+      "loss": 0.3738,
+      "step": 3079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.722824576035586e-05,
+      "loss": 0.3068,
+      "step": 3080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.721434528773979e-05,
+      "loss": 0.3477,
+      "step": 3081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.720044481512372e-05,
+      "loss": 0.3585,
+      "step": 3082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.718654434250765e-05,
+      "loss": 0.3913,
+      "step": 3083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.717264386989158e-05,
+      "loss": 0.3505,
+      "step": 3084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7158743397275504e-05,
+      "loss": 0.3602,
+      "step": 3085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7144842924659435e-05,
+      "loss": 0.3725,
+      "step": 3086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7130942452043365e-05,
+      "loss": 0.3964,
+      "step": 3087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.71170419794273e-05,
+      "loss": 0.36,
+      "step": 3088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.710314150681123e-05,
+      "loss": 0.4069,
+      "step": 3089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.708924103419516e-05,
+      "loss": 0.3905,
+      "step": 3090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.707534056157909e-05,
+      "loss": 0.3913,
+      "step": 3091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.706144008896303e-05,
+      "loss": 0.3286,
+      "step": 3092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.704753961634696e-05,
+      "loss": 0.3791,
+      "step": 3093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.703363914373089e-05,
+      "loss": 0.3705,
+      "step": 3094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.701973867111482e-05,
+      "loss": 0.3726,
+      "step": 3095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.700583819849875e-05,
+      "loss": 0.3481,
+      "step": 3096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6991937725882686e-05,
+      "loss": 0.402,
+      "step": 3097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6978037253266616e-05,
+      "loss": 0.4039,
+      "step": 3098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6964136780650546e-05,
+      "loss": 0.4196,
+      "step": 3099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6950236308034476e-05,
+      "loss": 0.3736,
+      "step": 3100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6936335835418406e-05,
+      "loss": 0.4058,
+      "step": 3101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.692243536280234e-05,
+      "loss": 0.4024,
+      "step": 3102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.690853489018627e-05,
+      "loss": 0.3581,
+      "step": 3103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.68946344175702e-05,
+      "loss": 0.3396,
+      "step": 3104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.688073394495413e-05,
+      "loss": 0.394,
+      "step": 3105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.686683347233806e-05,
+      "loss": 0.3613,
+      "step": 3106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6852932999722e-05,
+      "loss": 0.428,
+      "step": 3107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.683903252710593e-05,
+      "loss": 0.3892,
+      "step": 3108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.682513205448986e-05,
+      "loss": 0.3945,
+      "step": 3109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.681123158187379e-05,
+      "loss": 0.3669,
+      "step": 3110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.679733110925771e-05,
+      "loss": 0.4157,
+      "step": 3111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.678343063664164e-05,
+      "loss": 0.3895,
+      "step": 3112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.676953016402557e-05,
+      "loss": 0.3529,
+      "step": 3113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6755629691409503e-05,
+      "loss": 0.3296,
+      "step": 3114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.674172921879344e-05,
+      "loss": 0.4043,
+      "step": 3115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.672782874617737e-05,
+      "loss": 0.372,
+      "step": 3116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.67139282735613e-05,
+      "loss": 0.3608,
+      "step": 3117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.670002780094523e-05,
+      "loss": 0.3661,
+      "step": 3118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.668612732832916e-05,
+      "loss": 0.3874,
+      "step": 3119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.66722268557131e-05,
+      "loss": 0.3686,
+      "step": 3120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.665832638309703e-05,
+      "loss": 0.3888,
+      "step": 3121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.664442591048096e-05,
+      "loss": 0.423,
+      "step": 3122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.663052543786489e-05,
+      "loss": 0.4321,
+      "step": 3123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6616624965248824e-05,
+      "loss": 0.3495,
+      "step": 3124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6602724492632754e-05,
+      "loss": 0.3863,
+      "step": 3125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6588824020016685e-05,
+      "loss": 0.3253,
+      "step": 3126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6574923547400615e-05,
+      "loss": 0.3637,
+      "step": 3127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6561023074784545e-05,
+      "loss": 0.3335,
+      "step": 3128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.654712260216848e-05,
+      "loss": 0.3498,
+      "step": 3129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.653322212955241e-05,
+      "loss": 0.38,
+      "step": 3130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.651932165693634e-05,
+      "loss": 0.3725,
+      "step": 3131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.650542118432027e-05,
+      "loss": 0.3514,
+      "step": 3132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.64915207117042e-05,
+      "loss": 0.3366,
+      "step": 3133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.647762023908814e-05,
+      "loss": 0.3778,
+      "step": 3134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.646371976647207e-05,
+      "loss": 0.4125,
+      "step": 3135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6449819293856e-05,
+      "loss": 0.3649,
+      "step": 3136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.643591882123993e-05,
+      "loss": 0.4959,
+      "step": 3137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.642201834862385e-05,
+      "loss": 0.4232,
+      "step": 3138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.640811787600778e-05,
+      "loss": 0.4524,
+      "step": 3139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.639421740339171e-05,
+      "loss": 0.4013,
+      "step": 3140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.638031693077564e-05,
+      "loss": 0.4006,
+      "step": 3141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.636641645815958e-05,
+      "loss": 0.3649,
+      "step": 3142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.635251598554351e-05,
+      "loss": 0.3617,
+      "step": 3143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.633861551292744e-05,
+      "loss": 0.3409,
+      "step": 3144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.632471504031137e-05,
+      "loss": 0.3243,
+      "step": 3145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.63108145676953e-05,
+      "loss": 0.468,
+      "step": 3146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6296914095079236e-05,
+      "loss": 0.3415,
+      "step": 3147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6283013622463166e-05,
+      "loss": 0.3794,
+      "step": 3148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6269113149847096e-05,
+      "loss": 0.3746,
+      "step": 3149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6255212677231026e-05,
+      "loss": 0.3548,
+      "step": 3150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.624131220461496e-05,
+      "loss": 0.3852,
+      "step": 3151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.622741173199889e-05,
+      "loss": 0.3876,
+      "step": 3152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.621351125938282e-05,
+      "loss": 0.3914,
+      "step": 3153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6199610786766753e-05,
+      "loss": 0.3574,
+      "step": 3154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6185710314150683e-05,
+      "loss": 0.3256,
+      "step": 3155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.617180984153462e-05,
+      "loss": 0.4027,
+      "step": 3156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.615790936891855e-05,
+      "loss": 0.3583,
+      "step": 3157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.614400889630248e-05,
+      "loss": 0.3645,
+      "step": 3158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.613010842368641e-05,
+      "loss": 0.373,
+      "step": 3159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.611620795107034e-05,
+      "loss": 0.3483,
+      "step": 3160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.610230747845428e-05,
+      "loss": 0.3807,
+      "step": 3161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.608840700583821e-05,
+      "loss": 0.3475,
+      "step": 3162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.607450653322214e-05,
+      "loss": 0.3822,
+      "step": 3163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.606060606060606e-05,
+      "loss": 0.3903,
+      "step": 3164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.604670558798999e-05,
+      "loss": 0.3625,
+      "step": 3165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.603280511537392e-05,
+      "loss": 0.3519,
+      "step": 3166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.601890464275785e-05,
+      "loss": 0.3914,
+      "step": 3167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.600500417014178e-05,
+      "loss": 0.3426,
+      "step": 3168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.599110369752572e-05,
+      "loss": 0.403,
+      "step": 3169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.597720322490965e-05,
+      "loss": 0.4082,
+      "step": 3170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.596330275229358e-05,
+      "loss": 0.3948,
+      "step": 3171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.594940227967751e-05,
+      "loss": 0.3576,
+      "step": 3172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.593550180706144e-05,
+      "loss": 0.3529,
+      "step": 3173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5921601334445375e-05,
+      "loss": 0.3748,
+      "step": 3174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5907700861829305e-05,
+      "loss": 0.3791,
+      "step": 3175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5893800389213235e-05,
+      "loss": 0.3867,
+      "step": 3176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5879899916597165e-05,
+      "loss": 0.4119,
+      "step": 3177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5865999443981095e-05,
+      "loss": 0.3618,
+      "step": 3178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.585209897136503e-05,
+      "loss": 0.3428,
+      "step": 3179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.583819849874896e-05,
+      "loss": 0.3176,
+      "step": 3180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.582429802613289e-05,
+      "loss": 0.4046,
+      "step": 3181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.581039755351682e-05,
+      "loss": 0.3988,
+      "step": 3182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.579649708090076e-05,
+      "loss": 0.3592,
+      "step": 3183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.578259660828469e-05,
+      "loss": 0.3372,
+      "step": 3184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.576869613566862e-05,
+      "loss": 0.3069,
+      "step": 3185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.575479566305255e-05,
+      "loss": 0.3919,
+      "step": 3186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.574089519043648e-05,
+      "loss": 0.3484,
+      "step": 3187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5726994717820416e-05,
+      "loss": 0.3591,
+      "step": 3188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5713094245204346e-05,
+      "loss": 0.3665,
+      "step": 3189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5699193772588276e-05,
+      "loss": 0.382,
+      "step": 3190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.56852932999722e-05,
+      "loss": 0.3556,
+      "step": 3191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.567139282735613e-05,
+      "loss": 0.3911,
+      "step": 3192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.565749235474006e-05,
+      "loss": 0.3557,
+      "step": 3193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.564359188212399e-05,
+      "loss": 0.3815,
+      "step": 3194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.562969140950792e-05,
+      "loss": 0.3531,
+      "step": 3195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.561579093689186e-05,
+      "loss": 0.3753,
+      "step": 3196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.560189046427579e-05,
+      "loss": 0.352,
+      "step": 3197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.558798999165972e-05,
+      "loss": 0.3156,
+      "step": 3198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.557408951904365e-05,
+      "loss": 0.4046,
+      "step": 3199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.556018904642758e-05,
+      "loss": 0.402,
+      "step": 3200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5546288573811514e-05,
+      "loss": 0.4032,
+      "step": 3201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5532388101195444e-05,
+      "loss": 0.32,
+      "step": 3202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5518487628579374e-05,
+      "loss": 0.3629,
+      "step": 3203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5504587155963304e-05,
+      "loss": 0.3934,
+      "step": 3204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5490686683347234e-05,
+      "loss": 0.3753,
+      "step": 3205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.547678621073117e-05,
+      "loss": 0.4155,
+      "step": 3206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.54628857381151e-05,
+      "loss": 0.3185,
+      "step": 3207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.544898526549903e-05,
+      "loss": 0.3989,
+      "step": 3208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.543508479288296e-05,
+      "loss": 0.3996,
+      "step": 3209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.542118432026689e-05,
+      "loss": 0.365,
+      "step": 3210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.540728384765083e-05,
+      "loss": 0.3657,
+      "step": 3211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.539338337503476e-05,
+      "loss": 0.3975,
+      "step": 3212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.537948290241869e-05,
+      "loss": 0.4231,
+      "step": 3213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.536558242980262e-05,
+      "loss": 0.3673,
+      "step": 3214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5351681957186555e-05,
+      "loss": 0.3551,
+      "step": 3215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5337781484570485e-05,
+      "loss": 0.4004,
+      "step": 3216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.53238810119544e-05,
+      "loss": 0.3733,
+      "step": 3217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.530998053933834e-05,
+      "loss": 0.3711,
+      "step": 3218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.529608006672227e-05,
+      "loss": 0.3541,
+      "step": 3219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.52821795941062e-05,
+      "loss": 0.3551,
+      "step": 3220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.526827912149013e-05,
+      "loss": 0.3953,
+      "step": 3221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.525437864887406e-05,
+      "loss": 0.3817,
+      "step": 3222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5240478176257996e-05,
+      "loss": 0.3864,
+      "step": 3223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5226577703641926e-05,
+      "loss": 0.3986,
+      "step": 3224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5212677231025856e-05,
+      "loss": 0.3811,
+      "step": 3225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5198776758409786e-05,
+      "loss": 0.4159,
+      "step": 3226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5184876285793716e-05,
+      "loss": 0.3624,
+      "step": 3227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.517097581317765e-05,
+      "loss": 0.3954,
+      "step": 3228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.515707534056158e-05,
+      "loss": 0.324,
+      "step": 3229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.514317486794551e-05,
+      "loss": 0.383,
+      "step": 3230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.512927439532944e-05,
+      "loss": 0.4116,
+      "step": 3231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.511537392271337e-05,
+      "loss": 0.3642,
+      "step": 3232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.510147345009731e-05,
+      "loss": 0.385,
+      "step": 3233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.508757297748124e-05,
+      "loss": 0.3745,
+      "step": 3234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.507367250486517e-05,
+      "loss": 0.3897,
+      "step": 3235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.50597720322491e-05,
+      "loss": 0.3686,
+      "step": 3236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.504587155963303e-05,
+      "loss": 0.4148,
+      "step": 3237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.503197108701697e-05,
+      "loss": 0.3678,
+      "step": 3238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.50180706144009e-05,
+      "loss": 0.3482,
+      "step": 3239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.500417014178483e-05,
+      "loss": 0.3994,
+      "step": 3240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.499026966916876e-05,
+      "loss": 0.3813,
+      "step": 3241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4976369196552694e-05,
+      "loss": 0.3829,
+      "step": 3242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.496246872393661e-05,
+      "loss": 0.3451,
+      "step": 3243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.494856825132054e-05,
+      "loss": 0.357,
+      "step": 3244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.493466777870447e-05,
+      "loss": 0.3618,
+      "step": 3245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.492076730608841e-05,
+      "loss": 0.3891,
+      "step": 3246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.490686683347234e-05,
+      "loss": 0.3651,
+      "step": 3247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.489296636085627e-05,
+      "loss": 0.3286,
+      "step": 3248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.48790658882402e-05,
+      "loss": 0.3914,
+      "step": 3249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4865165415624134e-05,
+      "loss": 0.356,
+      "step": 3250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4851264943008064e-05,
+      "loss": 0.4536,
+      "step": 3251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4837364470391995e-05,
+      "loss": 0.3199,
+      "step": 3252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4823463997775925e-05,
+      "loss": 0.39,
+      "step": 3253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4809563525159855e-05,
+      "loss": 0.3255,
+      "step": 3254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.479566305254379e-05,
+      "loss": 0.3587,
+      "step": 3255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.478176257992772e-05,
+      "loss": 0.3976,
+      "step": 3256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.476786210731165e-05,
+      "loss": 0.403,
+      "step": 3257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.475396163469558e-05,
+      "loss": 0.3403,
+      "step": 3258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.474006116207951e-05,
+      "loss": 0.3179,
+      "step": 3259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.472616068946345e-05,
+      "loss": 0.4032,
+      "step": 3260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.471226021684738e-05,
+      "loss": 0.4049,
+      "step": 3261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.469835974423131e-05,
+      "loss": 0.3883,
+      "step": 3262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.468445927161524e-05,
+      "loss": 0.309,
+      "step": 3263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.467055879899917e-05,
+      "loss": 0.3203,
+      "step": 3264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4656658326383106e-05,
+      "loss": 0.3561,
+      "step": 3265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4642757853767036e-05,
+      "loss": 0.4691,
+      "step": 3266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4628857381150966e-05,
+      "loss": 0.3797,
+      "step": 3267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4614956908534896e-05,
+      "loss": 0.4333,
+      "step": 3268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4601056435918826e-05,
+      "loss": 0.3506,
+      "step": 3269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.458715596330275e-05,
+      "loss": 0.4501,
+      "step": 3270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.457325549068668e-05,
+      "loss": 0.3846,
+      "step": 3271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.455935501807061e-05,
+      "loss": 0.3956,
+      "step": 3272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 0.3054,
+      "step": 3273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4531554072838476e-05,
+      "loss": 0.3575,
+      "step": 3274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4517653600222406e-05,
+      "loss": 0.3785,
+      "step": 3275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4503753127606336e-05,
+      "loss": 0.3435,
+      "step": 3276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.448985265499027e-05,
+      "loss": 0.4254,
+      "step": 3277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.44759521823742e-05,
+      "loss": 0.3108,
+      "step": 3278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.446205170975813e-05,
+      "loss": 0.3357,
+      "step": 3279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4448151237142063e-05,
+      "loss": 0.41,
+      "step": 3280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4434250764525994e-05,
+      "loss": 0.3749,
+      "step": 3281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.442035029190993e-05,
+      "loss": 0.3766,
+      "step": 3282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.440644981929386e-05,
+      "loss": 0.3782,
+      "step": 3283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.439254934667779e-05,
+      "loss": 0.3781,
+      "step": 3284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.437864887406172e-05,
+      "loss": 0.3857,
+      "step": 3285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.436474840144565e-05,
+      "loss": 0.3741,
+      "step": 3286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.435084792882959e-05,
+      "loss": 0.3412,
+      "step": 3287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.433694745621352e-05,
+      "loss": 0.3289,
+      "step": 3288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.432304698359745e-05,
+      "loss": 0.3703,
+      "step": 3289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.430914651098138e-05,
+      "loss": 0.3726,
+      "step": 3290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.429524603836531e-05,
+      "loss": 0.3711,
+      "step": 3291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4281345565749245e-05,
+      "loss": 0.2872,
+      "step": 3292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4267445093133175e-05,
+      "loss": 0.4182,
+      "step": 3293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4253544620517105e-05,
+      "loss": 0.3061,
+      "step": 3294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4239644147901035e-05,
+      "loss": 0.3444,
+      "step": 3295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.422574367528496e-05,
+      "loss": 0.4246,
+      "step": 3296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.421184320266889e-05,
+      "loss": 0.3973,
+      "step": 3297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.419794273005282e-05,
+      "loss": 0.3419,
+      "step": 3298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.418404225743675e-05,
+      "loss": 0.3694,
+      "step": 3299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4170141784820685e-05,
+      "loss": 0.3845,
+      "step": 3300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4156241312204615e-05,
+      "loss": 0.4093,
+      "step": 3301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4142340839588545e-05,
+      "loss": 0.3077,
+      "step": 3302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4128440366972475e-05,
+      "loss": 0.3817,
+      "step": 3303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4114539894356405e-05,
+      "loss": 0.3022,
+      "step": 3304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.410063942174034e-05,
+      "loss": 0.421,
+      "step": 3305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.408673894912427e-05,
+      "loss": 0.3432,
+      "step": 3306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.40728384765082e-05,
+      "loss": 0.3794,
+      "step": 3307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.405893800389213e-05,
+      "loss": 0.3349,
+      "step": 3308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.404503753127607e-05,
+      "loss": 0.4065,
+      "step": 3309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.403113705866e-05,
+      "loss": 0.3515,
+      "step": 3310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.401723658604393e-05,
+      "loss": 0.3652,
+      "step": 3311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.400333611342786e-05,
+      "loss": 0.3803,
+      "step": 3312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.398943564081179e-05,
+      "loss": 0.3735,
+      "step": 3313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3975535168195726e-05,
+      "loss": 0.3144,
+      "step": 3314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3961634695579656e-05,
+      "loss": 0.325,
+      "step": 3315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3947734222963586e-05,
+      "loss": 0.3886,
+      "step": 3316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3933833750347516e-05,
+      "loss": 0.32,
+      "step": 3317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3919933277731447e-05,
+      "loss": 0.4111,
+      "step": 3318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.390603280511538e-05,
+      "loss": 0.3642,
+      "step": 3319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3892132332499313e-05,
+      "loss": 0.3923,
+      "step": 3320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3878231859883244e-05,
+      "loss": 0.3829,
+      "step": 3321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3864331387267174e-05,
+      "loss": 0.3652,
+      "step": 3322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.38504309146511e-05,
+      "loss": 0.3164,
+      "step": 3323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.383653044203503e-05,
+      "loss": 0.3971,
+      "step": 3324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.382262996941896e-05,
+      "loss": 0.4761,
+      "step": 3325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.380872949680289e-05,
+      "loss": 0.3634,
+      "step": 3326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3794829024186824e-05,
+      "loss": 0.3122,
+      "step": 3327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3780928551570754e-05,
+      "loss": 0.3149,
+      "step": 3328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3767028078954684e-05,
+      "loss": 0.3246,
+      "step": 3329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3753127606338614e-05,
+      "loss": 0.3619,
+      "step": 3330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3739227133722544e-05,
+      "loss": 0.3328,
+      "step": 3331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.372532666110648e-05,
+      "loss": 0.3892,
+      "step": 3332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.371142618849041e-05,
+      "loss": 0.4064,
+      "step": 3333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.369752571587434e-05,
+      "loss": 0.3295,
+      "step": 3334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.368362524325827e-05,
+      "loss": 0.3655,
+      "step": 3335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.36697247706422e-05,
+      "loss": 0.3603,
+      "step": 3336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.365582429802614e-05,
+      "loss": 0.3507,
+      "step": 3337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.364192382541007e-05,
+      "loss": 0.3385,
+      "step": 3338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3628023352794e-05,
+      "loss": 0.3858,
+      "step": 3339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.361412288017793e-05,
+      "loss": 0.3233,
+      "step": 3340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3600222407561865e-05,
+      "loss": 0.4447,
+      "step": 3341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3586321934945795e-05,
+      "loss": 0.374,
+      "step": 3342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3572421462329725e-05,
+      "loss": 0.3126,
+      "step": 3343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3558520989713655e-05,
+      "loss": 0.3302,
+      "step": 3344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3544620517097585e-05,
+      "loss": 0.415,
+      "step": 3345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.353072004448152e-05,
+      "loss": 0.3741,
+      "step": 3346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.351681957186545e-05,
+      "loss": 0.345,
+      "step": 3347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.350291909924938e-05,
+      "loss": 0.4687,
+      "step": 3348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3489018626633306e-05,
+      "loss": 0.3991,
+      "step": 3349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3475118154017236e-05,
+      "loss": 0.4027,
+      "step": 3350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3461217681401166e-05,
+      "loss": 0.3243,
+      "step": 3351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3447317208785096e-05,
+      "loss": 0.361,
+      "step": 3352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3433416736169026e-05,
+      "loss": 0.389,
+      "step": 3353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.341951626355296e-05,
+      "loss": 0.2982,
+      "step": 3354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.340561579093689e-05,
+      "loss": 0.4242,
+      "step": 3355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.339171531832082e-05,
+      "loss": 0.3541,
+      "step": 3356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.337781484570475e-05,
+      "loss": 0.388,
+      "step": 3357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.336391437308868e-05,
+      "loss": 0.3465,
+      "step": 3358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.335001390047262e-05,
+      "loss": 0.3513,
+      "step": 3359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.333611342785655e-05,
+      "loss": 0.3287,
+      "step": 3360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.332221295524048e-05,
+      "loss": 0.3697,
+      "step": 3361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.330831248262441e-05,
+      "loss": 0.3796,
+      "step": 3362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.329441201000834e-05,
+      "loss": 0.3656,
+      "step": 3363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.328051153739228e-05,
+      "loss": 0.356,
+      "step": 3364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.326661106477621e-05,
+      "loss": 0.4045,
+      "step": 3365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.325271059216014e-05,
+      "loss": 0.3835,
+      "step": 3366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.323881011954407e-05,
+      "loss": 0.3752,
+      "step": 3367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3224909646928004e-05,
+      "loss": 0.4052,
+      "step": 3368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3211009174311934e-05,
+      "loss": 0.362,
+      "step": 3369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3197108701695864e-05,
+      "loss": 0.3154,
+      "step": 3370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3183208229079794e-05,
+      "loss": 0.3818,
+      "step": 3371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3169307756463724e-05,
+      "loss": 0.3113,
+      "step": 3372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.315540728384766e-05,
+      "loss": 0.3643,
+      "step": 3373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.314150681123159e-05,
+      "loss": 0.3925,
+      "step": 3374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.312760633861552e-05,
+      "loss": 0.3177,
+      "step": 3375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3113705865999444e-05,
+      "loss": 0.3451,
+      "step": 3376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3099805393383375e-05,
+      "loss": 0.3666,
+      "step": 3377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3085904920767305e-05,
+      "loss": 0.3776,
+      "step": 3378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3072004448151235e-05,
+      "loss": 0.2971,
+      "step": 3379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3058103975535165e-05,
+      "loss": 0.3241,
+      "step": 3380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.30442035029191e-05,
+      "loss": 0.386,
+      "step": 3381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.303030303030303e-05,
+      "loss": 0.3651,
+      "step": 3382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.301640255768696e-05,
+      "loss": 0.3278,
+      "step": 3383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.300250208507089e-05,
+      "loss": 0.3718,
+      "step": 3384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.298860161245482e-05,
+      "loss": 0.4717,
+      "step": 3385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.297470113983876e-05,
+      "loss": 0.397,
+      "step": 3386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.296080066722269e-05,
+      "loss": 0.326,
+      "step": 3387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.294690019460662e-05,
+      "loss": 0.3494,
+      "step": 3388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.293299972199055e-05,
+      "loss": 0.3812,
+      "step": 3389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.291909924937448e-05,
+      "loss": 0.3814,
+      "step": 3390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2905198776758416e-05,
+      "loss": 0.3149,
+      "step": 3391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2891298304142346e-05,
+      "loss": 0.3992,
+      "step": 3392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2877397831526276e-05,
+      "loss": 0.3159,
+      "step": 3393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2863497358910206e-05,
+      "loss": 0.3344,
+      "step": 3394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2849596886294136e-05,
+      "loss": 0.3661,
+      "step": 3395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.283569641367807e-05,
+      "loss": 0.3644,
+      "step": 3396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2821795941062e-05,
+      "loss": 0.3533,
+      "step": 3397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.280789546844593e-05,
+      "loss": 0.3374,
+      "step": 3398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.279399499582986e-05,
+      "loss": 0.3983,
+      "step": 3399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.27800945232138e-05,
+      "loss": 0.3161,
+      "step": 3400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.276619405059773e-05,
+      "loss": 0.3871,
+      "step": 3401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2752293577981646e-05,
+      "loss": 0.2977,
+      "step": 3402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.273839310536558e-05,
+      "loss": 0.3477,
+      "step": 3403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.272449263274951e-05,
+      "loss": 0.3465,
+      "step": 3404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2710592160133443e-05,
+      "loss": 0.3789,
+      "step": 3405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2696691687517373e-05,
+      "loss": 0.3361,
+      "step": 3406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2682791214901304e-05,
+      "loss": 0.3425,
+      "step": 3407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.266889074228524e-05,
+      "loss": 0.3597,
+      "step": 3408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.265499026966917e-05,
+      "loss": 0.3482,
+      "step": 3409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.26410897970531e-05,
+      "loss": 0.3867,
+      "step": 3410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.262718932443703e-05,
+      "loss": 0.393,
+      "step": 3411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.261328885182096e-05,
+      "loss": 0.3314,
+      "step": 3412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.25993883792049e-05,
+      "loss": 0.3697,
+      "step": 3413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.258548790658883e-05,
+      "loss": 0.3449,
+      "step": 3414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.257158743397276e-05,
+      "loss": 0.336,
+      "step": 3415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.255768696135669e-05,
+      "loss": 0.3477,
+      "step": 3416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.254378648874062e-05,
+      "loss": 0.385,
+      "step": 3417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2529886016124555e-05,
+      "loss": 0.3683,
+      "step": 3418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2515985543508485e-05,
+      "loss": 0.3869,
+      "step": 3419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2502085070892415e-05,
+      "loss": 0.3209,
+      "step": 3420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2488184598276345e-05,
+      "loss": 0.3448,
+      "step": 3421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2474284125660275e-05,
+      "loss": 0.3819,
+      "step": 3422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.246038365304421e-05,
+      "loss": 0.3735,
+      "step": 3423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.244648318042814e-05,
+      "loss": 0.3882,
+      "step": 3424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.243258270781207e-05,
+      "loss": 0.3406,
+      "step": 3425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2418682235196e-05,
+      "loss": 0.3709,
+      "step": 3426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.240478176257994e-05,
+      "loss": 0.3502,
+      "step": 3427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.239088128996387e-05,
+      "loss": 0.4061,
+      "step": 3428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2376980817347785e-05,
+      "loss": 0.3955,
+      "step": 3429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2363080344731715e-05,
+      "loss": 0.3438,
+      "step": 3430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.234917987211565e-05,
+      "loss": 0.3205,
+      "step": 3431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.233527939949958e-05,
+      "loss": 0.3419,
+      "step": 3432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.232137892688351e-05,
+      "loss": 0.3955,
+      "step": 3433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.230747845426744e-05,
+      "loss": 0.4395,
+      "step": 3434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.229357798165138e-05,
+      "loss": 0.3267,
+      "step": 3435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.227967750903531e-05,
+      "loss": 0.3629,
+      "step": 3436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.226577703641924e-05,
+      "loss": 0.4125,
+      "step": 3437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.225187656380317e-05,
+      "loss": 0.3876,
+      "step": 3438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.22379760911871e-05,
+      "loss": 0.3497,
+      "step": 3439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2224075618571036e-05,
+      "loss": 0.3214,
+      "step": 3440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2210175145954966e-05,
+      "loss": 0.3528,
+      "step": 3441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2196274673338896e-05,
+      "loss": 0.3908,
+      "step": 3442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2182374200722827e-05,
+      "loss": 0.3824,
+      "step": 3443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2168473728106757e-05,
+      "loss": 0.2897,
+      "step": 3444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2154573255490693e-05,
+      "loss": 0.3492,
+      "step": 3445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2140672782874623e-05,
+      "loss": 0.3589,
+      "step": 3446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2126772310258554e-05,
+      "loss": 0.3767,
+      "step": 3447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2112871837642484e-05,
+      "loss": 0.3576,
+      "step": 3448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2098971365026414e-05,
+      "loss": 0.3884,
+      "step": 3449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.208507089241035e-05,
+      "loss": 0.3315,
+      "step": 3450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.207117041979428e-05,
+      "loss": 0.4004,
+      "step": 3451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.205726994717821e-05,
+      "loss": 0.4234,
+      "step": 3452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.204336947456214e-05,
+      "loss": 0.3571,
+      "step": 3453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.202946900194607e-05,
+      "loss": 0.3236,
+      "step": 3454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2015568529329994e-05,
+      "loss": 0.3473,
+      "step": 3455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2001668056713924e-05,
+      "loss": 0.3619,
+      "step": 3456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1987767584097854e-05,
+      "loss": 0.3867,
+      "step": 3457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.197386711148179e-05,
+      "loss": 0.3873,
+      "step": 3458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.195996663886572e-05,
+      "loss": 0.3235,
+      "step": 3459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.194606616624965e-05,
+      "loss": 0.3415,
+      "step": 3460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.193216569363358e-05,
+      "loss": 0.3991,
+      "step": 3461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.191826522101752e-05,
+      "loss": 0.3532,
+      "step": 3462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.190436474840145e-05,
+      "loss": 0.3372,
+      "step": 3463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.189046427578538e-05,
+      "loss": 0.3515,
+      "step": 3464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.187656380316931e-05,
+      "loss": 0.3296,
+      "step": 3465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.186266333055324e-05,
+      "loss": 0.3636,
+      "step": 3466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1848762857937175e-05,
+      "loss": 0.4124,
+      "step": 3467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1834862385321105e-05,
+      "loss": 0.3705,
+      "step": 3468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1820961912705035e-05,
+      "loss": 0.3925,
+      "step": 3469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1807061440088965e-05,
+      "loss": 0.3306,
+      "step": 3470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.1793160967472895e-05,
+      "loss": 0.354,
+      "step": 3471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.177926049485683e-05,
+      "loss": 0.3274,
+      "step": 3472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.176536002224076e-05,
+      "loss": 0.3899,
+      "step": 3473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.175145954962469e-05,
+      "loss": 0.3352,
+      "step": 3474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.173755907700862e-05,
+      "loss": 0.3678,
+      "step": 3475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.172365860439255e-05,
+      "loss": 0.3591,
+      "step": 3476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.170975813177649e-05,
+      "loss": 0.3707,
+      "step": 3477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.169585765916042e-05,
+      "loss": 0.3069,
+      "step": 3478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.168195718654435e-05,
+      "loss": 0.3595,
+      "step": 3479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.166805671392828e-05,
+      "loss": 0.3564,
+      "step": 3480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.16541562413122e-05,
+      "loss": 0.4073,
+      "step": 3481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.164025576869613e-05,
+      "loss": 0.3913,
+      "step": 3482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.162635529608006e-05,
+      "loss": 0.3319,
+      "step": 3483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.161245482346399e-05,
+      "loss": 0.2906,
+      "step": 3484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.159855435084793e-05,
+      "loss": 0.3457,
+      "step": 3485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.158465387823186e-05,
+      "loss": 0.3606,
+      "step": 3486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.157075340561579e-05,
+      "loss": 0.3456,
+      "step": 3487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.155685293299972e-05,
+      "loss": 0.4085,
+      "step": 3488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.154295246038365e-05,
+      "loss": 0.433,
+      "step": 3489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.152905198776759e-05,
+      "loss": 0.3482,
+      "step": 3490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.151515151515152e-05,
+      "loss": 0.3957,
+      "step": 3491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.150125104253545e-05,
+      "loss": 0.3487,
+      "step": 3492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.148735056991938e-05,
+      "loss": 0.3401,
+      "step": 3493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1473450097303314e-05,
+      "loss": 0.3176,
+      "step": 3494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1459549624687244e-05,
+      "loss": 0.3359,
+      "step": 3495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1445649152071174e-05,
+      "loss": 0.4133,
+      "step": 3496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1431748679455104e-05,
+      "loss": 0.3635,
+      "step": 3497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1417848206839034e-05,
+      "loss": 0.3536,
+      "step": 3498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.140394773422297e-05,
+      "loss": 0.3298,
+      "step": 3499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.13900472616069e-05,
+      "loss": 0.3784,
+      "step": 3500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.137614678899083e-05,
+      "loss": 0.4158,
+      "step": 3501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.136224631637476e-05,
+      "loss": 0.3738,
+      "step": 3502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.134834584375869e-05,
+      "loss": 0.3665,
+      "step": 3503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.133444537114263e-05,
+      "loss": 0.3743,
+      "step": 3504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.132054489852656e-05,
+      "loss": 0.3956,
+      "step": 3505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.130664442591049e-05,
+      "loss": 0.3406,
+      "step": 3506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.129274395329442e-05,
+      "loss": 0.4027,
+      "step": 3507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.127884348067834e-05,
+      "loss": 0.3672,
+      "step": 3508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.126494300806227e-05,
+      "loss": 0.3882,
+      "step": 3509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.12510425354462e-05,
+      "loss": 0.4074,
+      "step": 3510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.123714206283013e-05,
+      "loss": 0.3855,
+      "step": 3511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.122324159021407e-05,
+      "loss": 0.3617,
+      "step": 3512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1209341117598e-05,
+      "loss": 0.3882,
+      "step": 3513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.119544064498193e-05,
+      "loss": 0.3717,
+      "step": 3514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.118154017236586e-05,
+      "loss": 0.3537,
+      "step": 3515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.116763969974979e-05,
+      "loss": 0.326,
+      "step": 3516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1153739227133726e-05,
+      "loss": 0.3164,
+      "step": 3517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1139838754517656e-05,
+      "loss": 0.3094,
+      "step": 3518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1125938281901586e-05,
+      "loss": 0.338,
+      "step": 3519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1112037809285516e-05,
+      "loss": 0.3852,
+      "step": 3520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1098137336669446e-05,
+      "loss": 0.3624,
+      "step": 3521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.108423686405338e-05,
+      "loss": 0.366,
+      "step": 3522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.107033639143731e-05,
+      "loss": 0.3907,
+      "step": 3523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.105643591882124e-05,
+      "loss": 0.4109,
+      "step": 3524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.104253544620517e-05,
+      "loss": 0.3913,
+      "step": 3525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.102863497358911e-05,
+      "loss": 0.358,
+      "step": 3526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.101473450097304e-05,
+      "loss": 0.3526,
+      "step": 3527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.100083402835697e-05,
+      "loss": 0.4131,
+      "step": 3528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.09869335557409e-05,
+      "loss": 0.3903,
+      "step": 3529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.097303308312483e-05,
+      "loss": 0.3921,
+      "step": 3530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.095913261050877e-05,
+      "loss": 0.3556,
+      "step": 3531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.09452321378927e-05,
+      "loss": 0.3726,
+      "step": 3532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.093133166527663e-05,
+      "loss": 0.3681,
+      "step": 3533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.091743119266055e-05,
+      "loss": 0.4005,
+      "step": 3534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.090353072004448e-05,
+      "loss": 0.3758,
+      "step": 3535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.088963024742841e-05,
+      "loss": 0.355,
+      "step": 3536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.087572977481234e-05,
+      "loss": 0.3772,
+      "step": 3537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.086182930219627e-05,
+      "loss": 0.3444,
+      "step": 3538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.084792882958021e-05,
+      "loss": 0.3391,
+      "step": 3539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.083402835696414e-05,
+      "loss": 0.3333,
+      "step": 3540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.082012788434807e-05,
+      "loss": 0.3412,
+      "step": 3541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0806227411732e-05,
+      "loss": 0.3568,
+      "step": 3542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.079232693911593e-05,
+      "loss": 0.3282,
+      "step": 3543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0778426466499865e-05,
+      "loss": 0.3517,
+      "step": 3544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0764525993883795e-05,
+      "loss": 0.3371,
+      "step": 3545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0750625521267725e-05,
+      "loss": 0.3049,
+      "step": 3546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0736725048651655e-05,
+      "loss": 0.3599,
+      "step": 3547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0722824576035585e-05,
+      "loss": 0.3637,
+      "step": 3548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.070892410341952e-05,
+      "loss": 0.3651,
+      "step": 3549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.069502363080345e-05,
+      "loss": 0.3559,
+      "step": 3550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.068112315818738e-05,
+      "loss": 0.3607,
+      "step": 3551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.066722268557131e-05,
+      "loss": 0.3884,
+      "step": 3552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.065332221295525e-05,
+      "loss": 0.3372,
+      "step": 3553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.063942174033918e-05,
+      "loss": 0.3806,
+      "step": 3554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.062552126772311e-05,
+      "loss": 0.3882,
+      "step": 3555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.061162079510704e-05,
+      "loss": 0.266,
+      "step": 3556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.059772032249097e-05,
+      "loss": 0.3667,
+      "step": 3557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0583819849874906e-05,
+      "loss": 0.3419,
+      "step": 3558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0569919377258836e-05,
+      "loss": 0.374,
+      "step": 3559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0556018904642766e-05,
+      "loss": 0.3466,
+      "step": 3560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.054211843202669e-05,
+      "loss": 0.3643,
+      "step": 3561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.052821795941062e-05,
+      "loss": 0.3087,
+      "step": 3562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.051431748679455e-05,
+      "loss": 0.3508,
+      "step": 3563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.050041701417848e-05,
+      "loss": 0.3481,
+      "step": 3564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.048651654156241e-05,
+      "loss": 0.4653,
+      "step": 3565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0472616068946346e-05,
+      "loss": 0.402,
+      "step": 3566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0458715596330276e-05,
+      "loss": 0.3832,
+      "step": 3567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0444815123714206e-05,
+      "loss": 0.318,
+      "step": 3568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0430914651098137e-05,
+      "loss": 0.3502,
+      "step": 3569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0417014178482067e-05,
+      "loss": 0.399,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0403113705866003e-05,
+      "loss": 0.3003,
+      "step": 3571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0389213233249934e-05,
+      "loss": 0.3918,
+      "step": 3572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0375312760633864e-05,
+      "loss": 0.3616,
+      "step": 3573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0361412288017794e-05,
+      "loss": 0.3562,
+      "step": 3574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0347511815401724e-05,
+      "loss": 0.3268,
+      "step": 3575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.033361134278566e-05,
+      "loss": 0.3481,
+      "step": 3576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.031971087016959e-05,
+      "loss": 0.3373,
+      "step": 3577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.030581039755352e-05,
+      "loss": 0.405,
+      "step": 3578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.029190992493745e-05,
+      "loss": 0.3954,
+      "step": 3579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.027800945232138e-05,
+      "loss": 0.3261,
+      "step": 3580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.026410897970532e-05,
+      "loss": 0.3408,
+      "step": 3581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.025020850708925e-05,
+      "loss": 0.3402,
+      "step": 3582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.023630803447318e-05,
+      "loss": 0.3077,
+      "step": 3583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.022240756185711e-05,
+      "loss": 0.3142,
+      "step": 3584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0208507089241045e-05,
+      "loss": 0.3748,
+      "step": 3585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0194606616624975e-05,
+      "loss": 0.3799,
+      "step": 3586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.018070614400889e-05,
+      "loss": 0.3148,
+      "step": 3587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.016680567139283e-05,
+      "loss": 0.3294,
+      "step": 3588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.015290519877676e-05,
+      "loss": 0.3777,
+      "step": 3589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.013900472616069e-05,
+      "loss": 0.3385,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.012510425354462e-05,
+      "loss": 0.3637,
+      "step": 3591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.011120378092855e-05,
+      "loss": 0.3265,
+      "step": 3592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0097303308312485e-05,
+      "loss": 0.3518,
+      "step": 3593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0083402835696415e-05,
+      "loss": 0.3293,
+      "step": 3594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0069502363080345e-05,
+      "loss": 0.3702,
+      "step": 3595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0055601890464275e-05,
+      "loss": 0.3943,
+      "step": 3596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0041701417848205e-05,
+      "loss": 0.3783,
+      "step": 3597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.002780094523214e-05,
+      "loss": 0.4264,
+      "step": 3598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.001390047261607e-05,
+      "loss": 0.379,
+      "step": 3599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-05,
+      "loss": 0.3313,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.998609952738393e-05,
+      "loss": 0.3029,
+      "step": 3601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.997219905476786e-05,
+      "loss": 0.3156,
+      "step": 3602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.99582985821518e-05,
+      "loss": 0.3217,
+      "step": 3603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.994439810953573e-05,
+      "loss": 0.4112,
+      "step": 3604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.993049763691966e-05,
+      "loss": 0.4161,
+      "step": 3605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.991659716430359e-05,
+      "loss": 0.371,
+      "step": 3606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.990269669168752e-05,
+      "loss": 0.3568,
+      "step": 3607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.988879621907145e-05,
+      "loss": 0.3417,
+      "step": 3608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.987489574645538e-05,
+      "loss": 0.3781,
+      "step": 3609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.986099527383931e-05,
+      "loss": 0.33,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.984709480122324e-05,
+      "loss": 0.3082,
+      "step": 3611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.983319432860718e-05,
+      "loss": 0.3389,
+      "step": 3612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.981929385599111e-05,
+      "loss": 0.3704,
+      "step": 3613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.980539338337504e-05,
+      "loss": 0.3912,
+      "step": 3614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.979149291075897e-05,
+      "loss": 0.3569,
+      "step": 3615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9777592438142904e-05,
+      "loss": 0.3946,
+      "step": 3616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9763691965526834e-05,
+      "loss": 0.3398,
+      "step": 3617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9749791492910764e-05,
+      "loss": 0.3829,
+      "step": 3618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9735891020294694e-05,
+      "loss": 0.3226,
+      "step": 3619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9721990547678624e-05,
+      "loss": 0.3239,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9708090075062554e-05,
+      "loss": 0.3672,
+      "step": 3621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9694189602446484e-05,
+      "loss": 0.4092,
+      "step": 3622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9680289129830414e-05,
+      "loss": 0.3374,
+      "step": 3623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9666388657214344e-05,
+      "loss": 0.3213,
+      "step": 3624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.965248818459828e-05,
+      "loss": 0.3904,
+      "step": 3625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.963858771198221e-05,
+      "loss": 0.3757,
+      "step": 3626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.962468723936614e-05,
+      "loss": 0.3519,
+      "step": 3627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.961078676675007e-05,
+      "loss": 0.3529,
+      "step": 3628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9596886294134e-05,
+      "loss": 0.3527,
+      "step": 3629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.958298582151794e-05,
+      "loss": 0.3561,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.956908534890187e-05,
+      "loss": 0.304,
+      "step": 3631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.95551848762858e-05,
+      "loss": 0.375,
+      "step": 3632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.954128440366973e-05,
+      "loss": 0.2893,
+      "step": 3633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.952738393105366e-05,
+      "loss": 0.3488,
+      "step": 3634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.951348345843759e-05,
+      "loss": 0.3694,
+      "step": 3635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.949958298582152e-05,
+      "loss": 0.3579,
+      "step": 3636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.948568251320545e-05,
+      "loss": 0.4527,
+      "step": 3637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.947178204058938e-05,
+      "loss": 0.406,
+      "step": 3638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9457881567973316e-05,
+      "loss": 0.4066,
+      "step": 3639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9443981095357246e-05,
+      "loss": 0.3282,
+      "step": 3640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9430080622741176e-05,
+      "loss": 0.4009,
+      "step": 3641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9416180150125106e-05,
+      "loss": 0.3559,
+      "step": 3642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9402279677509036e-05,
+      "loss": 0.3877,
+      "step": 3643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.938837920489297e-05,
+      "loss": 0.4028,
+      "step": 3644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93744787322769e-05,
+      "loss": 0.3415,
+      "step": 3645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.936057825966083e-05,
+      "loss": 0.3784,
+      "step": 3646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9346677787044756e-05,
+      "loss": 0.3719,
+      "step": 3647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933277731442869e-05,
+      "loss": 0.4411,
+      "step": 3648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931887684181262e-05,
+      "loss": 0.3944,
+      "step": 3649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930497636919655e-05,
+      "loss": 0.3178,
+      "step": 3650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.929107589658048e-05,
+      "loss": 0.3429,
+      "step": 3651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.927717542396442e-05,
+      "loss": 0.3251,
+      "step": 3652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.926327495134835e-05,
+      "loss": 0.2947,
+      "step": 3653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.924937447873228e-05,
+      "loss": 0.4026,
+      "step": 3654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.923547400611621e-05,
+      "loss": 0.3495,
+      "step": 3655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.922157353350014e-05,
+      "loss": 0.2835,
+      "step": 3656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.920767306088408e-05,
+      "loss": 0.3557,
+      "step": 3657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919377258826801e-05,
+      "loss": 0.3742,
+      "step": 3658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.917987211565194e-05,
+      "loss": 0.3501,
+      "step": 3659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.916597164303586e-05,
+      "loss": 0.3746,
+      "step": 3660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.91520711704198e-05,
+      "loss": 0.3781,
+      "step": 3661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.913817069780373e-05,
+      "loss": 0.3503,
+      "step": 3662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.912427022518766e-05,
+      "loss": 0.4098,
+      "step": 3663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.911036975257159e-05,
+      "loss": 0.3509,
+      "step": 3664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.909646927995552e-05,
+      "loss": 0.3379,
+      "step": 3665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9082568807339454e-05,
+      "loss": 0.3787,
+      "step": 3666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9068668334723384e-05,
+      "loss": 0.408,
+      "step": 3667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9054767862107315e-05,
+      "loss": 0.3279,
+      "step": 3668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9040867389491245e-05,
+      "loss": 0.4235,
+      "step": 3669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9026966916875175e-05,
+      "loss": 0.3518,
+      "step": 3670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.901306644425911e-05,
+      "loss": 0.4635,
+      "step": 3671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.899916597164304e-05,
+      "loss": 0.3283,
+      "step": 3672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8985265499026965e-05,
+      "loss": 0.3819,
+      "step": 3673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8971365026410895e-05,
+      "loss": 0.3965,
+      "step": 3674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.895746455379483e-05,
+      "loss": 0.3918,
+      "step": 3675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.894356408117876e-05,
+      "loss": 0.3698,
+      "step": 3676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.892966360856269e-05,
+      "loss": 0.4226,
+      "step": 3677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.891576313594662e-05,
+      "loss": 0.3364,
+      "step": 3678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.890186266333056e-05,
+      "loss": 0.3807,
+      "step": 3679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.888796219071449e-05,
+      "loss": 0.3742,
+      "step": 3680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.887406171809842e-05,
+      "loss": 0.358,
+      "step": 3681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.886016124548235e-05,
+      "loss": 0.3687,
+      "step": 3682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.884626077286628e-05,
+      "loss": 0.4211,
+      "step": 3683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8832360300250216e-05,
+      "loss": 0.3366,
+      "step": 3684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8818459827634146e-05,
+      "loss": 0.3219,
+      "step": 3685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.880455935501807e-05,
+      "loss": 0.3615,
+      "step": 3686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8790658882402e-05,
+      "loss": 0.3583,
+      "step": 3687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8776758409785936e-05,
+      "loss": 0.4002,
+      "step": 3688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8762857937169866e-05,
+      "loss": 0.4118,
+      "step": 3689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8748957464553796e-05,
+      "loss": 0.3364,
+      "step": 3690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8735056991937726e-05,
+      "loss": 0.3986,
+      "step": 3691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8721156519321656e-05,
+      "loss": 0.3448,
+      "step": 3692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.870725604670559e-05,
+      "loss": 0.3462,
+      "step": 3693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.869335557408952e-05,
+      "loss": 0.3916,
+      "step": 3694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.867945510147345e-05,
+      "loss": 0.4523,
+      "step": 3695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8665554628857383e-05,
+      "loss": 0.3551,
+      "step": 3696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8651654156241313e-05,
+      "loss": 0.3228,
+      "step": 3697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.863775368362525e-05,
+      "loss": 0.3235,
+      "step": 3698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.862385321100918e-05,
+      "loss": 0.3809,
+      "step": 3699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8609952738393104e-05,
+      "loss": 0.3523,
+      "step": 3700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8596052265777034e-05,
+      "loss": 0.3907,
+      "step": 3701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.858215179316097e-05,
+      "loss": 0.3605,
+      "step": 3702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.85682513205449e-05,
+      "loss": 0.4023,
+      "step": 3703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.855435084792883e-05,
+      "loss": 0.3046,
+      "step": 3704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.854045037531276e-05,
+      "loss": 0.2956,
+      "step": 3705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852654990269669e-05,
+      "loss": 0.3247,
+      "step": 3706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851264943008063e-05,
+      "loss": 0.351,
+      "step": 3707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849874895746456e-05,
+      "loss": 0.4012,
+      "step": 3708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 0.3063,
+      "step": 3709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847094801223242e-05,
+      "loss": 0.3529,
+      "step": 3710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8457047539616355e-05,
+      "loss": 0.3147,
+      "step": 3711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8443147067000285e-05,
+      "loss": 0.3167,
+      "step": 3712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.842924659438421e-05,
+      "loss": 0.3918,
+      "step": 3713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841534612176814e-05,
+      "loss": 0.3704,
+      "step": 3714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8401445649152075e-05,
+      "loss": 0.3691,
+      "step": 3715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8387545176536005e-05,
+      "loss": 0.371,
+      "step": 3716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8373644703919935e-05,
+      "loss": 0.3334,
+      "step": 3717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8359744231303865e-05,
+      "loss": 0.3691,
+      "step": 3718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8345843758687795e-05,
+      "loss": 0.3943,
+      "step": 3719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.833194328607173e-05,
+      "loss": 0.336,
+      "step": 3720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.831804281345566e-05,
+      "loss": 0.3683,
+      "step": 3721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.830414234083959e-05,
+      "loss": 0.3562,
+      "step": 3722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.829024186822352e-05,
+      "loss": 0.3125,
+      "step": 3723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.827634139560745e-05,
+      "loss": 0.3164,
+      "step": 3724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826244092299139e-05,
+      "loss": 0.3795,
+      "step": 3725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.824854045037531e-05,
+      "loss": 0.4025,
+      "step": 3726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.823463997775924e-05,
+      "loss": 0.358,
+      "step": 3727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.822073950514317e-05,
+      "loss": 0.3118,
+      "step": 3728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.820683903252711e-05,
+      "loss": 0.3486,
+      "step": 3729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.819293855991104e-05,
+      "loss": 0.3597,
+      "step": 3730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.817903808729497e-05,
+      "loss": 0.4083,
+      "step": 3731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.81651376146789e-05,
+      "loss": 0.3833,
+      "step": 3732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.815123714206283e-05,
+      "loss": 0.3518,
+      "step": 3733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8137336669446766e-05,
+      "loss": 0.3655,
+      "step": 3734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8123436196830697e-05,
+      "loss": 0.3549,
+      "step": 3735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.810953572421463e-05,
+      "loss": 0.3908,
+      "step": 3736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.809563525159856e-05,
+      "loss": 0.3537,
+      "step": 3737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.808173477898249e-05,
+      "loss": 0.3869,
+      "step": 3738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.806783430636642e-05,
+      "loss": 0.4075,
+      "step": 3739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.805393383375035e-05,
+      "loss": 0.3455,
+      "step": 3740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.804003336113428e-05,
+      "loss": 0.3241,
+      "step": 3741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8026132888518214e-05,
+      "loss": 0.2947,
+      "step": 3742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8012232415902144e-05,
+      "loss": 0.3924,
+      "step": 3743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7998331943286074e-05,
+      "loss": 0.3771,
+      "step": 3744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7984431470670004e-05,
+      "loss": 0.4211,
+      "step": 3745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7970530998053934e-05,
+      "loss": 0.3094,
+      "step": 3746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.795663052543787e-05,
+      "loss": 0.4276,
+      "step": 3747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.79427300528218e-05,
+      "loss": 0.3755,
+      "step": 3748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.792882958020573e-05,
+      "loss": 0.3025,
+      "step": 3749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.791492910758966e-05,
+      "loss": 0.4094,
+      "step": 3750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.790102863497359e-05,
+      "loss": 0.35,
+      "step": 3751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.788712816235753e-05,
+      "loss": 0.3348,
+      "step": 3752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.787322768974145e-05,
+      "loss": 0.3246,
+      "step": 3753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.785932721712538e-05,
+      "loss": 0.3417,
+      "step": 3754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.784542674450931e-05,
+      "loss": 0.3549,
+      "step": 3755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.783152627189325e-05,
+      "loss": 0.3368,
+      "step": 3756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.781762579927718e-05,
+      "loss": 0.3621,
+      "step": 3757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.780372532666111e-05,
+      "loss": 0.3355,
+      "step": 3758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.778982485404504e-05,
+      "loss": 0.3602,
+      "step": 3759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.777592438142897e-05,
+      "loss": 0.3493,
+      "step": 3760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7762023908812905e-05,
+      "loss": 0.341,
+      "step": 3761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7748123436196835e-05,
+      "loss": 0.4012,
+      "step": 3762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7734222963580765e-05,
+      "loss": 0.4074,
+      "step": 3763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7720322490964696e-05,
+      "loss": 0.3356,
+      "step": 3764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7706422018348626e-05,
+      "loss": 0.3114,
+      "step": 3765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7692521545732556e-05,
+      "loss": 0.32,
+      "step": 3766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7678621073116486e-05,
+      "loss": 0.3413,
+      "step": 3767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7664720600500416e-05,
+      "loss": 0.3169,
+      "step": 3768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7650820127884346e-05,
+      "loss": 0.3248,
+      "step": 3769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.763691965526828e-05,
+      "loss": 0.3399,
+      "step": 3770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.762301918265221e-05,
+      "loss": 0.3718,
+      "step": 3771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.760911871003614e-05,
+      "loss": 0.3471,
+      "step": 3772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.759521823742007e-05,
+      "loss": 0.3454,
+      "step": 3773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.758131776480401e-05,
+      "loss": 0.2941,
+      "step": 3774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.756741729218794e-05,
+      "loss": 0.3783,
+      "step": 3775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.755351681957187e-05,
+      "loss": 0.307,
+      "step": 3776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.75396163469558e-05,
+      "loss": 0.3423,
+      "step": 3777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.752571587433973e-05,
+      "loss": 0.3211,
+      "step": 3778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751181540172366e-05,
+      "loss": 0.3775,
+      "step": 3779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749791492910759e-05,
+      "loss": 0.353,
+      "step": 3780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.748401445649152e-05,
+      "loss": 0.3585,
+      "step": 3781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747011398387545e-05,
+      "loss": 0.3661,
+      "step": 3782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745621351125939e-05,
+      "loss": 0.3287,
+      "step": 3783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744231303864332e-05,
+      "loss": 0.3724,
+      "step": 3784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742841256602725e-05,
+      "loss": 0.3701,
+      "step": 3785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741451209341118e-05,
+      "loss": 0.4,
+      "step": 3786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740061162079511e-05,
+      "loss": 0.3748,
+      "step": 3787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7386711148179044e-05,
+      "loss": 0.3444,
+      "step": 3788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7372810675562974e-05,
+      "loss": 0.3693,
+      "step": 3789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7358910202946904e-05,
+      "loss": 0.3817,
+      "step": 3790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7345009730330834e-05,
+      "loss": 0.3504,
+      "step": 3791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7331109257714764e-05,
+      "loss": 0.3888,
+      "step": 3792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7317208785098694e-05,
+      "loss": 0.3421,
+      "step": 3793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7303308312482625e-05,
+      "loss": 0.3639,
+      "step": 3794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7289407839866555e-05,
+      "loss": 0.3659,
+      "step": 3795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7275507367250485e-05,
+      "loss": 0.3664,
+      "step": 3796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726160689463442e-05,
+      "loss": 0.3999,
+      "step": 3797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.724770642201835e-05,
+      "loss": 0.317,
+      "step": 3798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.723380594940228e-05,
+      "loss": 0.338,
+      "step": 3799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.721990547678621e-05,
+      "loss": 0.3716,
+      "step": 3800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720600500417015e-05,
+      "loss": 0.368,
+      "step": 3801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.719210453155408e-05,
+      "loss": 0.2807,
+      "step": 3802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717820405893801e-05,
+      "loss": 0.3502,
+      "step": 3803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.716430358632194e-05,
+      "loss": 0.3564,
+      "step": 3804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715040311370587e-05,
+      "loss": 0.3242,
+      "step": 3805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71365026410898e-05,
+      "loss": 0.3468,
+      "step": 3806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712260216847373e-05,
+      "loss": 0.4369,
+      "step": 3807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710870169585766e-05,
+      "loss": 0.3578,
+      "step": 3808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709480122324159e-05,
+      "loss": 0.3477,
+      "step": 3809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7080900750625526e-05,
+      "loss": 0.4038,
+      "step": 3810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7067000278009456e-05,
+      "loss": 0.3727,
+      "step": 3811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7053099805393386e-05,
+      "loss": 0.4073,
+      "step": 3812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7039199332777316e-05,
+      "loss": 0.3457,
+      "step": 3813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7025298860161246e-05,
+      "loss": 0.4389,
+      "step": 3814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.701139838754518e-05,
+      "loss": 0.2953,
+      "step": 3815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.699749791492911e-05,
+      "loss": 0.4289,
+      "step": 3816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698359744231304e-05,
+      "loss": 0.2676,
+      "step": 3817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696969696969697e-05,
+      "loss": 0.3665,
+      "step": 3818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.69557964970809e-05,
+      "loss": 0.343,
+      "step": 3819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694189602446483e-05,
+      "loss": 0.3497,
+      "step": 3820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.692799555184876e-05,
+      "loss": 0.287,
+      "step": 3821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6914095079232693e-05,
+      "loss": 0.3543,
+      "step": 3822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6900194606616623e-05,
+      "loss": 0.3433,
+      "step": 3823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.688629413400056e-05,
+      "loss": 0.3897,
+      "step": 3824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.687239366138449e-05,
+      "loss": 0.36,
+      "step": 3825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685849318876842e-05,
+      "loss": 0.3827,
+      "step": 3826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684459271615235e-05,
+      "loss": 0.3171,
+      "step": 3827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683069224353628e-05,
+      "loss": 0.3262,
+      "step": 3828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.681679177092022e-05,
+      "loss": 0.3343,
+      "step": 3829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.680289129830415e-05,
+      "loss": 0.403,
+      "step": 3830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.678899082568808e-05,
+      "loss": 0.3466,
+      "step": 3831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6775090353072e-05,
+      "loss": 0.3764,
+      "step": 3832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676118988045594e-05,
+      "loss": 0.321,
+      "step": 3833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.674728940783987e-05,
+      "loss": 0.3879,
+      "step": 3834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.67333889352238e-05,
+      "loss": 0.2971,
+      "step": 3835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.671948846260773e-05,
+      "loss": 0.3356,
+      "step": 3836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6705587989991665e-05,
+      "loss": 0.3061,
+      "step": 3837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6691687517375595e-05,
+      "loss": 0.3457,
+      "step": 3838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6677787044759525e-05,
+      "loss": 0.3926,
+      "step": 3839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6663886572143455e-05,
+      "loss": 0.3478,
+      "step": 3840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6649986099527385e-05,
+      "loss": 0.347,
+      "step": 3841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.663608562691132e-05,
+      "loss": 0.339,
+      "step": 3842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.662218515429525e-05,
+      "loss": 0.4142,
+      "step": 3843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.660828468167918e-05,
+      "loss": 0.3392,
+      "step": 3844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6594384209063105e-05,
+      "loss": 0.3752,
+      "step": 3845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.658048373644704e-05,
+      "loss": 0.3339,
+      "step": 3846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.656658326383097e-05,
+      "loss": 0.3381,
+      "step": 3847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.65526827912149e-05,
+      "loss": 0.3809,
+      "step": 3848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.653878231859883e-05,
+      "loss": 0.3304,
+      "step": 3849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.652488184598276e-05,
+      "loss": 0.3594,
+      "step": 3850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65109813733667e-05,
+      "loss": 0.3475,
+      "step": 3851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649708090075063e-05,
+      "loss": 0.2991,
+      "step": 3852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.648318042813456e-05,
+      "loss": 0.3529,
+      "step": 3853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646927995551849e-05,
+      "loss": 0.4256,
+      "step": 3854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.645537948290242e-05,
+      "loss": 0.3413,
+      "step": 3855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6441479010286356e-05,
+      "loss": 0.3502,
+      "step": 3856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6427578537670286e-05,
+      "loss": 0.3287,
+      "step": 3857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.641367806505421e-05,
+      "loss": 0.3849,
+      "step": 3858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.639977759243814e-05,
+      "loss": 0.3117,
+      "step": 3859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6385877119822077e-05,
+      "loss": 0.3267,
+      "step": 3860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6371976647206007e-05,
+      "loss": 0.3299,
+      "step": 3861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635807617458994e-05,
+      "loss": 0.325,
+      "step": 3862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634417570197387e-05,
+      "loss": 0.3428,
+      "step": 3863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6330275229357804e-05,
+      "loss": 0.3013,
+      "step": 3864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6316374756741734e-05,
+      "loss": 0.3467,
+      "step": 3865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6302474284125664e-05,
+      "loss": 0.3703,
+      "step": 3866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6288573811509594e-05,
+      "loss": 0.2913,
+      "step": 3867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6274673338893524e-05,
+      "loss": 0.365,
+      "step": 3868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.626077286627746e-05,
+      "loss": 0.3511,
+      "step": 3869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.624687239366139e-05,
+      "loss": 0.3715,
+      "step": 3870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623297192104532e-05,
+      "loss": 0.3502,
+      "step": 3871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6219071448429244e-05,
+      "loss": 0.3596,
+      "step": 3872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620517097581318e-05,
+      "loss": 0.3848,
+      "step": 3873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619127050319711e-05,
+      "loss": 0.3857,
+      "step": 3874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.617737003058104e-05,
+      "loss": 0.3546,
+      "step": 3875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616346955796497e-05,
+      "loss": 0.3268,
+      "step": 3876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.61495690853489e-05,
+      "loss": 0.3316,
+      "step": 3877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613566861273284e-05,
+      "loss": 0.3473,
+      "step": 3878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612176814011677e-05,
+      "loss": 0.3905,
+      "step": 3879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.61078676675007e-05,
+      "loss": 0.372,
+      "step": 3880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609396719488463e-05,
+      "loss": 0.3757,
+      "step": 3881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608006672226856e-05,
+      "loss": 0.3464,
+      "step": 3882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6066166249652495e-05,
+      "loss": 0.367,
+      "step": 3883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6052265777036425e-05,
+      "loss": 0.378,
+      "step": 3884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603836530442035e-05,
+      "loss": 0.3419,
+      "step": 3885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.602446483180428e-05,
+      "loss": 0.2828,
+      "step": 3886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6010564359188215e-05,
+      "loss": 0.3618,
+      "step": 3887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5996663886572145e-05,
+      "loss": 0.337,
+      "step": 3888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5982763413956075e-05,
+      "loss": 0.3757,
+      "step": 3889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5968862941340006e-05,
+      "loss": 0.3754,
+      "step": 3890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5954962468723936e-05,
+      "loss": 0.3236,
+      "step": 3891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594106199610787e-05,
+      "loss": 0.3563,
+      "step": 3892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59271615234918e-05,
+      "loss": 0.3432,
+      "step": 3893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.591326105087573e-05,
+      "loss": 0.3876,
+      "step": 3894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.589936057825966e-05,
+      "loss": 0.3256,
+      "step": 3895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.58854601056436e-05,
+      "loss": 0.3283,
+      "step": 3896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.587155963302753e-05,
+      "loss": 0.2858,
+      "step": 3897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585765916041145e-05,
+      "loss": 0.3705,
+      "step": 3898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.584375868779538e-05,
+      "loss": 0.408,
+      "step": 3899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.582985821517932e-05,
+      "loss": 0.3448,
+      "step": 3900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581595774256325e-05,
+      "loss": 0.375,
+      "step": 3901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.580205726994718e-05,
+      "loss": 0.3719,
+      "step": 3902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.578815679733111e-05,
+      "loss": 0.3811,
+      "step": 3903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.577425632471504e-05,
+      "loss": 0.3899,
+      "step": 3904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.576035585209898e-05,
+      "loss": 0.3779,
+      "step": 3905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.574645537948291e-05,
+      "loss": 0.3902,
+      "step": 3906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.573255490686684e-05,
+      "loss": 0.3327,
+      "step": 3907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.571865443425077e-05,
+      "loss": 0.3707,
+      "step": 3908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57047539616347e-05,
+      "loss": 0.3511,
+      "step": 3909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5690853489018634e-05,
+      "loss": 0.3503,
+      "step": 3910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.567695301640256e-05,
+      "loss": 0.3584,
+      "step": 3911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.566305254378649e-05,
+      "loss": 0.3533,
+      "step": 3912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.564915207117042e-05,
+      "loss": 0.3592,
+      "step": 3913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5635251598554354e-05,
+      "loss": 0.3139,
+      "step": 3914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5621351125938284e-05,
+      "loss": 0.3922,
+      "step": 3915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5607450653322214e-05,
+      "loss": 0.34,
+      "step": 3916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5593550180706144e-05,
+      "loss": 0.3225,
+      "step": 3917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5579649708090074e-05,
+      "loss": 0.3111,
+      "step": 3918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.556574923547401e-05,
+      "loss": 0.3795,
+      "step": 3919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555184876285794e-05,
+      "loss": 0.3072,
+      "step": 3920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.553794829024187e-05,
+      "loss": 0.3515,
+      "step": 3921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.55240478176258e-05,
+      "loss": 0.3332,
+      "step": 3922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.551014734500973e-05,
+      "loss": 0.3807,
+      "step": 3923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.549624687239366e-05,
+      "loss": 0.3138,
+      "step": 3924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.548234639977759e-05,
+      "loss": 0.3734,
+      "step": 3925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.546844592716152e-05,
+      "loss": 0.3422,
+      "step": 3926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.3451,
+      "step": 3927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.544064498192939e-05,
+      "loss": 0.353,
+      "step": 3928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.542674450931332e-05,
+      "loss": 0.3657,
+      "step": 3929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.541284403669725e-05,
+      "loss": 0.3251,
+      "step": 3930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.539894356408118e-05,
+      "loss": 0.3383,
+      "step": 3931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5385043091465116e-05,
+      "loss": 0.3364,
+      "step": 3932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5371142618849046e-05,
+      "loss": 0.3418,
+      "step": 3933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5357242146232976e-05,
+      "loss": 0.3773,
+      "step": 3934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5343341673616906e-05,
+      "loss": 0.3628,
+      "step": 3935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5329441201000836e-05,
+      "loss": 0.3819,
+      "step": 3936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.531554072838477e-05,
+      "loss": 0.3724,
+      "step": 3937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5301640255768696e-05,
+      "loss": 0.34,
+      "step": 3938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5287739783152626e-05,
+      "loss": 0.3271,
+      "step": 3939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5273839310536556e-05,
+      "loss": 0.3645,
+      "step": 3940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.525993883792049e-05,
+      "loss": 0.3975,
+      "step": 3941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.524603836530442e-05,
+      "loss": 0.3267,
+      "step": 3942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.523213789268835e-05,
+      "loss": 0.3449,
+      "step": 3943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.521823742007228e-05,
+      "loss": 0.3575,
+      "step": 3944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.520433694745621e-05,
+      "loss": 0.3498,
+      "step": 3945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.519043647484015e-05,
+      "loss": 0.3396,
+      "step": 3946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.517653600222408e-05,
+      "loss": 0.3814,
+      "step": 3947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.516263552960801e-05,
+      "loss": 0.3818,
+      "step": 3948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.514873505699194e-05,
+      "loss": 0.3406,
+      "step": 3949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513483458437587e-05,
+      "loss": 0.37,
+      "step": 3950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.51209341117598e-05,
+      "loss": 0.3803,
+      "step": 3951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510703363914373e-05,
+      "loss": 0.3939,
+      "step": 3952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.509313316652766e-05,
+      "loss": 0.3325,
+      "step": 3953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.507923269391159e-05,
+      "loss": 0.3633,
+      "step": 3954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.506533222129553e-05,
+      "loss": 0.3395,
+      "step": 3955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505143174867946e-05,
+      "loss": 0.3175,
+      "step": 3956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.503753127606339e-05,
+      "loss": 0.3595,
+      "step": 3957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.502363080344732e-05,
+      "loss": 0.367,
+      "step": 3958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5009730330831254e-05,
+      "loss": 0.3636,
+      "step": 3959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4995829858215185e-05,
+      "loss": 0.3874,
+      "step": 3960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4981929385599115e-05,
+      "loss": 0.3143,
+      "step": 3961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4968028912983045e-05,
+      "loss": 0.3845,
+      "step": 3962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4954128440366975e-05,
+      "loss": 0.3638,
+      "step": 3963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4940227967750905e-05,
+      "loss": 0.331,
+      "step": 3964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4926327495134835e-05,
+      "loss": 0.3149,
+      "step": 3965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4912427022518765e-05,
+      "loss": 0.3418,
+      "step": 3966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4898526549902695e-05,
+      "loss": 0.3308,
+      "step": 3967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.488462607728663e-05,
+      "loss": 0.3353,
+      "step": 3968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.487072560467056e-05,
+      "loss": 0.365,
+      "step": 3969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485682513205449e-05,
+      "loss": 0.4245,
+      "step": 3970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484292465943842e-05,
+      "loss": 0.3261,
+      "step": 3971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482902418682235e-05,
+      "loss": 0.3561,
+      "step": 3972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.481512371420629e-05,
+      "loss": 0.2896,
+      "step": 3973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.480122324159022e-05,
+      "loss": 0.3358,
+      "step": 3974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.478732276897415e-05,
+      "loss": 0.3896,
+      "step": 3975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477342229635808e-05,
+      "loss": 0.3232,
+      "step": 3976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.475952182374201e-05,
+      "loss": 0.3793,
+      "step": 3977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.474562135112594e-05,
+      "loss": 0.346,
+      "step": 3978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.473172087850987e-05,
+      "loss": 0.3704,
+      "step": 3979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47178204058938e-05,
+      "loss": 0.3364,
+      "step": 3980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470391993327773e-05,
+      "loss": 0.3672,
+      "step": 3981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4690019460661666e-05,
+      "loss": 0.3147,
+      "step": 3982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4676118988045596e-05,
+      "loss": 0.3229,
+      "step": 3983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4662218515429526e-05,
+      "loss": 0.353,
+      "step": 3984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4648318042813456e-05,
+      "loss": 0.3048,
+      "step": 3985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4634417570197387e-05,
+      "loss": 0.3534,
+      "step": 3986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4634417570197387e-05,
+      "loss": 0.3739,
+      "step": 3987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4620517097581323e-05,
+      "loss": 0.3945,
+      "step": 3988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4606616624965253e-05,
+      "loss": 0.3211,
+      "step": 3989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4592716152349184e-05,
+      "loss": 0.3368,
+      "step": 3990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4578815679733114e-05,
+      "loss": 0.3243,
+      "step": 3991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4564915207117044e-05,
+      "loss": 0.3213,
+      "step": 3992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4551014734500974e-05,
+      "loss": 0.3498,
+      "step": 3993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4537114261884904e-05,
+      "loss": 0.3533,
+      "step": 3994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4523213789268834e-05,
+      "loss": 0.3671,
+      "step": 3995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.450931331665277e-05,
+      "loss": 0.357,
+      "step": 3996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.44954128440367e-05,
+      "loss": 0.3899,
+      "step": 3997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448151237142063e-05,
+      "loss": 0.3429,
+      "step": 3998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446761189880456e-05,
+      "loss": 0.3537,
+      "step": 3999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445371142618849e-05,
+      "loss": 0.3482,
+      "step": 4000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443981095357243e-05,
+      "loss": 0.37,
+      "step": 4001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.442591048095636e-05,
+      "loss": 0.3727,
+      "step": 4002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.441201000834029e-05,
+      "loss": 0.3315,
+      "step": 4003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.439810953572422e-05,
+      "loss": 0.366,
+      "step": 4004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.438420906310815e-05,
+      "loss": 0.3716,
+      "step": 4005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.437030859049208e-05,
+      "loss": 0.3761,
+      "step": 4006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.435640811787601e-05,
+      "loss": 0.4068,
+      "step": 4007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.434250764525994e-05,
+      "loss": 0.3244,
+      "step": 4008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.432860717264387e-05,
+      "loss": 0.3115,
+      "step": 4009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4314706700027805e-05,
+      "loss": 0.3548,
+      "step": 4010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4300806227411735e-05,
+      "loss": 0.3326,
+      "step": 4011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4286905754795665e-05,
+      "loss": 0.3363,
+      "step": 4012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4273005282179595e-05,
+      "loss": 0.3346,
+      "step": 4013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4259104809563525e-05,
+      "loss": 0.4335,
+      "step": 4014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.424520433694746e-05,
+      "loss": 0.3572,
+      "step": 4015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.423130386433139e-05,
+      "loss": 0.3452,
+      "step": 4016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.421740339171532e-05,
+      "loss": 0.3353,
+      "step": 4017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4203502919099246e-05,
+      "loss": 0.3377,
+      "step": 4018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.418960244648318e-05,
+      "loss": 0.3134,
+      "step": 4019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.417570197386711e-05,
+      "loss": 0.3179,
+      "step": 4020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.416180150125104e-05,
+      "loss": 0.316,
+      "step": 4021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.414790102863497e-05,
+      "loss": 0.3515,
+      "step": 4022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.413400055601891e-05,
+      "loss": 0.3069,
+      "step": 4023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.412010008340284e-05,
+      "loss": 0.3087,
+      "step": 4024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.410619961078677e-05,
+      "loss": 0.3758,
+      "step": 4025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.40922991381707e-05,
+      "loss": 0.3863,
+      "step": 4026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.407839866555463e-05,
+      "loss": 0.332,
+      "step": 4027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4064498192938567e-05,
+      "loss": 0.3485,
+      "step": 4028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.40505977203225e-05,
+      "loss": 0.34,
+      "step": 4029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.403669724770643e-05,
+      "loss": 0.3763,
+      "step": 4030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.402279677509035e-05,
+      "loss": 0.3643,
+      "step": 4031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.400889630247429e-05,
+      "loss": 0.3152,
+      "step": 4032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.399499582985822e-05,
+      "loss": 0.3888,
+      "step": 4033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.398109535724215e-05,
+      "loss": 0.3689,
+      "step": 4034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.396719488462608e-05,
+      "loss": 0.388,
+      "step": 4035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.395329441201001e-05,
+      "loss": 0.3448,
+      "step": 4036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3939393939393944e-05,
+      "loss": 0.2845,
+      "step": 4037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3925493466777874e-05,
+      "loss": 0.3468,
+      "step": 4038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3911592994161804e-05,
+      "loss": 0.3874,
+      "step": 4039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3897692521545734e-05,
+      "loss": 0.3567,
+      "step": 4040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3883792048929664e-05,
+      "loss": 0.3445,
+      "step": 4041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.38698915763136e-05,
+      "loss": 0.2652,
+      "step": 4042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.385599110369753e-05,
+      "loss": 0.3425,
+      "step": 4043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.384209063108146e-05,
+      "loss": 0.3522,
+      "step": 4044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3828190158465384e-05,
+      "loss": 0.3641,
+      "step": 4045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.381428968584932e-05,
+      "loss": 0.3099,
+      "step": 4046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.380038921323325e-05,
+      "loss": 0.3793,
+      "step": 4047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.378648874061718e-05,
+      "loss": 0.3643,
+      "step": 4048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.377258826800111e-05,
+      "loss": 0.3348,
+      "step": 4049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.375868779538504e-05,
+      "loss": 0.3801,
+      "step": 4050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.374478732276898e-05,
+      "loss": 0.3185,
+      "step": 4051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.373088685015291e-05,
+      "loss": 0.3517,
+      "step": 4052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.371698637753684e-05,
+      "loss": 0.3517,
+      "step": 4053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.370308590492077e-05,
+      "loss": 0.3141,
+      "step": 4054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3689185432304705e-05,
+      "loss": 0.3651,
+      "step": 4055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3675284959688636e-05,
+      "loss": 0.3424,
+      "step": 4056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3661384487072566e-05,
+      "loss": 0.313,
+      "step": 4057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.364748401445649e-05,
+      "loss": 0.4013,
+      "step": 4058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3633583541840426e-05,
+      "loss": 0.434,
+      "step": 4059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3619683069224356e-05,
+      "loss": 0.3,
+      "step": 4060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3605782596608286e-05,
+      "loss": 0.367,
+      "step": 4061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3591882123992216e-05,
+      "loss": 0.3481,
+      "step": 4062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3577981651376146e-05,
+      "loss": 0.3701,
+      "step": 4063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.356408117876008e-05,
+      "loss": 0.2938,
+      "step": 4064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.355018070614401e-05,
+      "loss": 0.3675,
+      "step": 4065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.353628023352794e-05,
+      "loss": 0.373,
+      "step": 4066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.352237976091187e-05,
+      "loss": 0.3613,
+      "step": 4067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.35084792882958e-05,
+      "loss": 0.339,
+      "step": 4068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.349457881567974e-05,
+      "loss": 0.2903,
+      "step": 4069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.348067834306367e-05,
+      "loss": 0.2893,
+      "step": 4070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.346677787044759e-05,
+      "loss": 0.3721,
+      "step": 4071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.345287739783152e-05,
+      "loss": 0.3825,
+      "step": 4072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.343897692521546e-05,
+      "loss": 0.3355,
+      "step": 4073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.342507645259939e-05,
+      "loss": 0.3917,
+      "step": 4074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.341117597998332e-05,
+      "loss": 0.3237,
+      "step": 4075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.339727550736725e-05,
+      "loss": 0.2934,
+      "step": 4076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.338337503475118e-05,
+      "loss": 0.3207,
+      "step": 4077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.336947456213512e-05,
+      "loss": 0.3588,
+      "step": 4078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.335557408951905e-05,
+      "loss": 0.3558,
+      "step": 4079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.334167361690298e-05,
+      "loss": 0.3366,
+      "step": 4080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.332777314428691e-05,
+      "loss": 0.3538,
+      "step": 4081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3313872671670844e-05,
+      "loss": 0.3776,
+      "step": 4082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3299972199054774e-05,
+      "loss": 0.4053,
+      "step": 4083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.32860717264387e-05,
+      "loss": 0.3294,
+      "step": 4084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.327217125382263e-05,
+      "loss": 0.3741,
+      "step": 4085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3258270781206565e-05,
+      "loss": 0.3574,
+      "step": 4086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3244370308590495e-05,
+      "loss": 0.3521,
+      "step": 4087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3230469835974425e-05,
+      "loss": 0.3276,
+      "step": 4088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3216569363358355e-05,
+      "loss": 0.425,
+      "step": 4089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3202668890742285e-05,
+      "loss": 0.298,
+      "step": 4090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.318876841812622e-05,
+      "loss": 0.3545,
+      "step": 4091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.317486794551015e-05,
+      "loss": 0.4278,
+      "step": 4092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.316096747289408e-05,
+      "loss": 0.3817,
+      "step": 4093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.314706700027801e-05,
+      "loss": 0.3433,
+      "step": 4094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.313316652766194e-05,
+      "loss": 0.3196,
+      "step": 4095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.311926605504588e-05,
+      "loss": 0.2855,
+      "step": 4096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.31053655824298e-05,
+      "loss": 0.3257,
+      "step": 4097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.309146510981373e-05,
+      "loss": 0.2889,
+      "step": 4098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.307756463719766e-05,
+      "loss": 0.3574,
+      "step": 4099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.30636641645816e-05,
+      "loss": 0.3105,
+      "step": 4100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.304976369196553e-05,
+      "loss": 0.3292,
+      "step": 4101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.303586321934946e-05,
+      "loss": 0.3178,
+      "step": 4102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.302196274673339e-05,
+      "loss": 0.3623,
+      "step": 4103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.300806227411732e-05,
+      "loss": 0.3546,
+      "step": 4104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2994161801501256e-05,
+      "loss": 0.3827,
+      "step": 4105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2980261328885186e-05,
+      "loss": 0.3639,
+      "step": 4106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2966360856269116e-05,
+      "loss": 0.2868,
+      "step": 4107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2952460383653046e-05,
+      "loss": 0.3948,
+      "step": 4108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2938559911036976e-05,
+      "loss": 0.3444,
+      "step": 4109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.292465943842091e-05,
+      "loss": 0.3634,
+      "step": 4110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2910758965804836e-05,
+      "loss": 0.3331,
+      "step": 4111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2896858493188767e-05,
+      "loss": 0.2969,
+      "step": 4112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2882958020572697e-05,
+      "loss": 0.3082,
+      "step": 4113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2869057547956633e-05,
+      "loss": 0.3304,
+      "step": 4114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2855157075340563e-05,
+      "loss": 0.3786,
+      "step": 4115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2841256602724494e-05,
+      "loss": 0.3865,
+      "step": 4116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2827356130108424e-05,
+      "loss": 0.3562,
+      "step": 4117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.281345565749236e-05,
+      "loss": 0.3512,
+      "step": 4118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.279955518487629e-05,
+      "loss": 0.3588,
+      "step": 4119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.278565471226022e-05,
+      "loss": 0.3724,
+      "step": 4120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.277175423964415e-05,
+      "loss": 0.3746,
+      "step": 4121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.275785376702808e-05,
+      "loss": 0.2928,
+      "step": 4122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.274395329441202e-05,
+      "loss": 0.3378,
+      "step": 4123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.273005282179594e-05,
+      "loss": 0.3735,
+      "step": 4124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.271615234917987e-05,
+      "loss": 0.3312,
+      "step": 4125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.27022518765638e-05,
+      "loss": 0.3234,
+      "step": 4126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.268835140394774e-05,
+      "loss": 0.3099,
+      "step": 4127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.267445093133167e-05,
+      "loss": 0.3122,
+      "step": 4128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.26605504587156e-05,
+      "loss": 0.3487,
+      "step": 4129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.264664998609953e-05,
+      "loss": 0.3709,
+      "step": 4130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.263274951348346e-05,
+      "loss": 0.3783,
+      "step": 4131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2618849040867395e-05,
+      "loss": 0.4159,
+      "step": 4132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2604948568251325e-05,
+      "loss": 0.2961,
+      "step": 4133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2591048095635255e-05,
+      "loss": 0.3424,
+      "step": 4134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2577147623019185e-05,
+      "loss": 0.3416,
+      "step": 4135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2563247150403115e-05,
+      "loss": 0.38,
+      "step": 4136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2549346677787045e-05,
+      "loss": 0.3792,
+      "step": 4137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2535446205170975e-05,
+      "loss": 0.4089,
+      "step": 4138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2521545732554905e-05,
+      "loss": 0.3238,
+      "step": 4139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2507645259938835e-05,
+      "loss": 0.3782,
+      "step": 4140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.249374478732277e-05,
+      "loss": 0.3486,
+      "step": 4141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.24798443147067e-05,
+      "loss": 0.3438,
+      "step": 4142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.246594384209063e-05,
+      "loss": 0.2678,
+      "step": 4143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.245204336947456e-05,
+      "loss": 0.3493,
+      "step": 4144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.24381428968585e-05,
+      "loss": 0.3342,
+      "step": 4145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 0.354,
+      "step": 4146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.241034195162636e-05,
+      "loss": 0.3465,
+      "step": 4147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.239644147901029e-05,
+      "loss": 0.3288,
+      "step": 4148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.238254100639422e-05,
+      "loss": 0.3791,
+      "step": 4149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.236864053377815e-05,
+      "loss": 0.3993,
+      "step": 4150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.235474006116208e-05,
+      "loss": 0.338,
+      "step": 4151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.234083958854601e-05,
+      "loss": 0.3593,
+      "step": 4152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.232693911592994e-05,
+      "loss": 0.4099,
+      "step": 4153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.231303864331388e-05,
+      "loss": 0.3367,
+      "step": 4154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.229913817069781e-05,
+      "loss": 0.3202,
+      "step": 4155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.228523769808174e-05,
+      "loss": 0.3118,
+      "step": 4156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.227133722546567e-05,
+      "loss": 0.3789,
+      "step": 4157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.22574367528496e-05,
+      "loss": 0.3683,
+      "step": 4158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2243536280233534e-05,
+      "loss": 0.3163,
+      "step": 4159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2229635807617464e-05,
+      "loss": 0.3639,
+      "step": 4160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2215735335001394e-05,
+      "loss": 0.3389,
+      "step": 4161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2201834862385324e-05,
+      "loss": 0.2851,
+      "step": 4162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2187934389769254e-05,
+      "loss": 0.3867,
+      "step": 4163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2174033917153184e-05,
+      "loss": 0.3124,
+      "step": 4164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2160133444537114e-05,
+      "loss": 0.3655,
+      "step": 4165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2146232971921044e-05,
+      "loss": 0.3939,
+      "step": 4166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2132332499304974e-05,
+      "loss": 0.4191,
+      "step": 4167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.211843202668891e-05,
+      "loss": 0.3441,
+      "step": 4168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.210453155407284e-05,
+      "loss": 0.3529,
+      "step": 4169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.209063108145677e-05,
+      "loss": 0.3442,
+      "step": 4170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.20767306088407e-05,
+      "loss": 0.3081,
+      "step": 4171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.206283013622463e-05,
+      "loss": 0.3666,
+      "step": 4172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.204892966360857e-05,
+      "loss": 0.3501,
+      "step": 4173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.20350291909925e-05,
+      "loss": 0.341,
+      "step": 4174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.202112871837643e-05,
+      "loss": 0.2905,
+      "step": 4175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.200722824576036e-05,
+      "loss": 0.3285,
+      "step": 4176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.199332777314429e-05,
+      "loss": 0.3422,
+      "step": 4177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.197942730052822e-05,
+      "loss": 0.3755,
+      "step": 4178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.196552682791215e-05,
+      "loss": 0.3564,
+      "step": 4179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.195162635529608e-05,
+      "loss": 0.2984,
+      "step": 4180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1937725882680015e-05,
+      "loss": 0.3391,
+      "step": 4181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1923825410063946e-05,
+      "loss": 0.3398,
+      "step": 4182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1909924937447876e-05,
+      "loss": 0.382,
+      "step": 4183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1896024464831806e-05,
+      "loss": 0.3522,
+      "step": 4184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1882123992215736e-05,
+      "loss": 0.3267,
+      "step": 4185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.186822351959967e-05,
+      "loss": 0.3468,
+      "step": 4186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.18543230469836e-05,
+      "loss": 0.3741,
+      "step": 4187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.184042257436753e-05,
+      "loss": 0.3412,
+      "step": 4188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.182652210175146e-05,
+      "loss": 0.3743,
+      "step": 4189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.181262162913539e-05,
+      "loss": 0.3322,
+      "step": 4190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.179872115651932e-05,
+      "loss": 0.381,
+      "step": 4191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.178482068390325e-05,
+      "loss": 0.3351,
+      "step": 4192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.177092021128718e-05,
+      "loss": 0.35,
+      "step": 4193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.175701973867111e-05,
+      "loss": 0.3692,
+      "step": 4194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.174311926605505e-05,
+      "loss": 0.3899,
+      "step": 4195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.172921879343898e-05,
+      "loss": 0.381,
+      "step": 4196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.171531832082291e-05,
+      "loss": 0.3245,
+      "step": 4197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.170141784820684e-05,
+      "loss": 0.3275,
+      "step": 4198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.168751737559077e-05,
+      "loss": 0.3783,
+      "step": 4199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.167361690297471e-05,
+      "loss": 0.3678,
+      "step": 4200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.165971643035864e-05,
+      "loss": 0.3348,
+      "step": 4201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.164581595774257e-05,
+      "loss": 0.336,
+      "step": 4202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.163191548512649e-05,
+      "loss": 0.4305,
+      "step": 4203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.161801501251043e-05,
+      "loss": 0.3843,
+      "step": 4204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.160411453989436e-05,
+      "loss": 0.3591,
+      "step": 4205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.159021406727829e-05,
+      "loss": 0.3277,
+      "step": 4206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.157631359466222e-05,
+      "loss": 0.3938,
+      "step": 4207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1562413122046154e-05,
+      "loss": 0.3372,
+      "step": 4208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1548512649430084e-05,
+      "loss": 0.3581,
+      "step": 4209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1534612176814014e-05,
+      "loss": 0.3252,
+      "step": 4210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1520711704197944e-05,
+      "loss": 0.3121,
+      "step": 4211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1506811231581875e-05,
+      "loss": 0.3312,
+      "step": 4212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.149291075896581e-05,
+      "loss": 0.3008,
+      "step": 4213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.147901028634974e-05,
+      "loss": 0.3726,
+      "step": 4214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.146510981373367e-05,
+      "loss": 0.2908,
+      "step": 4215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1451209341117595e-05,
+      "loss": 0.3285,
+      "step": 4216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.143730886850153e-05,
+      "loss": 0.3672,
+      "step": 4217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.142340839588546e-05,
+      "loss": 0.3913,
+      "step": 4218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.140950792326939e-05,
+      "loss": 0.3278,
+      "step": 4219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.139560745065332e-05,
+      "loss": 0.3727,
+      "step": 4220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.138170697803725e-05,
+      "loss": 0.2948,
+      "step": 4221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.136780650542119e-05,
+      "loss": 0.3095,
+      "step": 4222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.135390603280512e-05,
+      "loss": 0.35,
+      "step": 4223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.134000556018905e-05,
+      "loss": 0.3891,
+      "step": 4224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.132610508757298e-05,
+      "loss": 0.3603,
+      "step": 4225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.131220461495691e-05,
+      "loss": 0.3462,
+      "step": 4226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1298304142340846e-05,
+      "loss": 0.4034,
+      "step": 4227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1284403669724776e-05,
+      "loss": 0.3921,
+      "step": 4228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1270503197108706e-05,
+      "loss": 0.3622,
+      "step": 4229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.125660272449263e-05,
+      "loss": 0.3451,
+      "step": 4230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1242702251876566e-05,
+      "loss": 0.3569,
+      "step": 4231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1228801779260496e-05,
+      "loss": 0.3805,
+      "step": 4232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1214901306644426e-05,
+      "loss": 0.3946,
+      "step": 4233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1201000834028356e-05,
+      "loss": 0.3544,
+      "step": 4234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1187100361412286e-05,
+      "loss": 0.3507,
+      "step": 4235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.117319988879622e-05,
+      "loss": 0.3374,
+      "step": 4236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.115929941618015e-05,
+      "loss": 0.3277,
+      "step": 4237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.114539894356408e-05,
+      "loss": 0.3093,
+      "step": 4238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1131498470948013e-05,
+      "loss": 0.3627,
+      "step": 4239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.111759799833195e-05,
+      "loss": 0.3175,
+      "step": 4240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.110369752571588e-05,
+      "loss": 0.3039,
+      "step": 4241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.108979705309981e-05,
+      "loss": 0.3333,
+      "step": 4242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1075896580483734e-05,
+      "loss": 0.2933,
+      "step": 4243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.106199610786767e-05,
+      "loss": 0.3478,
+      "step": 4244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.10480956352516e-05,
+      "loss": 0.3215,
+      "step": 4245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.103419516263553e-05,
+      "loss": 0.34,
+      "step": 4246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.102029469001946e-05,
+      "loss": 0.3545,
+      "step": 4247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.100639421740339e-05,
+      "loss": 0.3633,
+      "step": 4248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.099249374478733e-05,
+      "loss": 0.3429,
+      "step": 4249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.097859327217126e-05,
+      "loss": 0.366,
+      "step": 4250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.096469279955519e-05,
+      "loss": 0.4029,
+      "step": 4251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.095079232693912e-05,
+      "loss": 0.3179,
+      "step": 4252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.093689185432305e-05,
+      "loss": 0.3575,
+      "step": 4253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0922991381706985e-05,
+      "loss": 0.3641,
+      "step": 4254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.3961,
+      "step": 4255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.089519043647484e-05,
+      "loss": 0.336,
+      "step": 4256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.088128996385877e-05,
+      "loss": 0.3665,
+      "step": 4257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0867389491242705e-05,
+      "loss": 0.3047,
+      "step": 4258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0853489018626635e-05,
+      "loss": 0.3351,
+      "step": 4259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0839588546010565e-05,
+      "loss": 0.3381,
+      "step": 4260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0825688073394495e-05,
+      "loss": 0.2994,
+      "step": 4261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0811787600778425e-05,
+      "loss": 0.4007,
+      "step": 4262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.079788712816236e-05,
+      "loss": 0.351,
+      "step": 4263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.078398665554629e-05,
+      "loss": 0.3505,
+      "step": 4264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.077008618293022e-05,
+      "loss": 0.3346,
+      "step": 4265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.075618571031415e-05,
+      "loss": 0.4138,
+      "step": 4266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.074228523769809e-05,
+      "loss": 0.2755,
+      "step": 4267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.072838476508202e-05,
+      "loss": 0.3219,
+      "step": 4268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.071448429246594e-05,
+      "loss": 0.3336,
+      "step": 4269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.070058381984987e-05,
+      "loss": 0.3278,
+      "step": 4270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.068668334723381e-05,
+      "loss": 0.3578,
+      "step": 4271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.067278287461774e-05,
+      "loss": 0.3259,
+      "step": 4272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.065888240200167e-05,
+      "loss": 0.3256,
+      "step": 4273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.06449819293856e-05,
+      "loss": 0.3217,
+      "step": 4274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.063108145676953e-05,
+      "loss": 0.309,
+      "step": 4275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0617180984153466e-05,
+      "loss": 0.3291,
+      "step": 4276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0603280511537396e-05,
+      "loss": 0.3133,
+      "step": 4277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0589380038921327e-05,
+      "loss": 0.3668,
+      "step": 4278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0575479566305257e-05,
+      "loss": 0.3821,
+      "step": 4279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.056157909368919e-05,
+      "loss": 0.3603,
+      "step": 4280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0547678621073124e-05,
+      "loss": 0.3433,
+      "step": 4281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0533778148457054e-05,
+      "loss": 0.33,
+      "step": 4282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.051987767584098e-05,
+      "loss": 0.3445,
+      "step": 4283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.050597720322491e-05,
+      "loss": 0.3354,
+      "step": 4284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0492076730608844e-05,
+      "loss": 0.2566,
+      "step": 4285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0478176257992774e-05,
+      "loss": 0.3315,
+      "step": 4286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0464275785376704e-05,
+      "loss": 0.3745,
+      "step": 4287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0450375312760634e-05,
+      "loss": 0.286,
+      "step": 4288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0436474840144564e-05,
+      "loss": 0.3468,
+      "step": 4289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.04225743675285e-05,
+      "loss": 0.3334,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.040867389491243e-05,
+      "loss": 0.3424,
+      "step": 4291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.039477342229636e-05,
+      "loss": 0.3853,
+      "step": 4292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.038087294968029e-05,
+      "loss": 0.3782,
+      "step": 4293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.036697247706422e-05,
+      "loss": 0.3183,
+      "step": 4294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.035307200444816e-05,
+      "loss": 0.4223,
+      "step": 4295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.033917153183208e-05,
+      "loss": 0.3572,
+      "step": 4296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.032527105921601e-05,
+      "loss": 0.3537,
+      "step": 4297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.031137058659994e-05,
+      "loss": 0.3857,
+      "step": 4298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.029747011398388e-05,
+      "loss": 0.3314,
+      "step": 4299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.028356964136781e-05,
+      "loss": 0.3785,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.026966916875174e-05,
+      "loss": 0.3313,
+      "step": 4301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.025576869613567e-05,
+      "loss": 0.3268,
+      "step": 4302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0241868223519605e-05,
+      "loss": 0.3371,
+      "step": 4303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0227967750903535e-05,
+      "loss": 0.3412,
+      "step": 4304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0214067278287465e-05,
+      "loss": 0.3411,
+      "step": 4305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0200166805671395e-05,
+      "loss": 0.4177,
+      "step": 4306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0186266333055326e-05,
+      "loss": 0.39,
+      "step": 4307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.017236586043926e-05,
+      "loss": 0.3537,
+      "step": 4308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0158465387823186e-05,
+      "loss": 0.3544,
+      "step": 4309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0144564915207116e-05,
+      "loss": 0.3298,
+      "step": 4310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0130664442591046e-05,
+      "loss": 0.324,
+      "step": 4311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.011676396997498e-05,
+      "loss": 0.2943,
+      "step": 4312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010286349735891e-05,
+      "loss": 0.3134,
+      "step": 4313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008896302474284e-05,
+      "loss": 0.336,
+      "step": 4314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007506255212677e-05,
+      "loss": 0.4227,
+      "step": 4315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00611620795107e-05,
+      "loss": 0.362,
+      "step": 4316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004726160689464e-05,
+      "loss": 0.3854,
+      "step": 4317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003336113427857e-05,
+      "loss": 0.3408,
+      "step": 4318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00194606616625e-05,
+      "loss": 0.31,
+      "step": 4319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000556018904643e-05,
+      "loss": 0.3614,
+      "step": 4320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999165971643036e-05,
+      "loss": 0.3483,
+      "step": 4321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997775924381429e-05,
+      "loss": 0.37,
+      "step": 4322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996385877119822e-05,
+      "loss": 0.3356,
+      "step": 4323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994995829858215e-05,
+      "loss": 0.326,
+      "step": 4324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.993605782596608e-05,
+      "loss": 0.3626,
+      "step": 4325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.992215735335002e-05,
+      "loss": 0.3769,
+      "step": 4326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990825688073395e-05,
+      "loss": 0.337,
+      "step": 4327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.989435640811788e-05,
+      "loss": 0.3272,
+      "step": 4328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.988045593550181e-05,
+      "loss": 0.3377,
+      "step": 4329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9866555462885744e-05,
+      "loss": 0.3431,
+      "step": 4330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9852654990269674e-05,
+      "loss": 0.3518,
+      "step": 4331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9838754517653604e-05,
+      "loss": 0.4265,
+      "step": 4332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9824854045037534e-05,
+      "loss": 0.3097,
+      "step": 4333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9810953572421464e-05,
+      "loss": 0.376,
+      "step": 4334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9797053099805394e-05,
+      "loss": 0.348,
+      "step": 4335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9783152627189324e-05,
+      "loss": 0.3819,
+      "step": 4336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9769252154573255e-05,
+      "loss": 0.3166,
+      "step": 4337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9755351681957185e-05,
+      "loss": 0.3171,
+      "step": 4338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.974145120934112e-05,
+      "loss": 0.3056,
+      "step": 4339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.972755073672505e-05,
+      "loss": 0.3149,
+      "step": 4340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.971365026410898e-05,
+      "loss": 0.3008,
+      "step": 4341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.969974979149291e-05,
+      "loss": 0.3219,
+      "step": 4342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.968584931887684e-05,
+      "loss": 0.3306,
+      "step": 4343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.967194884626078e-05,
+      "loss": 0.2818,
+      "step": 4344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.965804837364471e-05,
+      "loss": 0.3729,
+      "step": 4345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.964414790102864e-05,
+      "loss": 0.3121,
+      "step": 4346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.963024742841257e-05,
+      "loss": 0.3321,
+      "step": 4347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.96163469557965e-05,
+      "loss": 0.4049,
+      "step": 4348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.960244648318043e-05,
+      "loss": 0.2938,
+      "step": 4349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.958854601056436e-05,
+      "loss": 0.3646,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.957464553794829e-05,
+      "loss": 0.2968,
+      "step": 4351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.956074506533222e-05,
+      "loss": 0.2983,
+      "step": 4352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9546844592716156e-05,
+      "loss": 0.3314,
+      "step": 4353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9532944120100086e-05,
+      "loss": 0.3131,
+      "step": 4354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9519043647484016e-05,
+      "loss": 0.352,
+      "step": 4355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9505143174867946e-05,
+      "loss": 0.311,
+      "step": 4356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9491242702251876e-05,
+      "loss": 0.3354,
+      "step": 4357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.947734222963581e-05,
+      "loss": 0.3202,
+      "step": 4358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.946344175701974e-05,
+      "loss": 0.3051,
+      "step": 4359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.944954128440367e-05,
+      "loss": 0.3483,
+      "step": 4360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.94356408117876e-05,
+      "loss": 0.345,
+      "step": 4361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.942174033917153e-05,
+      "loss": 0.3317,
+      "step": 4362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.940783986655546e-05,
+      "loss": 0.3596,
+      "step": 4363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.939393939393939e-05,
+      "loss": 0.3835,
+      "step": 4364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9380038921323323e-05,
+      "loss": 0.3295,
+      "step": 4365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.936613844870726e-05,
+      "loss": 0.3325,
+      "step": 4366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.935223797609119e-05,
+      "loss": 0.3706,
+      "step": 4367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.933833750347512e-05,
+      "loss": 0.4029,
+      "step": 4368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.932443703085905e-05,
+      "loss": 0.3393,
+      "step": 4369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.931053655824298e-05,
+      "loss": 0.3833,
+      "step": 4370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.929663608562692e-05,
+      "loss": 0.3482,
+      "step": 4371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.928273561301085e-05,
+      "loss": 0.3503,
+      "step": 4372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.926883514039478e-05,
+      "loss": 0.3682,
+      "step": 4373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.925493466777871e-05,
+      "loss": 0.3276,
+      "step": 4374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.924103419516264e-05,
+      "loss": 0.4172,
+      "step": 4375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.922713372254657e-05,
+      "loss": 0.3185,
+      "step": 4376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.92132332499305e-05,
+      "loss": 0.3626,
+      "step": 4377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.919933277731443e-05,
+      "loss": 0.3757,
+      "step": 4378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.918543230469836e-05,
+      "loss": 0.3126,
+      "step": 4379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9171531832082295e-05,
+      "loss": 0.3937,
+      "step": 4380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9157631359466225e-05,
+      "loss": 0.277,
+      "step": 4381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9143730886850155e-05,
+      "loss": 0.2986,
+      "step": 4382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9129830414234085e-05,
+      "loss": 0.3745,
+      "step": 4383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9115929941618015e-05,
+      "loss": 0.3263,
+      "step": 4384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.910202946900195e-05,
+      "loss": 0.3148,
+      "step": 4385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.908812899638588e-05,
+      "loss": 0.3361,
+      "step": 4386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.907422852376981e-05,
+      "loss": 0.3027,
+      "step": 4387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9060328051153735e-05,
+      "loss": 0.3602,
+      "step": 4388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.904642757853767e-05,
+      "loss": 0.3585,
+      "step": 4389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.90325271059216e-05,
+      "loss": 0.279,
+      "step": 4390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.901862663330553e-05,
+      "loss": 0.3062,
+      "step": 4391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.900472616068946e-05,
+      "loss": 0.348,
+      "step": 4392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.89908256880734e-05,
+      "loss": 0.322,
+      "step": 4393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.897692521545733e-05,
+      "loss": 0.4035,
+      "step": 4394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.896302474284126e-05,
+      "loss": 0.2851,
+      "step": 4395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.894912427022519e-05,
+      "loss": 0.345,
+      "step": 4396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.893522379760912e-05,
+      "loss": 0.417,
+      "step": 4397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8921323324993056e-05,
+      "loss": 0.3683,
+      "step": 4398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8907422852376986e-05,
+      "loss": 0.3727,
+      "step": 4399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8893522379760916e-05,
+      "loss": 0.3386,
+      "step": 4400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8879621907144846e-05,
+      "loss": 0.3484,
+      "step": 4401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8865721434528776e-05,
+      "loss": 0.3122,
+      "step": 4402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8851820961912707e-05,
+      "loss": 0.3502,
+      "step": 4403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8837920489296637e-05,
+      "loss": 0.3025,
+      "step": 4404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.882402001668057e-05,
+      "loss": 0.3425,
+      "step": 4405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.88101195440645e-05,
+      "loss": 0.3853,
+      "step": 4406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8796219071448434e-05,
+      "loss": 0.3912,
+      "step": 4407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8782318598832364e-05,
+      "loss": 0.3716,
+      "step": 4408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8768418126216294e-05,
+      "loss": 0.3477,
+      "step": 4409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8754517653600224e-05,
+      "loss": 0.3746,
+      "step": 4410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8740617180984154e-05,
+      "loss": 0.3719,
+      "step": 4411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.872671670836809e-05,
+      "loss": 0.3622,
+      "step": 4412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.871281623575202e-05,
+      "loss": 0.3581,
+      "step": 4413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.869891576313595e-05,
+      "loss": 0.3918,
+      "step": 4414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8685015290519874e-05,
+      "loss": 0.3624,
+      "step": 4415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.867111481790381e-05,
+      "loss": 0.356,
+      "step": 4416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.865721434528774e-05,
+      "loss": 0.3145,
+      "step": 4417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.864331387267167e-05,
+      "loss": 0.4182,
+      "step": 4418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.86294134000556e-05,
+      "loss": 0.3821,
+      "step": 4419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.861551292743953e-05,
+      "loss": 0.346,
+      "step": 4420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.860161245482347e-05,
+      "loss": 0.3283,
+      "step": 4421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.85877119822074e-05,
+      "loss": 0.3533,
+      "step": 4422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.857381150959133e-05,
+      "loss": 0.3573,
+      "step": 4423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.855991103697526e-05,
+      "loss": 0.3674,
+      "step": 4424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8546010564359195e-05,
+      "loss": 0.3791,
+      "step": 4425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8532110091743125e-05,
+      "loss": 0.3584,
+      "step": 4426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8518209619127055e-05,
+      "loss": 0.3502,
+      "step": 4427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.850430914651098e-05,
+      "loss": 0.3326,
+      "step": 4428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8490408673894915e-05,
+      "loss": 0.3411,
+      "step": 4429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8476508201278845e-05,
+      "loss": 0.321,
+      "step": 4430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8462607728662775e-05,
+      "loss": 0.4052,
+      "step": 4431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8448707256046705e-05,
+      "loss": 0.4019,
+      "step": 4432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8434806783430636e-05,
+      "loss": 0.3616,
+      "step": 4433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.842090631081457e-05,
+      "loss": 0.337,
+      "step": 4434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.84070058381985e-05,
+      "loss": 0.3043,
+      "step": 4435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.839310536558243e-05,
+      "loss": 0.3404,
+      "step": 4436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.837920489296636e-05,
+      "loss": 0.3077,
+      "step": 4437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.836530442035029e-05,
+      "loss": 0.372,
+      "step": 4438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.835140394773423e-05,
+      "loss": 0.3234,
+      "step": 4439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.833750347511816e-05,
+      "loss": 0.3291,
+      "step": 4440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.832360300250208e-05,
+      "loss": 0.3513,
+      "step": 4441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.830970252988601e-05,
+      "loss": 0.3373,
+      "step": 4442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.829580205726995e-05,
+      "loss": 0.3215,
+      "step": 4443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.828190158465388e-05,
+      "loss": 0.3577,
+      "step": 4444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.826800111203781e-05,
+      "loss": 0.3394,
+      "step": 4445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.825410063942174e-05,
+      "loss": 0.3409,
+      "step": 4446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.824020016680567e-05,
+      "loss": 0.3861,
+      "step": 4447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.822629969418961e-05,
+      "loss": 0.3197,
+      "step": 4448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.821239922157354e-05,
+      "loss": 0.309,
+      "step": 4449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.819849874895747e-05,
+      "loss": 0.3635,
+      "step": 4450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.81845982763414e-05,
+      "loss": 0.3647,
+      "step": 4451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.817069780372533e-05,
+      "loss": 0.3117,
+      "step": 4452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8156797331109264e-05,
+      "loss": 0.3945,
+      "step": 4453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.814289685849319e-05,
+      "loss": 0.2952,
+      "step": 4454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.812899638587712e-05,
+      "loss": 0.3892,
+      "step": 4455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8115095913261054e-05,
+      "loss": 0.3361,
+      "step": 4456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8101195440644984e-05,
+      "loss": 0.3704,
+      "step": 4457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8087294968028914e-05,
+      "loss": 0.344,
+      "step": 4458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8073394495412844e-05,
+      "loss": 0.3608,
+      "step": 4459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8059494022796774e-05,
+      "loss": 0.3296,
+      "step": 4460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.804559355018071e-05,
+      "loss": 0.3358,
+      "step": 4461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.803169307756464e-05,
+      "loss": 0.3151,
+      "step": 4462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.801779260494857e-05,
+      "loss": 0.3624,
+      "step": 4463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.80038921323325e-05,
+      "loss": 0.331,
+      "step": 4464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.798999165971643e-05,
+      "loss": 0.2723,
+      "step": 4465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.797609118710037e-05,
+      "loss": 0.349,
+      "step": 4466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.79621907144843e-05,
+      "loss": 0.3234,
+      "step": 4467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.794829024186822e-05,
+      "loss": 0.3447,
+      "step": 4468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.793438976925215e-05,
+      "loss": 0.3387,
+      "step": 4469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.792048929663609e-05,
+      "loss": 0.3386,
+      "step": 4470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.790658882402002e-05,
+      "loss": 0.3313,
+      "step": 4471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.789268835140395e-05,
+      "loss": 0.3476,
+      "step": 4472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.3634,
+      "step": 4473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.786488740617181e-05,
+      "loss": 0.2788,
+      "step": 4474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7850986933555746e-05,
+      "loss": 0.3912,
+      "step": 4475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7837086460939676e-05,
+      "loss": 0.3735,
+      "step": 4476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7823185988323606e-05,
+      "loss": 0.3157,
+      "step": 4477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7809285515707536e-05,
+      "loss": 0.3714,
+      "step": 4478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7795385043091466e-05,
+      "loss": 0.286,
+      "step": 4479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.77814845704754e-05,
+      "loss": 0.3219,
+      "step": 4480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7767584097859326e-05,
+      "loss": 0.3718,
+      "step": 4481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7753683625243256e-05,
+      "loss": 0.351,
+      "step": 4482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7739783152627186e-05,
+      "loss": 0.3341,
+      "step": 4483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.772588268001112e-05,
+      "loss": 0.3717,
+      "step": 4484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.771198220739505e-05,
+      "loss": 0.3566,
+      "step": 4485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.769808173477898e-05,
+      "loss": 0.3311,
+      "step": 4486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.768418126216291e-05,
+      "loss": 0.3927,
+      "step": 4487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.767028078954685e-05,
+      "loss": 0.319,
+      "step": 4488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.765638031693078e-05,
+      "loss": 0.2784,
+      "step": 4489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.764247984431471e-05,
+      "loss": 0.3781,
+      "step": 4490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.762857937169864e-05,
+      "loss": 0.365,
+      "step": 4491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.761467889908257e-05,
+      "loss": 0.3168,
+      "step": 4492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.760077842646651e-05,
+      "loss": 0.2991,
+      "step": 4493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.758687795385043e-05,
+      "loss": 0.3496,
+      "step": 4494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.757297748123436e-05,
+      "loss": 0.3245,
+      "step": 4495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.755907700861829e-05,
+      "loss": 0.3457,
+      "step": 4496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.754517653600223e-05,
+      "loss": 0.3408,
+      "step": 4497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.753127606338616e-05,
+      "loss": 0.3823,
+      "step": 4498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.751737559077009e-05,
+      "loss": 0.3321,
+      "step": 4499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.750347511815402e-05,
+      "loss": 0.2996,
+      "step": 4500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.748957464553795e-05,
+      "loss": 0.2815,
+      "step": 4501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7475674172921884e-05,
+      "loss": 0.3261,
+      "step": 4502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7461773700305815e-05,
+      "loss": 0.3537,
+      "step": 4503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7447873227689745e-05,
+      "loss": 0.3507,
+      "step": 4504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7433972755073675e-05,
+      "loss": 0.3328,
+      "step": 4505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7420072282457605e-05,
+      "loss": 0.3277,
+      "step": 4506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7406171809841535e-05,
+      "loss": 0.333,
+      "step": 4507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7392271337225465e-05,
+      "loss": 0.3375,
+      "step": 4508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7378370864609395e-05,
+      "loss": 0.3096,
+      "step": 4509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7364470391993325e-05,
+      "loss": 0.3121,
+      "step": 4510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.735056991937726e-05,
+      "loss": 0.3744,
+      "step": 4511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.733666944676119e-05,
+      "loss": 0.3618,
+      "step": 4512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.732276897414512e-05,
+      "loss": 0.3731,
+      "step": 4513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.730886850152905e-05,
+      "loss": 0.302,
+      "step": 4514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.729496802891298e-05,
+      "loss": 0.291,
+      "step": 4515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.728106755629692e-05,
+      "loss": 0.321,
+      "step": 4516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.726716708368085e-05,
+      "loss": 0.3744,
+      "step": 4517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.725326661106478e-05,
+      "loss": 0.3756,
+      "step": 4518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.723936613844871e-05,
+      "loss": 0.3426,
+      "step": 4519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7225465665832646e-05,
+      "loss": 0.3339,
+      "step": 4520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.721156519321657e-05,
+      "loss": 0.3366,
+      "step": 4521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.71976647206005e-05,
+      "loss": 0.3114,
+      "step": 4522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.718376424798443e-05,
+      "loss": 0.3453,
+      "step": 4523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7169863775368366e-05,
+      "loss": 0.3002,
+      "step": 4524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7155963302752296e-05,
+      "loss": 0.3116,
+      "step": 4525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7142062830136226e-05,
+      "loss": 0.3582,
+      "step": 4526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7128162357520156e-05,
+      "loss": 0.3412,
+      "step": 4527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7114261884904086e-05,
+      "loss": 0.3377,
+      "step": 4528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.710036141228802e-05,
+      "loss": 0.3867,
+      "step": 4529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.708646093967195e-05,
+      "loss": 0.3421,
+      "step": 4530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7072560467055883e-05,
+      "loss": 0.342,
+      "step": 4531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7058659994439814e-05,
+      "loss": 0.3681,
+      "step": 4532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7044759521823744e-05,
+      "loss": 0.3808,
+      "step": 4533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7030859049207674e-05,
+      "loss": 0.3272,
+      "step": 4534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7016958576591604e-05,
+      "loss": 0.3547,
+      "step": 4535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7003058103975534e-05,
+      "loss": 0.2989,
+      "step": 4536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6989157631359464e-05,
+      "loss": 0.3452,
+      "step": 4537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.69752571587434e-05,
+      "loss": 0.3124,
+      "step": 4538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.696135668612733e-05,
+      "loss": 0.3238,
+      "step": 4539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.694745621351126e-05,
+      "loss": 0.3705,
+      "step": 4540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.693355574089519e-05,
+      "loss": 0.3054,
+      "step": 4541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.691965526827912e-05,
+      "loss": 0.3565,
+      "step": 4542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.690575479566306e-05,
+      "loss": 0.3707,
+      "step": 4543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.689185432304699e-05,
+      "loss": 0.3092,
+      "step": 4544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.687795385043092e-05,
+      "loss": 0.3022,
+      "step": 4545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.686405337781485e-05,
+      "loss": 0.2997,
+      "step": 4546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.685015290519878e-05,
+      "loss": 0.3754,
+      "step": 4547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.683625243258271e-05,
+      "loss": 0.3536,
+      "step": 4548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.682235195996664e-05,
+      "loss": 0.3727,
+      "step": 4549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.680845148735057e-05,
+      "loss": 0.359,
+      "step": 4550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6794551014734505e-05,
+      "loss": 0.3664,
+      "step": 4551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6780650542118435e-05,
+      "loss": 0.3881,
+      "step": 4552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6766750069502365e-05,
+      "loss": 0.3853,
+      "step": 4553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6752849596886295e-05,
+      "loss": 0.3711,
+      "step": 4554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6738949124270225e-05,
+      "loss": 0.3547,
+      "step": 4555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.672504865165416e-05,
+      "loss": 0.3255,
+      "step": 4556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.671114817903809e-05,
+      "loss": 0.356,
+      "step": 4557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.669724770642202e-05,
+      "loss": 0.3396,
+      "step": 4558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.668334723380595e-05,
+      "loss": 0.3559,
+      "step": 4559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.666944676118988e-05,
+      "loss": 0.3348,
+      "step": 4560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.665554628857381e-05,
+      "loss": 0.3021,
+      "step": 4561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.664164581595774e-05,
+      "loss": 0.3544,
+      "step": 4562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.662774534334167e-05,
+      "loss": 0.3264,
+      "step": 4563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.66138448707256e-05,
+      "loss": 0.3713,
+      "step": 4564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.659994439810954e-05,
+      "loss": 0.3425,
+      "step": 4565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.658604392549347e-05,
+      "loss": 0.3413,
+      "step": 4566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.65721434528774e-05,
+      "loss": 0.3197,
+      "step": 4567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.655824298026133e-05,
+      "loss": 0.3784,
+      "step": 4568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.654434250764526e-05,
+      "loss": 0.4031,
+      "step": 4569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6530442035029197e-05,
+      "loss": 0.2849,
+      "step": 4570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.651654156241313e-05,
+      "loss": 0.3532,
+      "step": 4571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.650264108979706e-05,
+      "loss": 0.3552,
+      "step": 4572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.648874061718098e-05,
+      "loss": 0.4451,
+      "step": 4573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.647484014456492e-05,
+      "loss": 0.3377,
+      "step": 4574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.646093967194885e-05,
+      "loss": 0.3499,
+      "step": 4575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.644703919933278e-05,
+      "loss": 0.3679,
+      "step": 4576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.643313872671671e-05,
+      "loss": 0.3132,
+      "step": 4577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.641923825410064e-05,
+      "loss": 0.3254,
+      "step": 4578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6405337781484574e-05,
+      "loss": 0.2971,
+      "step": 4579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6391437308868504e-05,
+      "loss": 0.3283,
+      "step": 4580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6377536836252434e-05,
+      "loss": 0.3201,
+      "step": 4581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.3416,
+      "step": 4582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.63497358910203e-05,
+      "loss": 0.3286,
+      "step": 4583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.633583541840423e-05,
+      "loss": 0.3383,
+      "step": 4584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.632193494578816e-05,
+      "loss": 0.3079,
+      "step": 4585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.630803447317209e-05,
+      "loss": 0.3213,
+      "step": 4586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.629413400055602e-05,
+      "loss": 0.3338,
+      "step": 4587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.628023352793995e-05,
+      "loss": 0.3382,
+      "step": 4588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.626633305532388e-05,
+      "loss": 0.3642,
+      "step": 4589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.625243258270781e-05,
+      "loss": 0.3246,
+      "step": 4590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.623853211009174e-05,
+      "loss": 0.3136,
+      "step": 4591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.622463163747568e-05,
+      "loss": 0.3878,
+      "step": 4592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.621073116485961e-05,
+      "loss": 0.3354,
+      "step": 4593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.619683069224354e-05,
+      "loss": 0.3284,
+      "step": 4594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.618293021962747e-05,
+      "loss": 0.3091,
+      "step": 4595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.61690297470114e-05,
+      "loss": 0.3322,
+      "step": 4596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6155129274395335e-05,
+      "loss": 0.338,
+      "step": 4597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6141228801779265e-05,
+      "loss": 0.2992,
+      "step": 4598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6127328329163196e-05,
+      "loss": 0.346,
+      "step": 4599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.611342785654712e-05,
+      "loss": 0.331,
+      "step": 4600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6099527383931056e-05,
+      "loss": 0.3319,
+      "step": 4601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6085626911314986e-05,
+      "loss": 0.3582,
+      "step": 4602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6071726438698916e-05,
+      "loss": 0.3434,
+      "step": 4603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6057825966082846e-05,
+      "loss": 0.3054,
+      "step": 4604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6043925493466776e-05,
+      "loss": 0.2907,
+      "step": 4605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.603002502085071e-05,
+      "loss": 0.3113,
+      "step": 4606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.601612454823464e-05,
+      "loss": 0.3743,
+      "step": 4607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.600222407561857e-05,
+      "loss": 0.3634,
+      "step": 4608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.59883236030025e-05,
+      "loss": 0.2932,
+      "step": 4609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.597442313038644e-05,
+      "loss": 0.3711,
+      "step": 4610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.596052265777037e-05,
+      "loss": 0.2869,
+      "step": 4611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.59466221851543e-05,
+      "loss": 0.2972,
+      "step": 4612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.593272171253822e-05,
+      "loss": 0.302,
+      "step": 4613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.591882123992216e-05,
+      "loss": 0.3695,
+      "step": 4614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.590492076730609e-05,
+      "loss": 0.3854,
+      "step": 4615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.589102029469002e-05,
+      "loss": 0.309,
+      "step": 4616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.587711982207395e-05,
+      "loss": 0.326,
+      "step": 4617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.586321934945788e-05,
+      "loss": 0.3003,
+      "step": 4618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.584931887684182e-05,
+      "loss": 0.3322,
+      "step": 4619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.583541840422575e-05,
+      "loss": 0.3605,
+      "step": 4620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.582151793160968e-05,
+      "loss": 0.3121,
+      "step": 4621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.580761745899361e-05,
+      "loss": 0.3156,
+      "step": 4622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.579371698637754e-05,
+      "loss": 0.3738,
+      "step": 4623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5779816513761474e-05,
+      "loss": 0.3588,
+      "step": 4624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5765916041145404e-05,
+      "loss": 0.3886,
+      "step": 4625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.575201556852933e-05,
+      "loss": 0.3892,
+      "step": 4626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.573811509591326e-05,
+      "loss": 0.3491,
+      "step": 4627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5724214623297195e-05,
+      "loss": 0.3837,
+      "step": 4628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5710314150681125e-05,
+      "loss": 0.3344,
+      "step": 4629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5696413678065055e-05,
+      "loss": 0.3428,
+      "step": 4630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5682513205448985e-05,
+      "loss": 0.3419,
+      "step": 4631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5668612732832915e-05,
+      "loss": 0.3802,
+      "step": 4632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.565471226021685e-05,
+      "loss": 0.3534,
+      "step": 4633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.564081178760078e-05,
+      "loss": 0.3897,
+      "step": 4634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.562691131498471e-05,
+      "loss": 0.2826,
+      "step": 4635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.561301084236864e-05,
+      "loss": 0.3173,
+      "step": 4636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.559911036975257e-05,
+      "loss": 0.284,
+      "step": 4637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.558520989713651e-05,
+      "loss": 0.353,
+      "step": 4638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.557130942452044e-05,
+      "loss": 0.3909,
+      "step": 4639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.555740895190436e-05,
+      "loss": 0.3545,
+      "step": 4640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.55435084792883e-05,
+      "loss": 0.3234,
+      "step": 4641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.552960800667223e-05,
+      "loss": 0.3275,
+      "step": 4642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.551570753405616e-05,
+      "loss": 0.3437,
+      "step": 4643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.550180706144009e-05,
+      "loss": 0.3489,
+      "step": 4644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.548790658882402e-05,
+      "loss": 0.3364,
+      "step": 4645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5474006116207956e-05,
+      "loss": 0.3377,
+      "step": 4646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5460105643591886e-05,
+      "loss": 0.3367,
+      "step": 4647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5446205170975816e-05,
+      "loss": 0.3258,
+      "step": 4648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5432304698359746e-05,
+      "loss": 0.2945,
+      "step": 4649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5418404225743676e-05,
+      "loss": 0.3742,
+      "step": 4650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.540450375312761e-05,
+      "loss": 0.3792,
+      "step": 4651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.539060328051154e-05,
+      "loss": 0.3459,
+      "step": 4652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5376702807895466e-05,
+      "loss": 0.2673,
+      "step": 4653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5362802335279397e-05,
+      "loss": 0.3319,
+      "step": 4654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.534890186266333e-05,
+      "loss": 0.3019,
+      "step": 4655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5335001390047263e-05,
+      "loss": 0.3483,
+      "step": 4656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5321100917431193e-05,
+      "loss": 0.3474,
+      "step": 4657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5307200444815124e-05,
+      "loss": 0.3461,
+      "step": 4658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5293299972199054e-05,
+      "loss": 0.3176,
+      "step": 4659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.527939949958299e-05,
+      "loss": 0.3244,
+      "step": 4660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.526549902696692e-05,
+      "loss": 0.3272,
+      "step": 4661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.525159855435085e-05,
+      "loss": 0.2822,
+      "step": 4662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.523769808173478e-05,
+      "loss": 0.3681,
+      "step": 4663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.522379760911871e-05,
+      "loss": 0.3546,
+      "step": 4664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.520989713650265e-05,
+      "loss": 0.3821,
+      "step": 4665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.519599666388657e-05,
+      "loss": 0.3755,
+      "step": 4666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.51820961912705e-05,
+      "loss": 0.3296,
+      "step": 4667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.516819571865443e-05,
+      "loss": 0.3786,
+      "step": 4668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.515429524603837e-05,
+      "loss": 0.332,
+      "step": 4669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.51403947734223e-05,
+      "loss": 0.3493,
+      "step": 4670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.512649430080623e-05,
+      "loss": 0.3128,
+      "step": 4671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.511259382819016e-05,
+      "loss": 0.3431,
+      "step": 4672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5098693355574095e-05,
+      "loss": 0.3346,
+      "step": 4673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5084792882958025e-05,
+      "loss": 0.3962,
+      "step": 4674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5070892410341955e-05,
+      "loss": 0.3634,
+      "step": 4675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5056991937725885e-05,
+      "loss": 0.4175,
+      "step": 4676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5043091465109815e-05,
+      "loss": 0.2863,
+      "step": 4677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.502919099249375e-05,
+      "loss": 0.2914,
+      "step": 4678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5015290519877675e-05,
+      "loss": 0.3498,
+      "step": 4679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5001390047261605e-05,
+      "loss": 0.3622,
+      "step": 4680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4987489574645535e-05,
+      "loss": 0.3281,
+      "step": 4681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.497358910202947e-05,
+      "loss": 0.3615,
+      "step": 4682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.49596886294134e-05,
+      "loss": 0.336,
+      "step": 4683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.494578815679733e-05,
+      "loss": 0.3654,
+      "step": 4684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.493188768418126e-05,
+      "loss": 0.3365,
+      "step": 4685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.491798721156519e-05,
+      "loss": 0.2903,
+      "step": 4686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.490408673894913e-05,
+      "loss": 0.3423,
+      "step": 4687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.489018626633306e-05,
+      "loss": 0.3394,
+      "step": 4688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.487628579371699e-05,
+      "loss": 0.3803,
+      "step": 4689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.486238532110092e-05,
+      "loss": 0.3242,
+      "step": 4690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 0.3288,
+      "step": 4691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4834584375868786e-05,
+      "loss": 0.3032,
+      "step": 4692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.482068390325271e-05,
+      "loss": 0.3745,
+      "step": 4693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.480678343063664e-05,
+      "loss": 0.3292,
+      "step": 4694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.479288295802057e-05,
+      "loss": 0.3571,
+      "step": 4695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.477898248540451e-05,
+      "loss": 0.3362,
+      "step": 4696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.476508201278844e-05,
+      "loss": 0.3206,
+      "step": 4697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.475118154017237e-05,
+      "loss": 0.3394,
+      "step": 4698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.47372810675563e-05,
+      "loss": 0.3194,
+      "step": 4699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.472338059494023e-05,
+      "loss": 0.3093,
+      "step": 4700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4709480122324164e-05,
+      "loss": 0.3504,
+      "step": 4701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4695579649708094e-05,
+      "loss": 0.3679,
+      "step": 4702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4681679177092024e-05,
+      "loss": 0.3245,
+      "step": 4703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4667778704475954e-05,
+      "loss": 0.2857,
+      "step": 4704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.465387823185989e-05,
+      "loss": 0.297,
+      "step": 4705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4639977759243814e-05,
+      "loss": 0.3565,
+      "step": 4706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4626077286627744e-05,
+      "loss": 0.3089,
+      "step": 4707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4612176814011674e-05,
+      "loss": 0.3281,
+      "step": 4708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.459827634139561e-05,
+      "loss": 0.3504,
+      "step": 4709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.458437586877954e-05,
+      "loss": 0.3278,
+      "step": 4710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.457047539616347e-05,
+      "loss": 0.3376,
+      "step": 4711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.45565749235474e-05,
+      "loss": 0.3362,
+      "step": 4712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.454267445093133e-05,
+      "loss": 0.3174,
+      "step": 4713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.452877397831527e-05,
+      "loss": 0.3465,
+      "step": 4714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.45148735056992e-05,
+      "loss": 0.3676,
+      "step": 4715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.450097303308313e-05,
+      "loss": 0.391,
+      "step": 4716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.448707256046706e-05,
+      "loss": 0.4401,
+      "step": 4717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.447317208785099e-05,
+      "loss": 0.3111,
+      "step": 4718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.445927161523492e-05,
+      "loss": 0.3137,
+      "step": 4719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.444537114261885e-05,
+      "loss": 0.3265,
+      "step": 4720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.443147067000278e-05,
+      "loss": 0.3102,
+      "step": 4721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.441757019738671e-05,
+      "loss": 0.3347,
+      "step": 4722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4403669724770645e-05,
+      "loss": 0.3693,
+      "step": 4723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4389769252154576e-05,
+      "loss": 0.3248,
+      "step": 4724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4375868779538506e-05,
+      "loss": 0.288,
+      "step": 4725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4361968306922436e-05,
+      "loss": 0.3596,
+      "step": 4726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4348067834306366e-05,
+      "loss": 0.3492,
+      "step": 4727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.43341673616903e-05,
+      "loss": 0.3315,
+      "step": 4728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.432026688907423e-05,
+      "loss": 0.3542,
+      "step": 4729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.430636641645816e-05,
+      "loss": 0.2751,
+      "step": 4730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.429246594384209e-05,
+      "loss": 0.3554,
+      "step": 4731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.427856547122602e-05,
+      "loss": 0.2764,
+      "step": 4732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.426466499860995e-05,
+      "loss": 0.3111,
+      "step": 4733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.425076452599388e-05,
+      "loss": 0.4048,
+      "step": 4734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.423686405337781e-05,
+      "loss": 0.3312,
+      "step": 4735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.422296358076175e-05,
+      "loss": 0.3137,
+      "step": 4736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.420906310814568e-05,
+      "loss": 0.299,
+      "step": 4737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.419516263552961e-05,
+      "loss": 0.3089,
+      "step": 4738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.418126216291354e-05,
+      "loss": 0.2993,
+      "step": 4739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.416736169029747e-05,
+      "loss": 0.2891,
+      "step": 4740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.415346121768141e-05,
+      "loss": 0.3299,
+      "step": 4741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.413956074506534e-05,
+      "loss": 0.3798,
+      "step": 4742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.412566027244927e-05,
+      "loss": 0.3618,
+      "step": 4743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.41117597998332e-05,
+      "loss": 0.3437,
+      "step": 4744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.409785932721713e-05,
+      "loss": 0.3421,
+      "step": 4745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.408395885460106e-05,
+      "loss": 0.3587,
+      "step": 4746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.407005838198499e-05,
+      "loss": 0.3247,
+      "step": 4747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.405615790936892e-05,
+      "loss": 0.3066,
+      "step": 4748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.404225743675285e-05,
+      "loss": 0.3102,
+      "step": 4749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4028356964136784e-05,
+      "loss": 0.3116,
+      "step": 4750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4014456491520714e-05,
+      "loss": 0.3168,
+      "step": 4751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4000556018904644e-05,
+      "loss": 0.3334,
+      "step": 4752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3986655546288574e-05,
+      "loss": 0.3557,
+      "step": 4753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3972755073672505e-05,
+      "loss": 0.3471,
+      "step": 4754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.395885460105644e-05,
+      "loss": 0.3197,
+      "step": 4755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.394495412844037e-05,
+      "loss": 0.3153,
+      "step": 4756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.39310536558243e-05,
+      "loss": 0.3294,
+      "step": 4757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.391715318320823e-05,
+      "loss": 0.2489,
+      "step": 4758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.390325271059216e-05,
+      "loss": 0.3289,
+      "step": 4759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.388935223797609e-05,
+      "loss": 0.3645,
+      "step": 4760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.387545176536002e-05,
+      "loss": 0.3399,
+      "step": 4761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.386155129274395e-05,
+      "loss": 0.3065,
+      "step": 4762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.384765082012788e-05,
+      "loss": 0.382,
+      "step": 4763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.383375034751182e-05,
+      "loss": 0.3601,
+      "step": 4764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.381984987489575e-05,
+      "loss": 0.3414,
+      "step": 4765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.380594940227968e-05,
+      "loss": 0.335,
+      "step": 4766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.379204892966361e-05,
+      "loss": 0.3644,
+      "step": 4767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3778148457047546e-05,
+      "loss": 0.3406,
+      "step": 4768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3764247984431476e-05,
+      "loss": 0.2835,
+      "step": 4769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3750347511815406e-05,
+      "loss": 0.2936,
+      "step": 4770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3736447039199336e-05,
+      "loss": 0.3412,
+      "step": 4771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3722546566583266e-05,
+      "loss": 0.3092,
+      "step": 4772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3708646093967196e-05,
+      "loss": 0.3197,
+      "step": 4773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3694745621351126e-05,
+      "loss": 0.3192,
+      "step": 4774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3680845148735056e-05,
+      "loss": 0.3369,
+      "step": 4775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3666944676118986e-05,
+      "loss": 0.3242,
+      "step": 4776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.365304420350292e-05,
+      "loss": 0.3147,
+      "step": 4777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.363914373088685e-05,
+      "loss": 0.4072,
+      "step": 4778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.362524325827078e-05,
+      "loss": 0.3641,
+      "step": 4779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.361134278565471e-05,
+      "loss": 0.3556,
+      "step": 4780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.359744231303864e-05,
+      "loss": 0.3011,
+      "step": 4781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.358354184042258e-05,
+      "loss": 0.2915,
+      "step": 4782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.356964136780651e-05,
+      "loss": 0.2986,
+      "step": 4783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.355574089519044e-05,
+      "loss": 0.366,
+      "step": 4784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3541840422574364e-05,
+      "loss": 0.2905,
+      "step": 4785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.35279399499583e-05,
+      "loss": 0.3713,
+      "step": 4786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.351403947734223e-05,
+      "loss": 0.3268,
+      "step": 4787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.350013900472616e-05,
+      "loss": 0.3524,
+      "step": 4788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.348623853211009e-05,
+      "loss": 0.3341,
+      "step": 4789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.347233805949402e-05,
+      "loss": 0.3202,
+      "step": 4790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.345843758687796e-05,
+      "loss": 0.3625,
+      "step": 4791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.344453711426189e-05,
+      "loss": 0.2477,
+      "step": 4792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.343063664164582e-05,
+      "loss": 0.305,
+      "step": 4793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.341673616902975e-05,
+      "loss": 0.3802,
+      "step": 4794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3402835696413685e-05,
+      "loss": 0.3029,
+      "step": 4795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3388935223797615e-05,
+      "loss": 0.293,
+      "step": 4796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3375034751181545e-05,
+      "loss": 0.3034,
+      "step": 4797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.336113427856547e-05,
+      "loss": 0.2888,
+      "step": 4798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3347233805949405e-05,
+      "loss": 0.2974,
+      "step": 4799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.3299,
+      "step": 4800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3319432860717265e-05,
+      "loss": 0.3135,
+      "step": 4801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3305532388101195e-05,
+      "loss": 0.3793,
+      "step": 4802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3291631915485125e-05,
+      "loss": 0.2793,
+      "step": 4803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.327773144286906e-05,
+      "loss": 0.2991,
+      "step": 4804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.326383097025299e-05,
+      "loss": 0.3064,
+      "step": 4805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.324993049763692e-05,
+      "loss": 0.3137,
+      "step": 4806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.323603002502085e-05,
+      "loss": 0.3054,
+      "step": 4807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.322212955240478e-05,
+      "loss": 0.3585,
+      "step": 4808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.320822907978872e-05,
+      "loss": 0.3816,
+      "step": 4809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.319432860717265e-05,
+      "loss": 0.2927,
+      "step": 4810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.318042813455658e-05,
+      "loss": 0.3148,
+      "step": 4811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.31665276619405e-05,
+      "loss": 0.3524,
+      "step": 4812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.315262718932444e-05,
+      "loss": 0.3646,
+      "step": 4813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.313872671670837e-05,
+      "loss": 0.358,
+      "step": 4814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.31248262440923e-05,
+      "loss": 0.3413,
+      "step": 4815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.311092577147623e-05,
+      "loss": 0.3612,
+      "step": 4816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.309702529886016e-05,
+      "loss": 0.3337,
+      "step": 4817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3083124826244096e-05,
+      "loss": 0.2854,
+      "step": 4818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3069224353628026e-05,
+      "loss": 0.3671,
+      "step": 4819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3055323881011957e-05,
+      "loss": 0.3081,
+      "step": 4820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3041423408395887e-05,
+      "loss": 0.3213,
+      "step": 4821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.302752293577982e-05,
+      "loss": 0.3235,
+      "step": 4822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3013622463163754e-05,
+      "loss": 0.2732,
+      "step": 4823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2999721990547684e-05,
+      "loss": 0.3209,
+      "step": 4824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.298582151793161e-05,
+      "loss": 0.3418,
+      "step": 4825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.297192104531554e-05,
+      "loss": 0.3489,
+      "step": 4826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2958020572699474e-05,
+      "loss": 0.3001,
+      "step": 4827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2944120100083404e-05,
+      "loss": 0.3651,
+      "step": 4828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2930219627467334e-05,
+      "loss": 0.284,
+      "step": 4829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2916319154851264e-05,
+      "loss": 0.2957,
+      "step": 4830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.29024186822352e-05,
+      "loss": 0.2848,
+      "step": 4831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.288851820961913e-05,
+      "loss": 0.3085,
+      "step": 4832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.287461773700306e-05,
+      "loss": 0.2768,
+      "step": 4833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.286071726438699e-05,
+      "loss": 0.3517,
+      "step": 4834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.284681679177092e-05,
+      "loss": 0.4154,
+      "step": 4835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.283291631915486e-05,
+      "loss": 0.358,
+      "step": 4836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.281901584653879e-05,
+      "loss": 0.299,
+      "step": 4837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.280511537392271e-05,
+      "loss": 0.3109,
+      "step": 4838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.279121490130664e-05,
+      "loss": 0.3154,
+      "step": 4839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.277731442869058e-05,
+      "loss": 0.3704,
+      "step": 4840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.276341395607451e-05,
+      "loss": 0.2931,
+      "step": 4841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.274951348345844e-05,
+      "loss": 0.2804,
+      "step": 4842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.273561301084237e-05,
+      "loss": 0.346,
+      "step": 4843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.27217125382263e-05,
+      "loss": 0.2881,
+      "step": 4844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2707812065610235e-05,
+      "loss": 0.317,
+      "step": 4845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2693911592994165e-05,
+      "loss": 0.2742,
+      "step": 4846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2680011120378095e-05,
+      "loss": 0.3448,
+      "step": 4847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2666110647762025e-05,
+      "loss": 0.3514,
+      "step": 4848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2652210175145955e-05,
+      "loss": 0.3603,
+      "step": 4849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.263830970252989e-05,
+      "loss": 0.3476,
+      "step": 4850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2624409229913816e-05,
+      "loss": 0.4529,
+      "step": 4851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2610508757297746e-05,
+      "loss": 0.3783,
+      "step": 4852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2596608284681676e-05,
+      "loss": 0.307,
+      "step": 4853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.258270781206561e-05,
+      "loss": 0.3207,
+      "step": 4854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.256880733944954e-05,
+      "loss": 0.3454,
+      "step": 4855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.255490686683347e-05,
+      "loss": 0.2822,
+      "step": 4856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.25410063942174e-05,
+      "loss": 0.3728,
+      "step": 4857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.252710592160134e-05,
+      "loss": 0.3354,
+      "step": 4858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.251320544898527e-05,
+      "loss": 0.3275,
+      "step": 4859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.24993049763692e-05,
+      "loss": 0.3803,
+      "step": 4860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.248540450375313e-05,
+      "loss": 0.3143,
+      "step": 4861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.247150403113706e-05,
+      "loss": 0.3225,
+      "step": 4862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2457603558521e-05,
+      "loss": 0.3498,
+      "step": 4863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.244370308590492e-05,
+      "loss": 0.2791,
+      "step": 4864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.242980261328885e-05,
+      "loss": 0.3706,
+      "step": 4865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.241590214067278e-05,
+      "loss": 0.416,
+      "step": 4866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.240200166805672e-05,
+      "loss": 0.2947,
+      "step": 4867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.238810119544065e-05,
+      "loss": 0.3241,
+      "step": 4868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.237420072282458e-05,
+      "loss": 0.318,
+      "step": 4869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.236030025020851e-05,
+      "loss": 0.2984,
+      "step": 4870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.234639977759244e-05,
+      "loss": 0.3213,
+      "step": 4871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2332499304976374e-05,
+      "loss": 0.356,
+      "step": 4872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2318598832360304e-05,
+      "loss": 0.3712,
+      "step": 4873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2304698359744234e-05,
+      "loss": 0.3644,
+      "step": 4874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2290797887128164e-05,
+      "loss": 0.3523,
+      "step": 4875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2276897414512094e-05,
+      "loss": 0.3323,
+      "step": 4876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.226299694189603e-05,
+      "loss": 0.3916,
+      "step": 4877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2249096469279954e-05,
+      "loss": 0.3324,
+      "step": 4878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2235195996663885e-05,
+      "loss": 0.3395,
+      "step": 4879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2221295524047815e-05,
+      "loss": 0.3365,
+      "step": 4880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.220739505143175e-05,
+      "loss": 0.318,
+      "step": 4881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.219349457881568e-05,
+      "loss": 0.3427,
+      "step": 4882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.217959410619961e-05,
+      "loss": 0.3396,
+      "step": 4883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.216569363358354e-05,
+      "loss": 0.3285,
+      "step": 4884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.215179316096747e-05,
+      "loss": 0.3682,
+      "step": 4885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.213789268835141e-05,
+      "loss": 0.3195,
+      "step": 4886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.212399221573534e-05,
+      "loss": 0.3421,
+      "step": 4887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.211009174311927e-05,
+      "loss": 0.3644,
+      "step": 4888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.20961912705032e-05,
+      "loss": 0.3504,
+      "step": 4889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2082290797887136e-05,
+      "loss": 0.2853,
+      "step": 4890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.206839032527106e-05,
+      "loss": 0.3347,
+      "step": 4891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.205448985265499e-05,
+      "loss": 0.3273,
+      "step": 4892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.204058938003892e-05,
+      "loss": 0.3829,
+      "step": 4893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2026688907422856e-05,
+      "loss": 0.3188,
+      "step": 4894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2012788434806786e-05,
+      "loss": 0.3881,
+      "step": 4895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1998887962190716e-05,
+      "loss": 0.3319,
+      "step": 4896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1984987489574646e-05,
+      "loss": 0.3626,
+      "step": 4897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1971087016958576e-05,
+      "loss": 0.2963,
+      "step": 4898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.195718654434251e-05,
+      "loss": 0.3449,
+      "step": 4899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.194328607172644e-05,
+      "loss": 0.3316,
+      "step": 4900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.192938559911037e-05,
+      "loss": 0.2969,
+      "step": 4901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.19154851264943e-05,
+      "loss": 0.3405,
+      "step": 4902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.190158465387823e-05,
+      "loss": 0.3642,
+      "step": 4903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.188768418126216e-05,
+      "loss": 0.3077,
+      "step": 4904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.187378370864609e-05,
+      "loss": 0.3245,
+      "step": 4905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.185988323603002e-05,
+      "loss": 0.3149,
+      "step": 4906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1845982763413953e-05,
+      "loss": 0.3587,
+      "step": 4907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.183208229079789e-05,
+      "loss": 0.2933,
+      "step": 4908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.3109,
+      "step": 4909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.180428134556575e-05,
+      "loss": 0.4069,
+      "step": 4910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.179038087294968e-05,
+      "loss": 0.319,
+      "step": 4911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.177648040033361e-05,
+      "loss": 0.3547,
+      "step": 4912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.176257992771755e-05,
+      "loss": 0.3071,
+      "step": 4913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.174867945510148e-05,
+      "loss": 0.3337,
+      "step": 4914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.173477898248541e-05,
+      "loss": 0.3354,
+      "step": 4915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.172087850986934e-05,
+      "loss": 0.3557,
+      "step": 4916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.170697803725327e-05,
+      "loss": 0.3275,
+      "step": 4917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.16930775646372e-05,
+      "loss": 0.3283,
+      "step": 4918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.167917709202113e-05,
+      "loss": 0.3516,
+      "step": 4919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.166527661940506e-05,
+      "loss": 0.2738,
+      "step": 4920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1651376146788995e-05,
+      "loss": 0.3472,
+      "step": 4921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1637475674172925e-05,
+      "loss": 0.3745,
+      "step": 4922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1623575201556855e-05,
+      "loss": 0.3597,
+      "step": 4923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1609674728940785e-05,
+      "loss": 0.3358,
+      "step": 4924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.1595774256324715e-05,
+      "loss": 0.3575,
+      "step": 4925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.158187378370865e-05,
+      "loss": 0.3517,
+      "step": 4926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.156797331109258e-05,
+      "loss": 0.3428,
+      "step": 4927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.155407283847651e-05,
+      "loss": 0.3262,
+      "step": 4928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.154017236586044e-05,
+      "loss": 0.3495,
+      "step": 4929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.152627189324437e-05,
+      "loss": 0.2773,
+      "step": 4930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.15123714206283e-05,
+      "loss": 0.3378,
+      "step": 4931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.149847094801223e-05,
+      "loss": 0.3181,
+      "step": 4932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.148457047539616e-05,
+      "loss": 0.2504,
+      "step": 4933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.147067000278009e-05,
+      "loss": 0.3405,
+      "step": 4934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.145676953016403e-05,
+      "loss": 0.3065,
+      "step": 4935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.144286905754796e-05,
+      "loss": 0.3166,
+      "step": 4936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.142896858493189e-05,
+      "loss": 0.3542,
+      "step": 4937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.141506811231582e-05,
+      "loss": 0.3586,
+      "step": 4938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.140116763969975e-05,
+      "loss": 0.3232,
+      "step": 4939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1387267167083686e-05,
+      "loss": 0.3551,
+      "step": 4940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1373366694467616e-05,
+      "loss": 0.3242,
+      "step": 4941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1359466221851546e-05,
+      "loss": 0.395,
+      "step": 4942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1345565749235476e-05,
+      "loss": 0.3197,
+      "step": 4943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1331665276619406e-05,
+      "loss": 0.3453,
+      "step": 4944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1317764804003337e-05,
+      "loss": 0.3165,
+      "step": 4945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1303864331387267e-05,
+      "loss": 0.3219,
+      "step": 4946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.12899638587712e-05,
+      "loss": 0.2854,
+      "step": 4947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.127606338615513e-05,
+      "loss": 0.3647,
+      "step": 4948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1262162913539064e-05,
+      "loss": 0.3503,
+      "step": 4949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1248262440922994e-05,
+      "loss": 0.2912,
+      "step": 4950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1234361968306924e-05,
+      "loss": 0.3504,
+      "step": 4951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1220461495690854e-05,
+      "loss": 0.3262,
+      "step": 4952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.120656102307479e-05,
+      "loss": 0.2967,
+      "step": 4953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.119266055045872e-05,
+      "loss": 0.32,
+      "step": 4954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.117876007784265e-05,
+      "loss": 0.369,
+      "step": 4955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.116485960522658e-05,
+      "loss": 0.3666,
+      "step": 4956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.115095913261051e-05,
+      "loss": 0.3114,
+      "step": 4957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.113705865999444e-05,
+      "loss": 0.2779,
+      "step": 4958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.112315818737837e-05,
+      "loss": 0.3379,
+      "step": 4959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.11092577147623e-05,
+      "loss": 0.3066,
+      "step": 4960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.109535724214623e-05,
+      "loss": 0.3245,
+      "step": 4961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.108145676953017e-05,
+      "loss": 0.3068,
+      "step": 4962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.10675562969141e-05,
+      "loss": 0.3316,
+      "step": 4963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.105365582429803e-05,
+      "loss": 0.3286,
+      "step": 4964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.103975535168196e-05,
+      "loss": 0.3481,
+      "step": 4965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.102585487906589e-05,
+      "loss": 0.331,
+      "step": 4966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1011954406449825e-05,
+      "loss": 0.3399,
+      "step": 4967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0998053933833755e-05,
+      "loss": 0.2853,
+      "step": 4968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0984153461217685e-05,
+      "loss": 0.3727,
+      "step": 4969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.097025298860161e-05,
+      "loss": 0.3056,
+      "step": 4970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0956352515985545e-05,
+      "loss": 0.3554,
+      "step": 4971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0942452043369475e-05,
+      "loss": 0.3309,
+      "step": 4972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0928551570753405e-05,
+      "loss": 0.3213,
+      "step": 4973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0914651098137335e-05,
+      "loss": 0.346,
+      "step": 4974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0900750625521266e-05,
+      "loss": 0.3301,
+      "step": 4975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.08868501529052e-05,
+      "loss": 0.2889,
+      "step": 4976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.087294968028913e-05,
+      "loss": 0.3276,
+      "step": 4977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.085904920767306e-05,
+      "loss": 0.3079,
+      "step": 4978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.084514873505699e-05,
+      "loss": 0.3243,
+      "step": 4979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.083124826244092e-05,
+      "loss": 0.3316,
+      "step": 4980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.081734778982486e-05,
+      "loss": 0.3341,
+      "step": 4981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.080344731720879e-05,
+      "loss": 0.3399,
+      "step": 4982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.078954684459271e-05,
+      "loss": 0.3005,
+      "step": 4983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.077564637197665e-05,
+      "loss": 0.3321,
+      "step": 4984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.076174589936058e-05,
+      "loss": 0.3047,
+      "step": 4985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.074784542674451e-05,
+      "loss": 0.3156,
+      "step": 4986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.073394495412844e-05,
+      "loss": 0.3435,
+      "step": 4987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.072004448151237e-05,
+      "loss": 0.3493,
+      "step": 4988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.070614400889631e-05,
+      "loss": 0.309,
+      "step": 4989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.069224353628024e-05,
+      "loss": 0.3511,
+      "step": 4990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.067834306366417e-05,
+      "loss": 0.3375,
+      "step": 4991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.06644425910481e-05,
+      "loss": 0.364,
+      "step": 4992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.065054211843203e-05,
+      "loss": 0.3577,
+      "step": 4993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0636641645815964e-05,
+      "loss": 0.2966,
+      "step": 4994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0622741173199894e-05,
+      "loss": 0.3109,
+      "step": 4995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0608840700583824e-05,
+      "loss": 0.2931,
+      "step": 4996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.059494022796775e-05,
+      "loss": 0.3056,
+      "step": 4997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0581039755351684e-05,
+      "loss": 0.3611,
+      "step": 4998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0567139282735614e-05,
+      "loss": 0.3406,
+      "step": 4999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0553238810119544e-05,
+      "loss": 0.2867,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0539338337503474e-05,
+      "loss": 0.3209,
+      "step": 5001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0525437864887404e-05,
+      "loss": 0.3653,
+      "step": 5002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0511537392271338e-05,
+      "loss": 0.3677,
+      "step": 5003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.049763691965527e-05,
+      "loss": 0.3188,
+      "step": 5004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.04837364470392e-05,
+      "loss": 0.2854,
+      "step": 5005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0469835974423135e-05,
+      "loss": 0.3878,
+      "step": 5006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0455935501807065e-05,
+      "loss": 0.2996,
+      "step": 5007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0442035029190995e-05,
+      "loss": 0.3147,
+      "step": 5008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.042813455657493e-05,
+      "loss": 0.3549,
+      "step": 5009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0414234083958855e-05,
+      "loss": 0.3192,
+      "step": 5010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0400333611342785e-05,
+      "loss": 0.335,
+      "step": 5011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0386433138726715e-05,
+      "loss": 0.3377,
+      "step": 5012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.037253266611065e-05,
+      "loss": 0.3369,
+      "step": 5013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.035863219349458e-05,
+      "loss": 0.2708,
+      "step": 5014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0344731720878512e-05,
+      "loss": 0.3303,
+      "step": 5015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0330831248262442e-05,
+      "loss": 0.2902,
+      "step": 5016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0316930775646372e-05,
+      "loss": 0.3518,
+      "step": 5017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.3165,
+      "step": 5018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0289129830414236e-05,
+      "loss": 0.3458,
+      "step": 5019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.027522935779817e-05,
+      "loss": 0.3202,
+      "step": 5020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.02613288851821e-05,
+      "loss": 0.3181,
+      "step": 5021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0247428412566033e-05,
+      "loss": 0.3276,
+      "step": 5022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0233527939949956e-05,
+      "loss": 0.3596,
+      "step": 5023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.021962746733389e-05,
+      "loss": 0.2998,
+      "step": 5024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.020572699471782e-05,
+      "loss": 0.3196,
+      "step": 5025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0191826522101753e-05,
+      "loss": 0.3143,
+      "step": 5026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0177926049485683e-05,
+      "loss": 0.3116,
+      "step": 5027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0164025576869613e-05,
+      "loss": 0.3302,
+      "step": 5028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0150125104253547e-05,
+      "loss": 0.3802,
+      "step": 5029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0136224631637477e-05,
+      "loss": 0.3862,
+      "step": 5030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.012232415902141e-05,
+      "loss": 0.3358,
+      "step": 5031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.010842368640534e-05,
+      "loss": 0.3186,
+      "step": 5032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.009452321378927e-05,
+      "loss": 0.356,
+      "step": 5033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0080622741173204e-05,
+      "loss": 0.3612,
+      "step": 5034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0066722268557134e-05,
+      "loss": 0.2807,
+      "step": 5035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.005282179594106e-05,
+      "loss": 0.304,
+      "step": 5036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0038921323324994e-05,
+      "loss": 0.3646,
+      "step": 5037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0025020850708924e-05,
+      "loss": 0.3625,
+      "step": 5038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0011120378092854e-05,
+      "loss": 0.3949,
+      "step": 5039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9997219905476787e-05,
+      "loss": 0.3183,
+      "step": 5040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9983319432860718e-05,
+      "loss": 0.3091,
+      "step": 5041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.996941896024465e-05,
+      "loss": 0.3391,
+      "step": 5042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.995551848762858e-05,
+      "loss": 0.3672,
+      "step": 5043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.994161801501251e-05,
+      "loss": 0.2926,
+      "step": 5044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9927717542396445e-05,
+      "loss": 0.3695,
+      "step": 5045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9913817069780375e-05,
+      "loss": 0.3144,
+      "step": 5046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9899916597164308e-05,
+      "loss": 0.3075,
+      "step": 5047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9886016124548238e-05,
+      "loss": 0.3159,
+      "step": 5048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9872115651932168e-05,
+      "loss": 0.3819,
+      "step": 5049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9858215179316095e-05,
+      "loss": 0.3442,
+      "step": 5050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.984431470670003e-05,
+      "loss": 0.3195,
+      "step": 5051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.983041423408396e-05,
+      "loss": 0.2864,
+      "step": 5052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9816513761467892e-05,
+      "loss": 0.3003,
+      "step": 5053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9802613288851822e-05,
+      "loss": 0.3566,
+      "step": 5054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9788712816235752e-05,
+      "loss": 0.2804,
+      "step": 5055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9774812343619685e-05,
+      "loss": 0.3132,
+      "step": 5056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9760911871003615e-05,
+      "loss": 0.2836,
+      "step": 5057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.974701139838755e-05,
+      "loss": 0.3371,
+      "step": 5058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.973311092577148e-05,
+      "loss": 0.3288,
+      "step": 5059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.971921045315541e-05,
+      "loss": 0.4073,
+      "step": 5060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9705309980539343e-05,
+      "loss": 0.3428,
+      "step": 5061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9691409507923273e-05,
+      "loss": 0.3051,
+      "step": 5062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.96775090353072e-05,
+      "loss": 0.3063,
+      "step": 5063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.966360856269113e-05,
+      "loss": 0.2908,
+      "step": 5064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9649708090075063e-05,
+      "loss": 0.3542,
+      "step": 5065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9635807617458993e-05,
+      "loss": 0.3132,
+      "step": 5066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9621907144842926e-05,
+      "loss": 0.3467,
+      "step": 5067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9608006672226856e-05,
+      "loss": 0.3097,
+      "step": 5068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.959410619961079e-05,
+      "loss": 0.3487,
+      "step": 5069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.958020572699472e-05,
+      "loss": 0.3442,
+      "step": 5070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.956630525437865e-05,
+      "loss": 0.3174,
+      "step": 5071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9552404781762583e-05,
+      "loss": 0.3013,
+      "step": 5072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9538504309146513e-05,
+      "loss": 0.3556,
+      "step": 5073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9524603836530447e-05,
+      "loss": 0.342,
+      "step": 5074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9510703363914377e-05,
+      "loss": 0.3223,
+      "step": 5075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9496802891298304e-05,
+      "loss": 0.2904,
+      "step": 5076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9482902418682234e-05,
+      "loss": 0.3222,
+      "step": 5077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9469001946066167e-05,
+      "loss": 0.4333,
+      "step": 5078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9455101473450097e-05,
+      "loss": 0.3532,
+      "step": 5079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9441201000834027e-05,
+      "loss": 0.3062,
+      "step": 5080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.942730052821796e-05,
+      "loss": 0.3635,
+      "step": 5081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.941340005560189e-05,
+      "loss": 0.2955,
+      "step": 5082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9399499582985824e-05,
+      "loss": 0.3312,
+      "step": 5083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9385599110369754e-05,
+      "loss": 0.3188,
+      "step": 5084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9371698637753688e-05,
+      "loss": 0.2931,
+      "step": 5085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9357798165137618e-05,
+      "loss": 0.3311,
+      "step": 5086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9343897692521548e-05,
+      "loss": 0.3238,
+      "step": 5087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.932999721990548e-05,
+      "loss": 0.3107,
+      "step": 5088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9316096747289408e-05,
+      "loss": 0.3263,
+      "step": 5089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9302196274673338e-05,
+      "loss": 0.3094,
+      "step": 5090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9288295802057268e-05,
+      "loss": 0.3146,
+      "step": 5091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.92743953294412e-05,
+      "loss": 0.3019,
+      "step": 5092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.926049485682513e-05,
+      "loss": 0.2821,
+      "step": 5093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9246594384209065e-05,
+      "loss": 0.3642,
+      "step": 5094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9232693911592995e-05,
+      "loss": 0.3365,
+      "step": 5095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9218793438976925e-05,
+      "loss": 0.2996,
+      "step": 5096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.920489296636086e-05,
+      "loss": 0.2874,
+      "step": 5097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.919099249374479e-05,
+      "loss": 0.3487,
+      "step": 5098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9177092021128722e-05,
+      "loss": 0.3777,
+      "step": 5099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9163191548512652e-05,
+      "loss": 0.3124,
+      "step": 5100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9149291075896586e-05,
+      "loss": 0.347,
+      "step": 5101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.913539060328051e-05,
+      "loss": 0.3327,
+      "step": 5102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9121490130664442e-05,
+      "loss": 0.324,
+      "step": 5103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9107589658048373e-05,
+      "loss": 0.2843,
+      "step": 5104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9093689185432306e-05,
+      "loss": 0.3272,
+      "step": 5105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9079788712816236e-05,
+      "loss": 0.277,
+      "step": 5106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9065888240200166e-05,
+      "loss": 0.3516,
+      "step": 5107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.90519877675841e-05,
+      "loss": 0.3322,
+      "step": 5108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.903808729496803e-05,
+      "loss": 0.3321,
+      "step": 5109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9024186822351963e-05,
+      "loss": 0.3005,
+      "step": 5110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9010286349735893e-05,
+      "loss": 0.3407,
+      "step": 5111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8996385877119827e-05,
+      "loss": 0.3164,
+      "step": 5112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8982485404503757e-05,
+      "loss": 0.3418,
+      "step": 5113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8968584931887687e-05,
+      "loss": 0.3312,
+      "step": 5114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.895468445927162e-05,
+      "loss": 0.3511,
+      "step": 5115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8940783986655547e-05,
+      "loss": 0.343,
+      "step": 5116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8926883514039477e-05,
+      "loss": 0.3679,
+      "step": 5117
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8912983041423407e-05,
+      "loss": 0.3386,
+      "step": 5118
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.889908256880734e-05,
+      "loss": 0.3814,
+      "step": 5119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.888518209619127e-05,
+      "loss": 0.3269,
+      "step": 5120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8871281623575204e-05,
+      "loss": 0.3159,
+      "step": 5121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8857381150959134e-05,
+      "loss": 0.3173,
+      "step": 5122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8843480678343064e-05,
+      "loss": 0.3359,
+      "step": 5123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8829580205726998e-05,
+      "loss": 0.3223,
+      "step": 5124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8815679733110928e-05,
+      "loss": 0.3333,
+      "step": 5125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.880177926049486e-05,
+      "loss": 0.2949,
+      "step": 5126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.878787878787879e-05,
+      "loss": 0.3345,
+      "step": 5127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8773978315262725e-05,
+      "loss": 0.3469,
+      "step": 5128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8760077842646648e-05,
+      "loss": 0.381,
+      "step": 5129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.874617737003058e-05,
+      "loss": 0.3489,
+      "step": 5130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.873227689741451e-05,
+      "loss": 0.363,
+      "step": 5131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8718376424798445e-05,
+      "loss": 0.3634,
+      "step": 5132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8704475952182375e-05,
+      "loss": 0.2758,
+      "step": 5133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8690575479566305e-05,
+      "loss": 0.306,
+      "step": 5134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.867667500695024e-05,
+      "loss": 0.344,
+      "step": 5135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.866277453433417e-05,
+      "loss": 0.3416,
+      "step": 5136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8648874061718102e-05,
+      "loss": 0.293,
+      "step": 5137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8634973589102032e-05,
+      "loss": 0.3188,
+      "step": 5138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8621073116485962e-05,
+      "loss": 0.3083,
+      "step": 5139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8607172643869895e-05,
+      "loss": 0.3971,
+      "step": 5140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8593272171253826e-05,
+      "loss": 0.3529,
+      "step": 5141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8579371698637752e-05,
+      "loss": 0.3454,
+      "step": 5142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8565471226021682e-05,
+      "loss": 0.2946,
+      "step": 5143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8551570753405616e-05,
+      "loss": 0.3973,
+      "step": 5144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8537670280789546e-05,
+      "loss": 0.3096,
+      "step": 5145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.852376980817348e-05,
+      "loss": 0.3139,
+      "step": 5146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.850986933555741e-05,
+      "loss": 0.3071,
+      "step": 5147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8495968862941343e-05,
+      "loss": 0.3368,
+      "step": 5148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8482068390325273e-05,
+      "loss": 0.3441,
+      "step": 5149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8468167917709203e-05,
+      "loss": 0.3792,
+      "step": 5150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8454267445093136e-05,
+      "loss": 0.3447,
+      "step": 5151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8440366972477066e-05,
+      "loss": 0.303,
+      "step": 5152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8426466499861e-05,
+      "loss": 0.2812,
+      "step": 5153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.841256602724493e-05,
+      "loss": 0.2944,
+      "step": 5154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8398665554628857e-05,
+      "loss": 0.3205,
+      "step": 5155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8384765082012787e-05,
+      "loss": 0.353,
+      "step": 5156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.837086460939672e-05,
+      "loss": 0.3541,
+      "step": 5157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.835696413678065e-05,
+      "loss": 0.3168,
+      "step": 5158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.834306366416458e-05,
+      "loss": 0.2793,
+      "step": 5159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8329163191548514e-05,
+      "loss": 0.3495,
+      "step": 5160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8315262718932444e-05,
+      "loss": 0.2908,
+      "step": 5161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8301362246316377e-05,
+      "loss": 0.3441,
+      "step": 5162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8287461773700307e-05,
+      "loss": 0.3461,
+      "step": 5163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.827356130108424e-05,
+      "loss": 0.3299,
+      "step": 5164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.825966082846817e-05,
+      "loss": 0.3451,
+      "step": 5165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.82457603558521e-05,
+      "loss": 0.3049,
+      "step": 5166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8231859883236034e-05,
+      "loss": 0.3518,
+      "step": 5167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8217959410619964e-05,
+      "loss": 0.3449,
+      "step": 5168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.820405893800389e-05,
+      "loss": 0.3378,
+      "step": 5169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.819015846538782e-05,
+      "loss": 0.3362,
+      "step": 5170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8176257992771755e-05,
+      "loss": 0.3468,
+      "step": 5171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8162357520155685e-05,
+      "loss": 0.3322,
+      "step": 5172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8148457047539618e-05,
+      "loss": 0.3863,
+      "step": 5173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8134556574923548e-05,
+      "loss": 0.3743,
+      "step": 5174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.812065610230748e-05,
+      "loss": 0.2985,
+      "step": 5175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.810675562969141e-05,
+      "loss": 0.2952,
+      "step": 5176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8092855157075342e-05,
+      "loss": 0.3588,
+      "step": 5177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8078954684459275e-05,
+      "loss": 0.3256,
+      "step": 5178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8065054211843205e-05,
+      "loss": 0.364,
+      "step": 5179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.805115373922714e-05,
+      "loss": 0.3278,
+      "step": 5180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.803725326661107e-05,
+      "loss": 0.286,
+      "step": 5181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8023352793994995e-05,
+      "loss": 0.2993,
+      "step": 5182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8009452321378926e-05,
+      "loss": 0.3962,
+      "step": 5183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.799555184876286e-05,
+      "loss": 0.3547,
+      "step": 5184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.798165137614679e-05,
+      "loss": 0.3407,
+      "step": 5185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.796775090353072e-05,
+      "loss": 0.3507,
+      "step": 5186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7953850430914653e-05,
+      "loss": 0.3373,
+      "step": 5187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7939949958298583e-05,
+      "loss": 0.3134,
+      "step": 5188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7926049485682516e-05,
+      "loss": 0.3134,
+      "step": 5189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7912149013066446e-05,
+      "loss": 0.3463,
+      "step": 5190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.789824854045038e-05,
+      "loss": 0.3639,
+      "step": 5191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.788434806783431e-05,
+      "loss": 0.3485,
+      "step": 5192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.787044759521824e-05,
+      "loss": 0.3248,
+      "step": 5193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7856547122602173e-05,
+      "loss": 0.3812,
+      "step": 5194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.78426466499861e-05,
+      "loss": 0.3478,
+      "step": 5195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.782874617737003e-05,
+      "loss": 0.296,
+      "step": 5196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.781484570475396e-05,
+      "loss": 0.2956,
+      "step": 5197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7800945232137893e-05,
+      "loss": 0.3453,
+      "step": 5198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7787044759521823e-05,
+      "loss": 0.3158,
+      "step": 5199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7773144286905757e-05,
+      "loss": 0.3555,
+      "step": 5200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7759243814289687e-05,
+      "loss": 0.38,
+      "step": 5201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7745343341673617e-05,
+      "loss": 0.2688,
+      "step": 5202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.773144286905755e-05,
+      "loss": 0.3092,
+      "step": 5203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.771754239644148e-05,
+      "loss": 0.2977,
+      "step": 5204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7703641923825414e-05,
+      "loss": 0.3116,
+      "step": 5205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7689741451209344e-05,
+      "loss": 0.3555,
+      "step": 5206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7675840978593278e-05,
+      "loss": 0.3314,
+      "step": 5207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.76619405059772e-05,
+      "loss": 0.2634,
+      "step": 5208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7648040033361134e-05,
+      "loss": 0.3783,
+      "step": 5209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7634139560745064e-05,
+      "loss": 0.2945,
+      "step": 5210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7620239088128998e-05,
+      "loss": 0.2474,
+      "step": 5211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7606338615512928e-05,
+      "loss": 0.3414,
+      "step": 5212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7592438142896858e-05,
+      "loss": 0.3432,
+      "step": 5213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.757853767028079e-05,
+      "loss": 0.3003,
+      "step": 5214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.756463719766472e-05,
+      "loss": 0.304,
+      "step": 5215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7550736725048655e-05,
+      "loss": 0.3194,
+      "step": 5216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7536836252432585e-05,
+      "loss": 0.3345,
+      "step": 5217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 0.3629,
+      "step": 5218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.750903530720045e-05,
+      "loss": 0.421,
+      "step": 5219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.749513483458438e-05,
+      "loss": 0.3939,
+      "step": 5220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7481234361968305e-05,
+      "loss": 0.2845,
+      "step": 5221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7467333889352235e-05,
+      "loss": 0.3228,
+      "step": 5222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.745343341673617e-05,
+      "loss": 0.3082,
+      "step": 5223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.74395329441201e-05,
+      "loss": 0.2903,
+      "step": 5224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7425632471504032e-05,
+      "loss": 0.2933,
+      "step": 5225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7411731998887962e-05,
+      "loss": 0.36,
+      "step": 5226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7397831526271896e-05,
+      "loss": 0.2916,
+      "step": 5227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7383931053655826e-05,
+      "loss": 0.3703,
+      "step": 5228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7370030581039756e-05,
+      "loss": 0.313,
+      "step": 5229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.735613010842369e-05,
+      "loss": 0.3655,
+      "step": 5230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.734222963580762e-05,
+      "loss": 0.2979,
+      "step": 5231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7328329163191553e-05,
+      "loss": 0.3588,
+      "step": 5232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7314428690575483e-05,
+      "loss": 0.3343,
+      "step": 5233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7300528217959413e-05,
+      "loss": 0.3276,
+      "step": 5234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.728662774534334e-05,
+      "loss": 0.2809,
+      "step": 5235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.2704,
+      "step": 5236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7258826800111203e-05,
+      "loss": 0.3389,
+      "step": 5237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7244926327495137e-05,
+      "loss": 0.3679,
+      "step": 5238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7231025854879067e-05,
+      "loss": 0.2839,
+      "step": 5239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7217125382262997e-05,
+      "loss": 0.3118,
+      "step": 5240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.720322490964693e-05,
+      "loss": 0.2993,
+      "step": 5241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.718932443703086e-05,
+      "loss": 0.3075,
+      "step": 5242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7175423964414794e-05,
+      "loss": 0.3573,
+      "step": 5243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7161523491798724e-05,
+      "loss": 0.2804,
+      "step": 5244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7147623019182654e-05,
+      "loss": 0.3405,
+      "step": 5245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7133722546566587e-05,
+      "loss": 0.3286,
+      "step": 5246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7119822073950517e-05,
+      "loss": 0.3529,
+      "step": 5247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7105921601334444e-05,
+      "loss": 0.2932,
+      "step": 5248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7092021128718374e-05,
+      "loss": 0.3118,
+      "step": 5249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7078120656102308e-05,
+      "loss": 0.3288,
+      "step": 5250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7064220183486238e-05,
+      "loss": 0.3394,
+      "step": 5251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.705031971087017e-05,
+      "loss": 0.324,
+      "step": 5252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.70364192382541e-05,
+      "loss": 0.3366,
+      "step": 5253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7022518765638035e-05,
+      "loss": 0.353,
+      "step": 5254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7008618293021965e-05,
+      "loss": 0.3535,
+      "step": 5255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6994717820405895e-05,
+      "loss": 0.3507,
+      "step": 5256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6980817347789828e-05,
+      "loss": 0.3176,
+      "step": 5257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6966916875173758e-05,
+      "loss": 0.3295,
+      "step": 5258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.695301640255769e-05,
+      "loss": 0.293,
+      "step": 5259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6939115929941622e-05,
+      "loss": 0.271,
+      "step": 5260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.692521545732555e-05,
+      "loss": 0.3084,
+      "step": 5261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.691131498470948e-05,
+      "loss": 0.2938,
+      "step": 5262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6897414512093412e-05,
+      "loss": 0.3732,
+      "step": 5263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6883514039477342e-05,
+      "loss": 0.3815,
+      "step": 5264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6869613566861272e-05,
+      "loss": 0.311,
+      "step": 5265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6855713094245206e-05,
+      "loss": 0.3245,
+      "step": 5266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6841812621629136e-05,
+      "loss": 0.3183,
+      "step": 5267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.682791214901307e-05,
+      "loss": 0.3574,
+      "step": 5268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6814011676397e-05,
+      "loss": 0.288,
+      "step": 5269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6800111203780933e-05,
+      "loss": 0.3375,
+      "step": 5270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6786210731164863e-05,
+      "loss": 0.2862,
+      "step": 5271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6772310258548793e-05,
+      "loss": 0.3531,
+      "step": 5272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6758409785932726e-05,
+      "loss": 0.3215,
+      "step": 5273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6744509313316653e-05,
+      "loss": 0.3214,
+      "step": 5274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6730608840700583e-05,
+      "loss": 0.2924,
+      "step": 5275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6716708368084513e-05,
+      "loss": 0.3503,
+      "step": 5276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6702807895468446e-05,
+      "loss": 0.2768,
+      "step": 5277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6688907422852376e-05,
+      "loss": 0.2803,
+      "step": 5278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.667500695023631e-05,
+      "loss": 0.286,
+      "step": 5279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.666110647762024e-05,
+      "loss": 0.3002,
+      "step": 5280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.664720600500417e-05,
+      "loss": 0.3195,
+      "step": 5281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6633305532388103e-05,
+      "loss": 0.3527,
+      "step": 5282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6619405059772034e-05,
+      "loss": 0.3787,
+      "step": 5283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6605504587155967e-05,
+      "loss": 0.3097,
+      "step": 5284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6591604114539897e-05,
+      "loss": 0.3248,
+      "step": 5285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.657770364192383e-05,
+      "loss": 0.3313,
+      "step": 5286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.656380316930776e-05,
+      "loss": 0.3161,
+      "step": 5287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6549902696691687e-05,
+      "loss": 0.3185,
+      "step": 5288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6536002224075617e-05,
+      "loss": 0.2915,
+      "step": 5289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.652210175145955e-05,
+      "loss": 0.2462,
+      "step": 5290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.650820127884348e-05,
+      "loss": 0.2837,
+      "step": 5291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.649430080622741e-05,
+      "loss": 0.3056,
+      "step": 5292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6480400333611344e-05,
+      "loss": 0.3082,
+      "step": 5293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6466499860995274e-05,
+      "loss": 0.2923,
+      "step": 5294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6452599388379208e-05,
+      "loss": 0.3505,
+      "step": 5295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6438698915763138e-05,
+      "loss": 0.3508,
+      "step": 5296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6424798443147068e-05,
+      "loss": 0.3372,
+      "step": 5297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6410897970531e-05,
+      "loss": 0.3197,
+      "step": 5298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.639699749791493e-05,
+      "loss": 0.3381,
+      "step": 5299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6383097025298865e-05,
+      "loss": 0.2947,
+      "step": 5300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.636919655268279e-05,
+      "loss": 0.3659,
+      "step": 5301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6355296080066722e-05,
+      "loss": 0.317,
+      "step": 5302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6341395607450652e-05,
+      "loss": 0.3179,
+      "step": 5303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6327495134834585e-05,
+      "loss": 0.3381,
+      "step": 5304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6313594662218515e-05,
+      "loss": 0.3001,
+      "step": 5305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.629969418960245e-05,
+      "loss": 0.3882,
+      "step": 5306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.628579371698638e-05,
+      "loss": 0.3182,
+      "step": 5307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.627189324437031e-05,
+      "loss": 0.3099,
+      "step": 5308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6257992771754242e-05,
+      "loss": 0.3011,
+      "step": 5309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6244092299138172e-05,
+      "loss": 0.3172,
+      "step": 5310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6230191826522106e-05,
+      "loss": 0.3198,
+      "step": 5311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6216291353906036e-05,
+      "loss": 0.3525,
+      "step": 5312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.620239088128997e-05,
+      "loss": 0.309,
+      "step": 5313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6188490408673893e-05,
+      "loss": 0.2519,
+      "step": 5314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6174589936057826e-05,
+      "loss": 0.2973,
+      "step": 5315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6160689463441756e-05,
+      "loss": 0.2825,
+      "step": 5316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.614678899082569e-05,
+      "loss": 0.3254,
+      "step": 5317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.613288851820962e-05,
+      "loss": 0.3423,
+      "step": 5318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.611898804559355e-05,
+      "loss": 0.3422,
+      "step": 5319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6105087572977483e-05,
+      "loss": 0.2506,
+      "step": 5320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6091187100361413e-05,
+      "loss": 0.3195,
+      "step": 5321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6077286627745347e-05,
+      "loss": 0.3269,
+      "step": 5322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6063386155129277e-05,
+      "loss": 0.339,
+      "step": 5323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6049485682513207e-05,
+      "loss": 0.3318,
+      "step": 5324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.603558520989714e-05,
+      "loss": 0.3234,
+      "step": 5325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.602168473728107e-05,
+      "loss": 0.3334,
+      "step": 5326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6007784264664997e-05,
+      "loss": 0.3348,
+      "step": 5327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5993883792048927e-05,
+      "loss": 0.3436,
+      "step": 5328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.597998331943286e-05,
+      "loss": 0.3098,
+      "step": 5329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.596608284681679e-05,
+      "loss": 0.3327,
+      "step": 5330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5952182374200724e-05,
+      "loss": 0.3202,
+      "step": 5331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5938281901584654e-05,
+      "loss": 0.2978,
+      "step": 5332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5924381428968588e-05,
+      "loss": 0.3008,
+      "step": 5333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5910480956352518e-05,
+      "loss": 0.3546,
+      "step": 5334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5896580483736448e-05,
+      "loss": 0.2949,
+      "step": 5335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.588268001112038e-05,
+      "loss": 0.329,
+      "step": 5336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.586877953850431e-05,
+      "loss": 0.2867,
+      "step": 5337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5854879065888245e-05,
+      "loss": 0.3243,
+      "step": 5338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5840978593272175e-05,
+      "loss": 0.2827,
+      "step": 5339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.58270781206561e-05,
+      "loss": 0.3213,
+      "step": 5340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.581317764804003e-05,
+      "loss": 0.341,
+      "step": 5341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5799277175423965e-05,
+      "loss": 0.3669,
+      "step": 5342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5785376702807895e-05,
+      "loss": 0.3458,
+      "step": 5343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5771476230191825e-05,
+      "loss": 0.2769,
+      "step": 5344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.575757575757576e-05,
+      "loss": 0.3627,
+      "step": 5345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.574367528495969e-05,
+      "loss": 0.2658,
+      "step": 5346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5729774812343622e-05,
+      "loss": 0.3528,
+      "step": 5347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5715874339727552e-05,
+      "loss": 0.3073,
+      "step": 5348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5701973867111486e-05,
+      "loss": 0.2963,
+      "step": 5349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5688073394495416e-05,
+      "loss": 0.3378,
+      "step": 5350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5674172921879346e-05,
+      "loss": 0.3113,
+      "step": 5351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.566027244926328e-05,
+      "loss": 0.2956,
+      "step": 5352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.564637197664721e-05,
+      "loss": 0.2735,
+      "step": 5353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5632471504031136e-05,
+      "loss": 0.3066,
+      "step": 5354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5618571031415066e-05,
+      "loss": 0.3082,
+      "step": 5355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5604670558799e-05,
+      "loss": 0.3418,
+      "step": 5356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.559077008618293e-05,
+      "loss": 0.3435,
+      "step": 5357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5576869613566863e-05,
+      "loss": 0.346,
+      "step": 5358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5562969140950793e-05,
+      "loss": 0.2558,
+      "step": 5359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5549068668334723e-05,
+      "loss": 0.3154,
+      "step": 5360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5535168195718656e-05,
+      "loss": 0.3474,
+      "step": 5361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5521267723102587e-05,
+      "loss": 0.3578,
+      "step": 5362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.550736725048652e-05,
+      "loss": 0.3388,
+      "step": 5363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.549346677787045e-05,
+      "loss": 0.3335,
+      "step": 5364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5479566305254383e-05,
+      "loss": 0.3709,
+      "step": 5365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5465665832638314e-05,
+      "loss": 0.3367,
+      "step": 5366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.545176536002224e-05,
+      "loss": 0.2758,
+      "step": 5367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.543786488740617e-05,
+      "loss": 0.2935,
+      "step": 5368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5423964414790104e-05,
+      "loss": 0.2856,
+      "step": 5369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5410063942174034e-05,
+      "loss": 0.3376,
+      "step": 5370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5396163469557964e-05,
+      "loss": 0.322,
+      "step": 5371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5382262996941897e-05,
+      "loss": 0.3185,
+      "step": 5372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5368362524325827e-05,
+      "loss": 0.4147,
+      "step": 5373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.535446205170976e-05,
+      "loss": 0.3637,
+      "step": 5374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.534056157909369e-05,
+      "loss": 0.349,
+      "step": 5375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5326661106477624e-05,
+      "loss": 0.3316,
+      "step": 5376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5312760633861554e-05,
+      "loss": 0.3226,
+      "step": 5377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5298860161245484e-05,
+      "loss": 0.2703,
+      "step": 5378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5284959688629418e-05,
+      "loss": 0.2852,
+      "step": 5379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5271059216013345e-05,
+      "loss": 0.3572,
+      "step": 5380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5257158743397275e-05,
+      "loss": 0.3625,
+      "step": 5381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5243258270781205e-05,
+      "loss": 0.3471,
+      "step": 5382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5229357798165138e-05,
+      "loss": 0.3456,
+      "step": 5383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5215457325549068e-05,
+      "loss": 0.3258,
+      "step": 5384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5201556852933002e-05,
+      "loss": 0.3584,
+      "step": 5385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5187656380316932e-05,
+      "loss": 0.3152,
+      "step": 5386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5173755907700862e-05,
+      "loss": 0.3786,
+      "step": 5387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5159855435084795e-05,
+      "loss": 0.3029,
+      "step": 5388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5145954962468725e-05,
+      "loss": 0.3326,
+      "step": 5389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.513205448985266e-05,
+      "loss": 0.3004,
+      "step": 5390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.511815401723659e-05,
+      "loss": 0.3332,
+      "step": 5391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5104253544620522e-05,
+      "loss": 0.2923,
+      "step": 5392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5090353072004446e-05,
+      "loss": 0.2797,
+      "step": 5393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.507645259938838e-05,
+      "loss": 0.3187,
+      "step": 5394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.506255212677231e-05,
+      "loss": 0.331,
+      "step": 5395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5048651654156243e-05,
+      "loss": 0.3562,
+      "step": 5396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5034751181540173e-05,
+      "loss": 0.2613,
+      "step": 5397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5020850708924103e-05,
+      "loss": 0.2817,
+      "step": 5398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5006950236308036e-05,
+      "loss": 0.2941,
+      "step": 5399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4993049763691966e-05,
+      "loss": 0.3065,
+      "step": 5400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.49791492910759e-05,
+      "loss": 0.3812,
+      "step": 5401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.496524881845983e-05,
+      "loss": 0.3421,
+      "step": 5402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.495134834584376e-05,
+      "loss": 0.3216,
+      "step": 5403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.493744787322769e-05,
+      "loss": 0.2466,
+      "step": 5404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.492354740061162e-05,
+      "loss": 0.3034,
+      "step": 5405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4909646927995553e-05,
+      "loss": 0.3035,
+      "step": 5406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4895746455379483e-05,
+      "loss": 0.2887,
+      "step": 5407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4881845982763417e-05,
+      "loss": 0.341,
+      "step": 5408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4867945510147347e-05,
+      "loss": 0.3033,
+      "step": 5409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4854045037531277e-05,
+      "loss": 0.3444,
+      "step": 5410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4840144564915207e-05,
+      "loss": 0.3304,
+      "step": 5411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.482624409229914e-05,
+      "loss": 0.3388,
+      "step": 5412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.481234361968307e-05,
+      "loss": 0.3432,
+      "step": 5413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4798443147067e-05,
+      "loss": 0.3148,
+      "step": 5414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4784542674450934e-05,
+      "loss": 0.2977,
+      "step": 5415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4770642201834864e-05,
+      "loss": 0.3149,
+      "step": 5416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4756741729218794e-05,
+      "loss": 0.3228,
+      "step": 5417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4742841256602724e-05,
+      "loss": 0.3112,
+      "step": 5418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4728940783986658e-05,
+      "loss": 0.3364,
+      "step": 5419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4715040311370588e-05,
+      "loss": 0.363,
+      "step": 5420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4701139838754518e-05,
+      "loss": 0.3298,
+      "step": 5421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.468723936613845e-05,
+      "loss": 0.3233,
+      "step": 5422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4673338893522378e-05,
+      "loss": 0.3348,
+      "step": 5423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.465943842090631e-05,
+      "loss": 0.2901,
+      "step": 5424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.464553794829024e-05,
+      "loss": 0.3368,
+      "step": 5425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4631637475674175e-05,
+      "loss": 0.3383,
+      "step": 5426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4617737003058105e-05,
+      "loss": 0.3005,
+      "step": 5427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.460383653044204e-05,
+      "loss": 0.319,
+      "step": 5428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.458993605782597e-05,
+      "loss": 0.3348,
+      "step": 5429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.45760355852099e-05,
+      "loss": 0.3086,
+      "step": 5430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.456213511259383e-05,
+      "loss": 0.3083,
+      "step": 5431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.454823463997776e-05,
+      "loss": 0.3043,
+      "step": 5432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4534334167361692e-05,
+      "loss": 0.2892,
+      "step": 5433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4520433694745622e-05,
+      "loss": 0.2949,
+      "step": 5434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4506533222129556e-05,
+      "loss": 0.3734,
+      "step": 5435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4492632749513482e-05,
+      "loss": 0.2951,
+      "step": 5436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4478732276897416e-05,
+      "loss": 0.3592,
+      "step": 5437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4464831804281346e-05,
+      "loss": 0.3365,
+      "step": 5438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.445093133166528e-05,
+      "loss": 0.3154,
+      "step": 5439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.443703085904921e-05,
+      "loss": 0.2893,
+      "step": 5440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.442313038643314e-05,
+      "loss": 0.3092,
+      "step": 5441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4409229913817073e-05,
+      "loss": 0.3047,
+      "step": 5442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4395329441201e-05,
+      "loss": 0.3159,
+      "step": 5443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4381428968584933e-05,
+      "loss": 0.3167,
+      "step": 5444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4367528495968863e-05,
+      "loss": 0.32,
+      "step": 5445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4353628023352797e-05,
+      "loss": 0.3313,
+      "step": 5446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4339727550736727e-05,
+      "loss": 0.3036,
+      "step": 5447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4325827078120657e-05,
+      "loss": 0.3038,
+      "step": 5448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.431192660550459e-05,
+      "loss": 0.2988,
+      "step": 5449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4298026132888517e-05,
+      "loss": 0.2761,
+      "step": 5450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.428412566027245e-05,
+      "loss": 0.3007,
+      "step": 5451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.427022518765638e-05,
+      "loss": 0.3083,
+      "step": 5452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4256324715040314e-05,
+      "loss": 0.2554,
+      "step": 5453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 0.296,
+      "step": 5454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4228523769808177e-05,
+      "loss": 0.271,
+      "step": 5455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4214623297192104e-05,
+      "loss": 0.2829,
+      "step": 5456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4200722824576037e-05,
+      "loss": 0.3328,
+      "step": 5457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4186822351959968e-05,
+      "loss": 0.3254,
+      "step": 5458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4172921879343898e-05,
+      "loss": 0.3821,
+      "step": 5459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.415902140672783e-05,
+      "loss": 0.2987,
+      "step": 5460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.414512093411176e-05,
+      "loss": 0.315,
+      "step": 5461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4131220461495695e-05,
+      "loss": 0.267,
+      "step": 5462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.411731998887962e-05,
+      "loss": 0.3149,
+      "step": 5463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4103419516263555e-05,
+      "loss": 0.2812,
+      "step": 5464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4089519043647485e-05,
+      "loss": 0.336,
+      "step": 5465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4075618571031415e-05,
+      "loss": 0.3195,
+      "step": 5466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4061718098415348e-05,
+      "loss": 0.3656,
+      "step": 5467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.404781762579928e-05,
+      "loss": 0.3426,
+      "step": 5468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.403391715318321e-05,
+      "loss": 0.3126,
+      "step": 5469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.402001668056714e-05,
+      "loss": 0.3087,
+      "step": 5470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4006116207951072e-05,
+      "loss": 0.2697,
+      "step": 5471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3992215735335002e-05,
+      "loss": 0.3573,
+      "step": 5472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3978315262718935e-05,
+      "loss": 0.3006,
+      "step": 5473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3964414790102866e-05,
+      "loss": 0.3002,
+      "step": 5474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3950514317486796e-05,
+      "loss": 0.3372,
+      "step": 5475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3936613844870726e-05,
+      "loss": 0.3494,
+      "step": 5476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3922713372254656e-05,
+      "loss": 0.2996,
+      "step": 5477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.390881289963859e-05,
+      "loss": 0.2953,
+      "step": 5478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.389491242702252e-05,
+      "loss": 0.3319,
+      "step": 5479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3881011954406453e-05,
+      "loss": 0.2954,
+      "step": 5480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3867111481790383e-05,
+      "loss": 0.3332,
+      "step": 5481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3853211009174313e-05,
+      "loss": 0.3216,
+      "step": 5482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3839310536558243e-05,
+      "loss": 0.3566,
+      "step": 5483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3825410063942173e-05,
+      "loss": 0.3068,
+      "step": 5484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3811509591326106e-05,
+      "loss": 0.3353,
+      "step": 5485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3797609118710036e-05,
+      "loss": 0.33,
+      "step": 5486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.378370864609397e-05,
+      "loss": 0.299,
+      "step": 5487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.37698081734779e-05,
+      "loss": 0.2896,
+      "step": 5488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.375590770086183e-05,
+      "loss": 0.3372,
+      "step": 5489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.374200722824576e-05,
+      "loss": 0.3373,
+      "step": 5490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3728106755629694e-05,
+      "loss": 0.3014,
+      "step": 5491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3714206283013624e-05,
+      "loss": 0.2775,
+      "step": 5492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3700305810397554e-05,
+      "loss": 0.304,
+      "step": 5493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3686405337781487e-05,
+      "loss": 0.3223,
+      "step": 5494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3672504865165417e-05,
+      "loss": 0.299,
+      "step": 5495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3658604392549347e-05,
+      "loss": 0.3183,
+      "step": 5496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3644703919933277e-05,
+      "loss": 0.3052,
+      "step": 5497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.363080344731721e-05,
+      "loss": 0.274,
+      "step": 5498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.361690297470114e-05,
+      "loss": 0.2449,
+      "step": 5499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3603002502085074e-05,
+      "loss": 0.3528,
+      "step": 5500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3589102029469004e-05,
+      "loss": 0.2518,
+      "step": 5501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3575201556852934e-05,
+      "loss": 0.2821,
+      "step": 5502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3561301084236864e-05,
+      "loss": 0.2854,
+      "step": 5503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3547400611620795e-05,
+      "loss": 0.3157,
+      "step": 5504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3533500139004728e-05,
+      "loss": 0.2936,
+      "step": 5505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3519599666388658e-05,
+      "loss": 0.3041,
+      "step": 5506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.350569919377259e-05,
+      "loss": 0.3952,
+      "step": 5507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.349179872115652e-05,
+      "loss": 0.3177,
+      "step": 5508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.347789824854045e-05,
+      "loss": 0.3221,
+      "step": 5509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.346399777592438e-05,
+      "loss": 0.2915,
+      "step": 5510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3450097303308312e-05,
+      "loss": 0.2716,
+      "step": 5511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3436196830692245e-05,
+      "loss": 0.3202,
+      "step": 5512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3422296358076175e-05,
+      "loss": 0.3298,
+      "step": 5513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.340839588546011e-05,
+      "loss": 0.2993,
+      "step": 5514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.339449541284404e-05,
+      "loss": 0.3095,
+      "step": 5515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.338059494022797e-05,
+      "loss": 0.3278,
+      "step": 5516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.33666944676119e-05,
+      "loss": 0.3103,
+      "step": 5517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3352793994995832e-05,
+      "loss": 0.3311,
+      "step": 5518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3338893522379762e-05,
+      "loss": 0.3404,
+      "step": 5519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3324993049763692e-05,
+      "loss": 0.3151,
+      "step": 5520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3311092577147626e-05,
+      "loss": 0.3542,
+      "step": 5521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3297192104531553e-05,
+      "loss": 0.2536,
+      "step": 5522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3283291631915486e-05,
+      "loss": 0.3208,
+      "step": 5523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3269391159299416e-05,
+      "loss": 0.2961,
+      "step": 5524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.325549068668335e-05,
+      "loss": 0.2801,
+      "step": 5525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.324159021406728e-05,
+      "loss": 0.2785,
+      "step": 5526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.322768974145121e-05,
+      "loss": 0.3264,
+      "step": 5527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3213789268835143e-05,
+      "loss": 0.3112,
+      "step": 5528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.319988879621907e-05,
+      "loss": 0.3474,
+      "step": 5529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3185988323603003e-05,
+      "loss": 0.3108,
+      "step": 5530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3172087850986933e-05,
+      "loss": 0.6804,
+      "step": 5531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3158187378370867e-05,
+      "loss": 0.2782,
+      "step": 5532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3144286905754797e-05,
+      "loss": 0.3067,
+      "step": 5533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.313038643313873e-05,
+      "loss": 0.2641,
+      "step": 5534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.311648596052266e-05,
+      "loss": 0.3031,
+      "step": 5535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.310258548790659e-05,
+      "loss": 0.3551,
+      "step": 5536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.308868501529052e-05,
+      "loss": 0.3132,
+      "step": 5537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.307478454267445e-05,
+      "loss": 0.3723,
+      "step": 5538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3060884070058384e-05,
+      "loss": 0.3565,
+      "step": 5539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3046983597442314e-05,
+      "loss": 0.3542,
+      "step": 5540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3033083124826248e-05,
+      "loss": 0.3249,
+      "step": 5541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3019182652210174e-05,
+      "loss": 0.3098,
+      "step": 5542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3005282179594108e-05,
+      "loss": 0.3372,
+      "step": 5543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2991381706978038e-05,
+      "loss": 0.2812,
+      "step": 5544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2977481234361968e-05,
+      "loss": 0.2946,
+      "step": 5545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.29635807617459e-05,
+      "loss": 0.3517,
+      "step": 5546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.294968028912983e-05,
+      "loss": 0.3001,
+      "step": 5547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2935779816513765e-05,
+      "loss": 0.3534,
+      "step": 5548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.292187934389769e-05,
+      "loss": 0.3116,
+      "step": 5549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2907978871281625e-05,
+      "loss": 0.345,
+      "step": 5550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2894078398665555e-05,
+      "loss": 0.3083,
+      "step": 5551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.288017792604949e-05,
+      "loss": 0.3599,
+      "step": 5552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.286627745343342e-05,
+      "loss": 0.295,
+      "step": 5553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.285237698081735e-05,
+      "loss": 0.2944,
+      "step": 5554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.283847650820128e-05,
+      "loss": 0.3226,
+      "step": 5555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.282457603558521e-05,
+      "loss": 0.3592,
+      "step": 5556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2810675562969142e-05,
+      "loss": 0.3381,
+      "step": 5557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2796775090353072e-05,
+      "loss": 0.3183,
+      "step": 5558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2782874617737006e-05,
+      "loss": 0.3126,
+      "step": 5559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2768974145120936e-05,
+      "loss": 0.3434,
+      "step": 5560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2755073672504866e-05,
+      "loss": 0.3307,
+      "step": 5561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2741173199888796e-05,
+      "loss": 0.3431,
+      "step": 5562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.3153,
+      "step": 5563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.271337225465666e-05,
+      "loss": 0.3282,
+      "step": 5564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.269947178204059e-05,
+      "loss": 0.3001,
+      "step": 5565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2685571309424523e-05,
+      "loss": 0.2498,
+      "step": 5566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2671670836808453e-05,
+      "loss": 0.2937,
+      "step": 5567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2657770364192386e-05,
+      "loss": 0.3129,
+      "step": 5568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2643869891576313e-05,
+      "loss": 0.3344,
+      "step": 5569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2629969418960247e-05,
+      "loss": 0.3289,
+      "step": 5570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2616068946344177e-05,
+      "loss": 0.3305,
+      "step": 5571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2602168473728107e-05,
+      "loss": 0.3075,
+      "step": 5572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.258826800111204e-05,
+      "loss": 0.3536,
+      "step": 5573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.257436752849597e-05,
+      "loss": 0.2933,
+      "step": 5574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.25604670558799e-05,
+      "loss": 0.3633,
+      "step": 5575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.254656658326383e-05,
+      "loss": 0.3682,
+      "step": 5576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2532666110647764e-05,
+      "loss": 0.3053,
+      "step": 5577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2518765638031694e-05,
+      "loss": 0.3283,
+      "step": 5578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2504865165415627e-05,
+      "loss": 0.3178,
+      "step": 5579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2490964692799557e-05,
+      "loss": 0.3167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2477064220183487e-05,
+      "loss": 0.2405,
+      "step": 5581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2463163747567417e-05,
+      "loss": 0.3464,
+      "step": 5582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2449263274951348e-05,
+      "loss": 0.3259,
+      "step": 5583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.243536280233528e-05,
+      "loss": 0.3246,
+      "step": 5584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.242146232971921e-05,
+      "loss": 0.336,
+      "step": 5585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2407561857103144e-05,
+      "loss": 0.3339,
+      "step": 5586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2393661384487075e-05,
+      "loss": 0.3141,
+      "step": 5587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2379760911871005e-05,
+      "loss": 0.3192,
+      "step": 5588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2365860439254935e-05,
+      "loss": 0.3657,
+      "step": 5589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2351959966638865e-05,
+      "loss": 0.3014,
+      "step": 5590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2338059494022798e-05,
+      "loss": 0.3012,
+      "step": 5591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2324159021406728e-05,
+      "loss": 0.3379,
+      "step": 5592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2310258548790662e-05,
+      "loss": 0.2975,
+      "step": 5593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2296358076174592e-05,
+      "loss": 0.3974,
+      "step": 5594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2282457603558522e-05,
+      "loss": 0.3798,
+      "step": 5595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2268557130942452e-05,
+      "loss": 0.4246,
+      "step": 5596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2254656658326385e-05,
+      "loss": 0.337,
+      "step": 5597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2240756185710315e-05,
+      "loss": 0.3007,
+      "step": 5598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2226855713094245e-05,
+      "loss": 0.2951,
+      "step": 5599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.221295524047818e-05,
+      "loss": 0.282,
+      "step": 5600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.219905476786211e-05,
+      "loss": 0.2849,
+      "step": 5601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.218515429524604e-05,
+      "loss": 0.3261,
+      "step": 5602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.217125382262997e-05,
+      "loss": 0.3361,
+      "step": 5603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2157353350013903e-05,
+      "loss": 0.3408,
+      "step": 5604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2143452877397833e-05,
+      "loss": 0.3584,
+      "step": 5605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2129552404781763e-05,
+      "loss": 0.3107,
+      "step": 5606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2115651932165696e-05,
+      "loss": 0.2848,
+      "step": 5607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2101751459549623e-05,
+      "loss": 0.3137,
+      "step": 5608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2087850986933556e-05,
+      "loss": 0.2678,
+      "step": 5609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2073950514317486e-05,
+      "loss": 0.2937,
+      "step": 5610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.206005004170142e-05,
+      "loss": 0.3468,
+      "step": 5611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.204614956908535e-05,
+      "loss": 0.2879,
+      "step": 5612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2032249096469283e-05,
+      "loss": 0.3188,
+      "step": 5613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2018348623853213e-05,
+      "loss": 0.313,
+      "step": 5614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2004448151237143e-05,
+      "loss": 0.2974,
+      "step": 5615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1990547678621073e-05,
+      "loss": 0.2933,
+      "step": 5616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1976647206005004e-05,
+      "loss": 0.3178,
+      "step": 5617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1962746733388937e-05,
+      "loss": 0.2752,
+      "step": 5618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1948846260772867e-05,
+      "loss": 0.3001,
+      "step": 5619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.19349457881568e-05,
+      "loss": 0.3444,
+      "step": 5620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.192104531554073e-05,
+      "loss": 0.2875,
+      "step": 5621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.190714484292466e-05,
+      "loss": 0.3499,
+      "step": 5622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.189324437030859e-05,
+      "loss": 0.3357,
+      "step": 5623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.187934389769252e-05,
+      "loss": 0.2828,
+      "step": 5624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1865443425076454e-05,
+      "loss": 0.326,
+      "step": 5625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1851542952460384e-05,
+      "loss": 0.3526,
+      "step": 5626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1837642479844318e-05,
+      "loss": 0.2839,
+      "step": 5627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1823742007228244e-05,
+      "loss": 0.3047,
+      "step": 5628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1809841534612178e-05,
+      "loss": 0.3454,
+      "step": 5629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1795941061996108e-05,
+      "loss": 0.2769,
+      "step": 5630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.178204058938004e-05,
+      "loss": 0.3226,
+      "step": 5631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.176814011676397e-05,
+      "loss": 0.3281,
+      "step": 5632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.17542396441479e-05,
+      "loss": 0.3149,
+      "step": 5633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1740339171531835e-05,
+      "loss": 0.2766,
+      "step": 5634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.172643869891576e-05,
+      "loss": 0.3152,
+      "step": 5635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1712538226299695e-05,
+      "loss": 0.3128,
+      "step": 5636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1698637753683625e-05,
+      "loss": 0.3203,
+      "step": 5637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.168473728106756e-05,
+      "loss": 0.3416,
+      "step": 5638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.167083680845149e-05,
+      "loss": 0.3484,
+      "step": 5639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1656936335835422e-05,
+      "loss": 0.356,
+      "step": 5640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.164303586321935e-05,
+      "loss": 0.2812,
+      "step": 5641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1629135390603282e-05,
+      "loss": 0.3485,
+      "step": 5642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1615234917987212e-05,
+      "loss": 0.307,
+      "step": 5643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1601334445371142e-05,
+      "loss": 0.2521,
+      "step": 5644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1587433972755076e-05,
+      "loss": 0.3093,
+      "step": 5645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1573533500139006e-05,
+      "loss": 0.2566,
+      "step": 5646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.155963302752294e-05,
+      "loss": 0.2964,
+      "step": 5647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1545732554906866e-05,
+      "loss": 0.306,
+      "step": 5648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15318320822908e-05,
+      "loss": 0.2774,
+      "step": 5649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151793160967473e-05,
+      "loss": 0.2656,
+      "step": 5650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150403113705866e-05,
+      "loss": 0.2883,
+      "step": 5651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1490130664442593e-05,
+      "loss": 0.2966,
+      "step": 5652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1476230191826523e-05,
+      "loss": 0.2737,
+      "step": 5653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1462329719210457e-05,
+      "loss": 0.3629,
+      "step": 5654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1448429246594383e-05,
+      "loss": 0.336,
+      "step": 5655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1434528773978317e-05,
+      "loss": 0.2837,
+      "step": 5656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1420628301362247e-05,
+      "loss": 0.3149,
+      "step": 5657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140672782874618e-05,
+      "loss": 0.3682,
+      "step": 5658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139282735613011e-05,
+      "loss": 0.2948,
+      "step": 5659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137892688351404e-05,
+      "loss": 0.2791,
+      "step": 5660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136502641089797e-05,
+      "loss": 0.3809,
+      "step": 5661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13511259382819e-05,
+      "loss": 0.3152,
+      "step": 5662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1337225465665834e-05,
+      "loss": 0.3074,
+      "step": 5663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1323324993049764e-05,
+      "loss": 0.3185,
+      "step": 5664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1309424520433697e-05,
+      "loss": 0.3201,
+      "step": 5665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1295524047817628e-05,
+      "loss": 0.2893,
+      "step": 5666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1281623575201558e-05,
+      "loss": 0.2902,
+      "step": 5667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1267723102585488e-05,
+      "loss": 0.3258,
+      "step": 5668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1253822629969418e-05,
+      "loss": 0.306,
+      "step": 5669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.123992215735335e-05,
+      "loss": 0.2854,
+      "step": 5670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.122602168473728e-05,
+      "loss": 0.3081,
+      "step": 5671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1212121212121215e-05,
+      "loss": 0.2964,
+      "step": 5672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1198220739505145e-05,
+      "loss": 0.3084,
+      "step": 5673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1184320266889075e-05,
+      "loss": 0.2904,
+      "step": 5674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1170419794273005e-05,
+      "loss": 0.2813,
+      "step": 5675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.115651932165694e-05,
+      "loss": 0.3446,
+      "step": 5676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.114261884904087e-05,
+      "loss": 0.3567,
+      "step": 5677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.11287183764248e-05,
+      "loss": 0.31,
+      "step": 5678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1114817903808732e-05,
+      "loss": 0.3472,
+      "step": 5679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1100917431192662e-05,
+      "loss": 0.2754,
+      "step": 5680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1087016958576592e-05,
+      "loss": 0.311,
+      "step": 5681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1073116485960522e-05,
+      "loss": 0.3027,
+      "step": 5682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1059216013344456e-05,
+      "loss": 0.3733,
+      "step": 5683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1045315540728386e-05,
+      "loss": 0.2605,
+      "step": 5684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1031415068112316e-05,
+      "loss": 0.3618,
+      "step": 5685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.101751459549625e-05,
+      "loss": 0.2983,
+      "step": 5686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.100361412288018e-05,
+      "loss": 0.3149,
+      "step": 5687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.098971365026411e-05,
+      "loss": 0.3389,
+      "step": 5688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.097581317764804e-05,
+      "loss": 0.3167,
+      "step": 5689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0961912705031973e-05,
+      "loss": 0.3311,
+      "step": 5690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0948012232415903e-05,
+      "loss": 0.3322,
+      "step": 5691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0934111759799836e-05,
+      "loss": 0.287,
+      "step": 5692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0920211287183766e-05,
+      "loss": 0.3393,
+      "step": 5693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0906310814567696e-05,
+      "loss": 0.3375,
+      "step": 5694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0892410341951626e-05,
+      "loss": 0.3167,
+      "step": 5695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0878509869335557e-05,
+      "loss": 0.3916,
+      "step": 5696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.086460939671949e-05,
+      "loss": 0.2979,
+      "step": 5697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.085070892410342e-05,
+      "loss": 0.3095,
+      "step": 5698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0836808451487354e-05,
+      "loss": 0.2956,
+      "step": 5699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0822907978871284e-05,
+      "loss": 0.307,
+      "step": 5700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0809007506255214e-05,
+      "loss": 0.3024,
+      "step": 5701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0795107033639144e-05,
+      "loss": 0.3108,
+      "step": 5702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0781206561023077e-05,
+      "loss": 0.3532,
+      "step": 5703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0767306088407007e-05,
+      "loss": 0.3373,
+      "step": 5704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0753405615790937e-05,
+      "loss": 0.2626,
+      "step": 5705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.073950514317487e-05,
+      "loss": 0.3149,
+      "step": 5706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0725604670558797e-05,
+      "loss": 0.3446,
+      "step": 5707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.071170419794273e-05,
+      "loss": 0.3279,
+      "step": 5708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069780372532666e-05,
+      "loss": 0.3574,
+      "step": 5709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0683903252710594e-05,
+      "loss": 0.3007,
+      "step": 5710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0670002780094524e-05,
+      "loss": 0.3001,
+      "step": 5711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0656102307478455e-05,
+      "loss": 0.2902,
+      "step": 5712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0642201834862388e-05,
+      "loss": 0.3585,
+      "step": 5713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0628301362246315e-05,
+      "loss": 0.3109,
+      "step": 5714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0614400889630248e-05,
+      "loss": 0.2779,
+      "step": 5715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0600500417014178e-05,
+      "loss": 0.3343,
+      "step": 5716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.058659994439811e-05,
+      "loss": 0.3223,
+      "step": 5717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.057269947178204e-05,
+      "loss": 0.296,
+      "step": 5718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0558798999165975e-05,
+      "loss": 0.2746,
+      "step": 5719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0544898526549905e-05,
+      "loss": 0.2933,
+      "step": 5720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0530998053933835e-05,
+      "loss": 0.2683,
+      "step": 5721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517097581317765e-05,
+      "loss": 0.342,
+      "step": 5722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0503197108701695e-05,
+      "loss": 0.3175,
+      "step": 5723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048929663608563e-05,
+      "loss": 0.2738,
+      "step": 5724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047539616346956e-05,
+      "loss": 0.2577,
+      "step": 5725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0461495690853492e-05,
+      "loss": 0.2913,
+      "step": 5726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044759521823742e-05,
+      "loss": 0.2764,
+      "step": 5727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0433694745621352e-05,
+      "loss": 0.2499,
+      "step": 5728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0419794273005283e-05,
+      "loss": 0.3248,
+      "step": 5729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0405893800389213e-05,
+      "loss": 0.3696,
+      "step": 5730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0391993327773146e-05,
+      "loss": 0.3272,
+      "step": 5731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0378092855157076e-05,
+      "loss": 0.3201,
+      "step": 5732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036419238254101e-05,
+      "loss": 0.3616,
+      "step": 5733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0350291909924936e-05,
+      "loss": 0.3022,
+      "step": 5734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033639143730887e-05,
+      "loss": 0.3321,
+      "step": 5735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03224909646928e-05,
+      "loss": 0.2983,
+      "step": 5736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0308590492076733e-05,
+      "loss": 0.3134,
+      "step": 5737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0294690019460663e-05,
+      "loss": 0.295,
+      "step": 5738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0280789546844593e-05,
+      "loss": 0.2437,
+      "step": 5739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0266889074228527e-05,
+      "loss": 0.3046,
+      "step": 5740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0252988601612453e-05,
+      "loss": 0.2763,
+      "step": 5741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0239088128996387e-05,
+      "loss": 0.305,
+      "step": 5742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0225187656380317e-05,
+      "loss": 0.3052,
+      "step": 5743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021128718376425e-05,
+      "loss": 0.3055,
+      "step": 5744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019738671114818e-05,
+      "loss": 0.3258,
+      "step": 5745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018348623853211e-05,
+      "loss": 0.307,
+      "step": 5746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016958576591604e-05,
+      "loss": 0.3101,
+      "step": 5747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015568529329997e-05,
+      "loss": 0.2934,
+      "step": 5748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0141784820683904e-05,
+      "loss": 0.2775,
+      "step": 5749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0127884348067834e-05,
+      "loss": 0.3105,
+      "step": 5750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0113983875451768e-05,
+      "loss": 0.3284,
+      "step": 5751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0100083402835698e-05,
+      "loss": 0.3117,
+      "step": 5752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008618293021963e-05,
+      "loss": 0.3292,
+      "step": 5753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0072282457603558e-05,
+      "loss": 0.2816,
+      "step": 5754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005838198498749e-05,
+      "loss": 0.3095,
+      "step": 5755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004448151237142e-05,
+      "loss": 0.3045,
+      "step": 5756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003058103975535e-05,
+      "loss": 0.3589,
+      "step": 5757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0016680567139285e-05,
+      "loss": 0.3408,
+      "step": 5758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0002780094523215e-05,
+      "loss": 0.3219,
+      "step": 5759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9988879621907145e-05,
+      "loss": 0.2995,
+      "step": 5760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9974979149291075e-05,
+      "loss": 0.3643,
+      "step": 5761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996107867667501e-05,
+      "loss": 0.3272,
+      "step": 5762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994717820405894e-05,
+      "loss": 0.2935,
+      "step": 5763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9933277731442872e-05,
+      "loss": 0.3572,
+      "step": 5764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9919377258826802e-05,
+      "loss": 0.3302,
+      "step": 5765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9905476786210732e-05,
+      "loss": 0.2955,
+      "step": 5766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9891576313594662e-05,
+      "loss": 0.3221,
+      "step": 5767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9877675840978592e-05,
+      "loss": 0.2942,
+      "step": 5768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9863775368362526e-05,
+      "loss": 0.3828,
+      "step": 5769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9849874895746456e-05,
+      "loss": 0.3062,
+      "step": 5770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983597442313039e-05,
+      "loss": 0.2881,
+      "step": 5771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982207395051432e-05,
+      "loss": 0.2955,
+      "step": 5772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980817347789825e-05,
+      "loss": 0.3178,
+      "step": 5773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979427300528218e-05,
+      "loss": 0.3531,
+      "step": 5774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978037253266611e-05,
+      "loss": 0.3769,
+      "step": 5775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9766472060050043e-05,
+      "loss": 0.36,
+      "step": 5776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9752571587433973e-05,
+      "loss": 0.2999,
+      "step": 5777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9738671114817906e-05,
+      "loss": 0.2623,
+      "step": 5778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9724770642201837e-05,
+      "loss": 0.2947,
+      "step": 5779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9710870169585767e-05,
+      "loss": 0.3163,
+      "step": 5780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 0.3566,
+      "step": 5781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968306922435363e-05,
+      "loss": 0.2919,
+      "step": 5782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966916875173756e-05,
+      "loss": 0.3508,
+      "step": 5783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965526827912149e-05,
+      "loss": 0.2951,
+      "step": 5784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9641367806505424e-05,
+      "loss": 0.3586,
+      "step": 5785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9627467333889354e-05,
+      "loss": 0.2647,
+      "step": 5786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9613566861273284e-05,
+      "loss": 0.301,
+      "step": 5787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9599666388657214e-05,
+      "loss": 0.3424,
+      "step": 5788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9585765916041147e-05,
+      "loss": 0.3017,
+      "step": 5789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9571865443425077e-05,
+      "loss": 0.2773,
+      "step": 5790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9557964970809007e-05,
+      "loss": 0.2497,
+      "step": 5791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.954406449819294e-05,
+      "loss": 0.3483,
+      "step": 5792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9530164025576868e-05,
+      "loss": 0.3272,
+      "step": 5793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.95162635529608e-05,
+      "loss": 0.2989,
+      "step": 5794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.950236308034473e-05,
+      "loss": 0.2803,
+      "step": 5795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9488462607728665e-05,
+      "loss": 0.369,
+      "step": 5796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9474562135112595e-05,
+      "loss": 0.3394,
+      "step": 5797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9460661662496528e-05,
+      "loss": 0.2937,
+      "step": 5798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9446761189880458e-05,
+      "loss": 0.3103,
+      "step": 5799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9432860717264388e-05,
+      "loss": 0.2336,
+      "step": 5800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9418960244648318e-05,
+      "loss": 0.2846,
+      "step": 5801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.940505977203225e-05,
+      "loss": 0.3194,
+      "step": 5802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9391159299416182e-05,
+      "loss": 0.3634,
+      "step": 5803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9377258826800112e-05,
+      "loss": 0.3126,
+      "step": 5804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9363358354184045e-05,
+      "loss": 0.327,
+      "step": 5805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9349457881567975e-05,
+      "loss": 0.3174,
+      "step": 5806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9335557408951905e-05,
+      "loss": 0.3558,
+      "step": 5807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9321656936335836e-05,
+      "loss": 0.3344,
+      "step": 5808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9307756463719766e-05,
+      "loss": 0.2992,
+      "step": 5809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92938559911037e-05,
+      "loss": 0.3196,
+      "step": 5810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927995551848763e-05,
+      "loss": 0.259,
+      "step": 5811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 0.2911,
+      "step": 5812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925215457325549e-05,
+      "loss": 0.3577,
+      "step": 5813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9238254100639423e-05,
+      "loss": 0.2887,
+      "step": 5814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9224353628023353e-05,
+      "loss": 0.3153,
+      "step": 5815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9210453155407286e-05,
+      "loss": 0.2692,
+      "step": 5816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9196552682791216e-05,
+      "loss": 0.4123,
+      "step": 5817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9182652210175146e-05,
+      "loss": 0.2897,
+      "step": 5818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916875173755908e-05,
+      "loss": 0.3569,
+      "step": 5819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9154851264943006e-05,
+      "loss": 0.3163,
+      "step": 5820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914095079232694e-05,
+      "loss": 0.3258,
+      "step": 5821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912705031971087e-05,
+      "loss": 0.3811,
+      "step": 5822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113149847094803e-05,
+      "loss": 0.3356,
+      "step": 5823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099249374478733e-05,
+      "loss": 0.3205,
+      "step": 5824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9085348901862664e-05,
+      "loss": 0.2741,
+      "step": 5825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9071448429246594e-05,
+      "loss": 0.3415,
+      "step": 5826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9057547956630527e-05,
+      "loss": 0.369,
+      "step": 5827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9043647484014457e-05,
+      "loss": 0.3359,
+      "step": 5828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9029747011398387e-05,
+      "loss": 0.2836,
+      "step": 5829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.901584653878232e-05,
+      "loss": 0.319,
+      "step": 5830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900194606616625e-05,
+      "loss": 0.348,
+      "step": 5831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8988045593550184e-05,
+      "loss": 0.3422,
+      "step": 5832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897414512093411e-05,
+      "loss": 0.3147,
+      "step": 5833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8960244648318044e-05,
+      "loss": 0.3069,
+      "step": 5834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8946344175701974e-05,
+      "loss": 0.3294,
+      "step": 5835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8932443703085904e-05,
+      "loss": 0.386,
+      "step": 5836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8918543230469838e-05,
+      "loss": 0.2703,
+      "step": 5837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8904642757853768e-05,
+      "loss": 0.3208,
+      "step": 5838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.88907422852377e-05,
+      "loss": 0.3626,
+      "step": 5839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8876841812621628e-05,
+      "loss": 0.2898,
+      "step": 5840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.886294134000556e-05,
+      "loss": 0.3635,
+      "step": 5841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884904086738949e-05,
+      "loss": 0.27,
+      "step": 5842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8835140394773425e-05,
+      "loss": 0.3077,
+      "step": 5843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8821239922157355e-05,
+      "loss": 0.2862,
+      "step": 5844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8807339449541285e-05,
+      "loss": 0.3682,
+      "step": 5845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8793438976925215e-05,
+      "loss": 0.2826,
+      "step": 5846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8779538504309145e-05,
+      "loss": 0.2747,
+      "step": 5847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876563803169308e-05,
+      "loss": 0.2705,
+      "step": 5848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875173755907701e-05,
+      "loss": 0.2872,
+      "step": 5849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8737837086460942e-05,
+      "loss": 0.2853,
+      "step": 5850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8723936613844872e-05,
+      "loss": 0.338,
+      "step": 5851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8710036141228802e-05,
+      "loss": 0.3314,
+      "step": 5852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8696135668612732e-05,
+      "loss": 0.3174,
+      "step": 5853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8682235195996663e-05,
+      "loss": 0.314,
+      "step": 5854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8668334723380596e-05,
+      "loss": 0.2831,
+      "step": 5855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8654434250764526e-05,
+      "loss": 0.3179,
+      "step": 5856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.864053377814846e-05,
+      "loss": 0.3374,
+      "step": 5857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.862663330553239e-05,
+      "loss": 0.3257,
+      "step": 5858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8612732832916323e-05,
+      "loss": 0.342,
+      "step": 5859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.859883236030025e-05,
+      "loss": 0.2854,
+      "step": 5860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8584931887684183e-05,
+      "loss": 0.3408,
+      "step": 5861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8571031415068113e-05,
+      "loss": 0.3409,
+      "step": 5862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8557130942452043e-05,
+      "loss": 0.3457,
+      "step": 5863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8543230469835977e-05,
+      "loss": 0.3501,
+      "step": 5864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8529329997219907e-05,
+      "loss": 0.3471,
+      "step": 5865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8515429524603837e-05,
+      "loss": 0.3216,
+      "step": 5866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8501529051987767e-05,
+      "loss": 0.3293,
+      "step": 5867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.84876285793717e-05,
+      "loss": 0.2795,
+      "step": 5868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.847372810675563e-05,
+      "loss": 0.3368,
+      "step": 5869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.845982763413956e-05,
+      "loss": 0.3831,
+      "step": 5870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8445927161523494e-05,
+      "loss": 0.3074,
+      "step": 5871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8432026688907424e-05,
+      "loss": 0.2651,
+      "step": 5872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8418126216291354e-05,
+      "loss": 0.3624,
+      "step": 5873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8404225743675284e-05,
+      "loss": 0.2957,
+      "step": 5874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8390325271059218e-05,
+      "loss": 0.2896,
+      "step": 5875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8376424798443148e-05,
+      "loss": 0.234,
+      "step": 5876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.836252432582708e-05,
+      "loss": 0.3153,
+      "step": 5877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 0.3375,
+      "step": 5878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.833472338059494e-05,
+      "loss": 0.2849,
+      "step": 5879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.832082290797887e-05,
+      "loss": 0.3495,
+      "step": 5880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.83069224353628e-05,
+      "loss": 0.2863,
+      "step": 5881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8293021962746735e-05,
+      "loss": 0.2891,
+      "step": 5882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8279121490130665e-05,
+      "loss": 0.3365,
+      "step": 5883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8265221017514598e-05,
+      "loss": 0.2987,
+      "step": 5884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.825132054489853e-05,
+      "loss": 0.3187,
+      "step": 5885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.823742007228246e-05,
+      "loss": 0.3282,
+      "step": 5886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.822351959966639e-05,
+      "loss": 0.3511,
+      "step": 5887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.820961912705032e-05,
+      "loss": 0.3396,
+      "step": 5888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8195718654434252e-05,
+      "loss": 0.3632,
+      "step": 5889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.3261,
+      "step": 5890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8167917709202116e-05,
+      "loss": 0.2988,
+      "step": 5891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8154017236586046e-05,
+      "loss": 0.2834,
+      "step": 5892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8140116763969976e-05,
+      "loss": 0.3166,
+      "step": 5893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8126216291353906e-05,
+      "loss": 0.3209,
+      "step": 5894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.811231581873784e-05,
+      "loss": 0.3071,
+      "step": 5895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.809841534612177e-05,
+      "loss": 0.3309,
+      "step": 5896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.80845148735057e-05,
+      "loss": 0.3134,
+      "step": 5897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8070614400889633e-05,
+      "loss": 0.3374,
+      "step": 5898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.805671392827356e-05,
+      "loss": 0.3148,
+      "step": 5899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8042813455657493e-05,
+      "loss": 0.3194,
+      "step": 5900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8028912983041423e-05,
+      "loss": 0.3184,
+      "step": 5901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8015012510425356e-05,
+      "loss": 0.3017,
+      "step": 5902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8001112037809286e-05,
+      "loss": 0.3491,
+      "step": 5903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.798721156519322e-05,
+      "loss": 0.278,
+      "step": 5904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.797331109257715e-05,
+      "loss": 0.307,
+      "step": 5905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.795941061996108e-05,
+      "loss": 0.308,
+      "step": 5906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.794551014734501e-05,
+      "loss": 0.3197,
+      "step": 5907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.793160967472894e-05,
+      "loss": 0.284,
+      "step": 5908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7917709202112874e-05,
+      "loss": 0.3446,
+      "step": 5909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7903808729496804e-05,
+      "loss": 0.2899,
+      "step": 5910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7889908256880737e-05,
+      "loss": 0.3109,
+      "step": 5911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7876007784264664e-05,
+      "loss": 0.3347,
+      "step": 5912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7862107311648597e-05,
+      "loss": 0.3406,
+      "step": 5913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7848206839032527e-05,
+      "loss": 0.3235,
+      "step": 5914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7834306366416457e-05,
+      "loss": 0.3218,
+      "step": 5915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.782040589380039e-05,
+      "loss": 0.2889,
+      "step": 5916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.780650542118432e-05,
+      "loss": 0.3761,
+      "step": 5917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7792604948568254e-05,
+      "loss": 0.3083,
+      "step": 5918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.777870447595218e-05,
+      "loss": 0.2701,
+      "step": 5919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7764804003336114e-05,
+      "loss": 0.3108,
+      "step": 5920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7750903530720045e-05,
+      "loss": 0.3444,
+      "step": 5921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7737003058103978e-05,
+      "loss": 0.3471,
+      "step": 5922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7723102585487908e-05,
+      "loss": 0.3174,
+      "step": 5923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7709202112871838e-05,
+      "loss": 0.2819,
+      "step": 5924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.769530164025577e-05,
+      "loss": 0.3476,
+      "step": 5925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7681401167639698e-05,
+      "loss": 0.3338,
+      "step": 5926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7667500695023632e-05,
+      "loss": 0.3042,
+      "step": 5927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7653600222407562e-05,
+      "loss": 0.3317,
+      "step": 5928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7639699749791495e-05,
+      "loss": 0.3047,
+      "step": 5929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7625799277175425e-05,
+      "loss": 0.3206,
+      "step": 5930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7611898804559355e-05,
+      "loss": 0.3213,
+      "step": 5931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7597998331943285e-05,
+      "loss": 0.3166,
+      "step": 5932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7584097859327215e-05,
+      "loss": 0.3119,
+      "step": 5933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.757019738671115e-05,
+      "loss": 0.3274,
+      "step": 5934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.755629691409508e-05,
+      "loss": 0.2861,
+      "step": 5935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7542396441479012e-05,
+      "loss": 0.2905,
+      "step": 5936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7528495968862943e-05,
+      "loss": 0.3431,
+      "step": 5937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7514595496246876e-05,
+      "loss": 0.3562,
+      "step": 5938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7500695023630803e-05,
+      "loss": 0.3144,
+      "step": 5939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7486794551014736e-05,
+      "loss": 0.3175,
+      "step": 5940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7472894078398666e-05,
+      "loss": 0.3541,
+      "step": 5941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7458993605782596e-05,
+      "loss": 0.3132,
+      "step": 5942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.744509313316653e-05,
+      "loss": 0.3015,
+      "step": 5943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.743119266055046e-05,
+      "loss": 0.2973,
+      "step": 5944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7417292187934393e-05,
+      "loss": 0.2903,
+      "step": 5945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.740339171531832e-05,
+      "loss": 0.3029,
+      "step": 5946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7389491242702253e-05,
+      "loss": 0.2756,
+      "step": 5947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7375590770086183e-05,
+      "loss": 0.295,
+      "step": 5948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7361690297470113e-05,
+      "loss": 0.3326,
+      "step": 5949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7347789824854047e-05,
+      "loss": 0.3128,
+      "step": 5950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7333889352237977e-05,
+      "loss": 0.3304,
+      "step": 5951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7319988879621907e-05,
+      "loss": 0.3664,
+      "step": 5952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7306088407005837e-05,
+      "loss": 0.3228,
+      "step": 5953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.729218793438977e-05,
+      "loss": 0.2908,
+      "step": 5954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.72782874617737e-05,
+      "loss": 0.3594,
+      "step": 5955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7264386989157634e-05,
+      "loss": 0.3217,
+      "step": 5956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7250486516541564e-05,
+      "loss": 0.2774,
+      "step": 5957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7236586043925494e-05,
+      "loss": 0.3158,
+      "step": 5958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7222685571309424e-05,
+      "loss": 0.281,
+      "step": 5959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7208785098693354e-05,
+      "loss": 0.3122,
+      "step": 5960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7194884626077288e-05,
+      "loss": 0.3323,
+      "step": 5961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7180984153461218e-05,
+      "loss": 0.3396,
+      "step": 5962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.716708368084515e-05,
+      "loss": 0.3736,
+      "step": 5963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.715318320822908e-05,
+      "loss": 0.3024,
+      "step": 5964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.713928273561301e-05,
+      "loss": 0.2994,
+      "step": 5965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.712538226299694e-05,
+      "loss": 0.3358,
+      "step": 5966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7111481790380875e-05,
+      "loss": 0.3943,
+      "step": 5967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7097581317764805e-05,
+      "loss": 0.3039,
+      "step": 5968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7083680845148735e-05,
+      "loss": 0.3379,
+      "step": 5969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.706978037253267e-05,
+      "loss": 0.284,
+      "step": 5970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70558798999166e-05,
+      "loss": 0.3329,
+      "step": 5971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.704197942730053e-05,
+      "loss": 0.2862,
+      "step": 5972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.702807895468446e-05,
+      "loss": 0.3176,
+      "step": 5973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7014178482068392e-05,
+      "loss": 0.3256,
+      "step": 5974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000278009452322e-05,
+      "loss": 0.3312,
+      "step": 5975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986377536836252e-05,
+      "loss": 0.3102,
+      "step": 5976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972477064220186e-05,
+      "loss": 0.2991,
+      "step": 5977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958576591604116e-05,
+      "loss": 0.3253,
+      "step": 5978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944676118988046e-05,
+      "loss": 0.2832,
+      "step": 5979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6930775646371976e-05,
+      "loss": 0.2962,
+      "step": 5980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.691687517375591e-05,
+      "loss": 0.3271,
+      "step": 5981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.690297470113984e-05,
+      "loss": 0.3273,
+      "step": 5982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6889074228523773e-05,
+      "loss": 0.34,
+      "step": 5983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6875173755907703e-05,
+      "loss": 0.2711,
+      "step": 5984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6861273283291633e-05,
+      "loss": 0.3038,
+      "step": 5985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6847372810675563e-05,
+      "loss": 0.2781,
+      "step": 5986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6833472338059493e-05,
+      "loss": 0.326,
+      "step": 5987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6819571865443427e-05,
+      "loss": 0.3091,
+      "step": 5988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6805671392827357e-05,
+      "loss": 0.3474,
+      "step": 5989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.679177092021129e-05,
+      "loss": 0.3428,
+      "step": 5990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.677787044759522e-05,
+      "loss": 0.3085,
+      "step": 5991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.676396997497915e-05,
+      "loss": 0.3463,
+      "step": 5992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.675006950236308e-05,
+      "loss": 0.2866,
+      "step": 5993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.673616902974701e-05,
+      "loss": 0.3327,
+      "step": 5994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6722268557130944e-05,
+      "loss": 0.3223,
+      "step": 5995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6708368084514874e-05,
+      "loss": 0.2771,
+      "step": 5996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6694467611898807e-05,
+      "loss": 0.2964,
+      "step": 5997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6680567139282734e-05,
+      "loss": 0.2643,
+      "step": 5998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.3017,
+      "step": 5999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6652766194050598e-05,
+      "loss": 0.3014,
+      "step": 6000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.663886572143453e-05,
+      "loss": 0.3169,
+      "step": 6001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.662496524881846e-05,
+      "loss": 0.3088,
+      "step": 6002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.661106477620239e-05,
+      "loss": 0.329,
+      "step": 6003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6597164303586325e-05,
+      "loss": 0.3347,
+      "step": 6004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.658326383097025e-05,
+      "loss": 0.3059,
+      "step": 6005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6569363358354185e-05,
+      "loss": 0.2498,
+      "step": 6006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6555462885738115e-05,
+      "loss": 0.3075,
+      "step": 6007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6541562413122048e-05,
+      "loss": 0.3327,
+      "step": 6008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6527661940505978e-05,
+      "loss": 0.3393,
+      "step": 6009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.651376146788991e-05,
+      "loss": 0.2581,
+      "step": 6010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6499860995273842e-05,
+      "loss": 0.2877,
+      "step": 6011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.648596052265777e-05,
+      "loss": 0.2854,
+      "step": 6012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6472060050041702e-05,
+      "loss": 0.3179,
+      "step": 6013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6458159577425632e-05,
+      "loss": 0.2777,
+      "step": 6014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6444259104809565e-05,
+      "loss": 0.3073,
+      "step": 6015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6430358632193495e-05,
+      "loss": 0.2559,
+      "step": 6016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.641645815957743e-05,
+      "loss": 0.3417,
+      "step": 6017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6402557686961356e-05,
+      "loss": 0.3286,
+      "step": 6018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.638865721434529e-05,
+      "loss": 0.2892,
+      "step": 6019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.637475674172922e-05,
+      "loss": 0.3543,
+      "step": 6020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.636085626911315e-05,
+      "loss": 0.3021,
+      "step": 6021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6346955796497083e-05,
+      "loss": 0.3207,
+      "step": 6022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6333055323881013e-05,
+      "loss": 0.2843,
+      "step": 6023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6319154851264946e-05,
+      "loss": 0.3163,
+      "step": 6024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6305254378648873e-05,
+      "loss": 0.2928,
+      "step": 6025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6291353906032806e-05,
+      "loss": 0.2571,
+      "step": 6026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6277453433416736e-05,
+      "loss": 0.2878,
+      "step": 6027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.626355296080067e-05,
+      "loss": 0.3027,
+      "step": 6028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.62496524881846e-05,
+      "loss": 0.3321,
+      "step": 6029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.623575201556853e-05,
+      "loss": 0.3183,
+      "step": 6030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.622185154295246e-05,
+      "loss": 0.3104,
+      "step": 6031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.620795107033639e-05,
+      "loss": 0.3476,
+      "step": 6032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6194050597720324e-05,
+      "loss": 0.3037,
+      "step": 6033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6180150125104254e-05,
+      "loss": 0.3241,
+      "step": 6034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6166249652488187e-05,
+      "loss": 0.3224,
+      "step": 6035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6152349179872117e-05,
+      "loss": 0.3162,
+      "step": 6036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6138448707256047e-05,
+      "loss": 0.3267,
+      "step": 6037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6124548234639977e-05,
+      "loss": 0.3429,
+      "step": 6038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6110647762023907e-05,
+      "loss": 0.3174,
+      "step": 6039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.609674728940784e-05,
+      "loss": 0.3508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.608284681679177e-05,
+      "loss": 0.2871,
+      "step": 6041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6068946344175704e-05,
+      "loss": 0.3884,
+      "step": 6042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6055045871559634e-05,
+      "loss": 0.2853,
+      "step": 6043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6041145398943568e-05,
+      "loss": 0.3497,
+      "step": 6044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6027244926327494e-05,
+      "loss": 0.3105,
+      "step": 6045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6013344453711428e-05,
+      "loss": 0.3532,
+      "step": 6046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5999443981095358e-05,
+      "loss": 0.3284,
+      "step": 6047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5985543508479288e-05,
+      "loss": 0.3339,
+      "step": 6048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.597164303586322e-05,
+      "loss": 0.3072,
+      "step": 6049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.595774256324715e-05,
+      "loss": 0.3302,
+      "step": 6050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.594384209063108e-05,
+      "loss": 0.3805,
+      "step": 6051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.592994161801501e-05,
+      "loss": 0.3022,
+      "step": 6052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5916041145398945e-05,
+      "loss": 0.2981,
+      "step": 6053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5902140672782875e-05,
+      "loss": 0.3374,
+      "step": 6054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5888240200166805e-05,
+      "loss": 0.2955,
+      "step": 6055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.587433972755074e-05,
+      "loss": 0.2906,
+      "step": 6056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.586043925493467e-05,
+      "loss": 0.2954,
+      "step": 6057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.58465387823186e-05,
+      "loss": 0.3388,
+      "step": 6058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.583263830970253e-05,
+      "loss": 0.2862,
+      "step": 6059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5818737837086462e-05,
+      "loss": 0.3104,
+      "step": 6060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5804837364470392e-05,
+      "loss": 0.2916,
+      "step": 6061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5790936891854326e-05,
+      "loss": 0.2377,
+      "step": 6062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5777036419238256e-05,
+      "loss": 0.2993,
+      "step": 6063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5763135946622186e-05,
+      "loss": 0.2983,
+      "step": 6064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5749235474006116e-05,
+      "loss": 0.3156,
+      "step": 6065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5735335001390046e-05,
+      "loss": 0.2865,
+      "step": 6066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.572143452877398e-05,
+      "loss": 0.3432,
+      "step": 6067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.570753405615791e-05,
+      "loss": 0.2699,
+      "step": 6068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5693633583541843e-05,
+      "loss": 0.3003,
+      "step": 6069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5679733110925773e-05,
+      "loss": 0.302,
+      "step": 6070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5665832638309703e-05,
+      "loss": 0.3174,
+      "step": 6071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5651932165693633e-05,
+      "loss": 0.3387,
+      "step": 6072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5638031693077563e-05,
+      "loss": 0.3452,
+      "step": 6073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5624131220461497e-05,
+      "loss": 0.3415,
+      "step": 6074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5610230747845427e-05,
+      "loss": 0.3034,
+      "step": 6075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.559633027522936e-05,
+      "loss": 0.279,
+      "step": 6076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.558242980261329e-05,
+      "loss": 0.358,
+      "step": 6077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.556852932999722e-05,
+      "loss": 0.3309,
+      "step": 6078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.555462885738115e-05,
+      "loss": 0.3025,
+      "step": 6079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5540728384765084e-05,
+      "loss": 0.3317,
+      "step": 6080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5526827912149014e-05,
+      "loss": 0.3108,
+      "step": 6081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5512927439532944e-05,
+      "loss": 0.32,
+      "step": 6082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5499026966916878e-05,
+      "loss": 0.3106,
+      "step": 6083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5485126494300804e-05,
+      "loss": 0.3561,
+      "step": 6084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5471226021684738e-05,
+      "loss": 0.337,
+      "step": 6085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5457325549068668e-05,
+      "loss": 0.32,
+      "step": 6086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.54434250764526e-05,
+      "loss": 0.2961,
+      "step": 6087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.542952460383653e-05,
+      "loss": 0.2645,
+      "step": 6088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.541562413122046e-05,
+      "loss": 0.2989,
+      "step": 6089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5401723658604395e-05,
+      "loss": 0.2762,
+      "step": 6090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5387823185988325e-05,
+      "loss": 0.3109,
+      "step": 6091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5373922713372255e-05,
+      "loss": 0.2866,
+      "step": 6092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5360022240756185e-05,
+      "loss": 0.2979,
+      "step": 6093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.534612176814012e-05,
+      "loss": 0.275,
+      "step": 6094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.533222129552405e-05,
+      "loss": 0.2765,
+      "step": 6095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5318320822907982e-05,
+      "loss": 0.2663,
+      "step": 6096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5304420350291912e-05,
+      "loss": 0.3045,
+      "step": 6097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5290519877675842e-05,
+      "loss": 0.3645,
+      "step": 6098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5276619405059772e-05,
+      "loss": 0.2481,
+      "step": 6099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5262718932443702e-05,
+      "loss": 0.2928,
+      "step": 6100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5248818459827636e-05,
+      "loss": 0.307,
+      "step": 6101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5234917987211567e-05,
+      "loss": 0.2642,
+      "step": 6102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5221017514595497e-05,
+      "loss": 0.3066,
+      "step": 6103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5207117041979428e-05,
+      "loss": 0.329,
+      "step": 6104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5193216569363358e-05,
+      "loss": 0.3427,
+      "step": 6105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.517931609674729e-05,
+      "loss": 0.3039,
+      "step": 6106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5165415624131221e-05,
+      "loss": 0.3383,
+      "step": 6107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.3078,
+      "step": 6108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5137614678899085e-05,
+      "loss": 0.2268,
+      "step": 6109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5123714206283016e-05,
+      "loss": 0.2752,
+      "step": 6110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5109813733666945e-05,
+      "loss": 0.3072,
+      "step": 6111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5095913261050877e-05,
+      "loss": 0.2885,
+      "step": 6112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5082012788434807e-05,
+      "loss": 0.2866,
+      "step": 6113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5068112315818738e-05,
+      "loss": 0.2736,
+      "step": 6114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.505421184320267e-05,
+      "loss": 0.3265,
+      "step": 6115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5040311370586602e-05,
+      "loss": 0.2696,
+      "step": 6116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.502641089797053e-05,
+      "loss": 0.3504,
+      "step": 6117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5012510425354462e-05,
+      "loss": 0.3954,
+      "step": 6118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4998609952738394e-05,
+      "loss": 0.2723,
+      "step": 6119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4984709480122325e-05,
+      "loss": 0.3355,
+      "step": 6120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4970809007506256e-05,
+      "loss": 0.2526,
+      "step": 6121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4956908534890187e-05,
+      "loss": 0.2834,
+      "step": 6122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4943008062274119e-05,
+      "loss": 0.2943,
+      "step": 6123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4929107589658047e-05,
+      "loss": 0.2852,
+      "step": 6124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.491520711704198e-05,
+      "loss": 0.2858,
+      "step": 6125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4901306644425911e-05,
+      "loss": 0.3119,
+      "step": 6126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4887406171809843e-05,
+      "loss": 0.2675,
+      "step": 6127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4873505699193774e-05,
+      "loss": 0.3035,
+      "step": 6128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4859605226577705e-05,
+      "loss": 0.3242,
+      "step": 6129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4845704753961636e-05,
+      "loss": 0.3311,
+      "step": 6130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4831804281345565e-05,
+      "loss": 0.2814,
+      "step": 6131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4817903808729496e-05,
+      "loss": 0.2794,
+      "step": 6132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4804003336113428e-05,
+      "loss": 0.3298,
+      "step": 6133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.479010286349736e-05,
+      "loss": 0.3772,
+      "step": 6134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4776202390881292e-05,
+      "loss": 0.2997,
+      "step": 6135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4762301918265223e-05,
+      "loss": 0.3131,
+      "step": 6136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4748401445649152e-05,
+      "loss": 0.3618,
+      "step": 6137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4734500973033084e-05,
+      "loss": 0.2941,
+      "step": 6138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4720600500417014e-05,
+      "loss": 0.3136,
+      "step": 6139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4706700027800945e-05,
+      "loss": 0.3487,
+      "step": 6140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4692799555184877e-05,
+      "loss": 0.3049,
+      "step": 6141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4678899082568809e-05,
+      "loss": 0.2679,
+      "step": 6142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.466499860995274e-05,
+      "loss": 0.2768,
+      "step": 6143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4651098137336669e-05,
+      "loss": 0.3068,
+      "step": 6144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.46371976647206e-05,
+      "loss": 0.2784,
+      "step": 6145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4623297192104533e-05,
+      "loss": 0.2736,
+      "step": 6146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4609396719488463e-05,
+      "loss": 0.2928,
+      "step": 6147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4595496246872394e-05,
+      "loss": 0.3725,
+      "step": 6148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4581595774256326e-05,
+      "loss": 0.3068,
+      "step": 6149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4567695301640255e-05,
+      "loss": 0.2893,
+      "step": 6150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4553794829024186e-05,
+      "loss": 0.3025,
+      "step": 6151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4539894356408118e-05,
+      "loss": 0.2799,
+      "step": 6152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.452599388379205e-05,
+      "loss": 0.3261,
+      "step": 6153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4512093411175982e-05,
+      "loss": 0.3403,
+      "step": 6154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4498192938559913e-05,
+      "loss": 0.2721,
+      "step": 6155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4484292465943843e-05,
+      "loss": 0.2373,
+      "step": 6156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4470391993327773e-05,
+      "loss": 0.2436,
+      "step": 6157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4456491520711703e-05,
+      "loss": 0.3194,
+      "step": 6158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4442591048095635e-05,
+      "loss": 0.3009,
+      "step": 6159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4428690575479567e-05,
+      "loss": 0.2842,
+      "step": 6160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4414790102863499e-05,
+      "loss": 0.3601,
+      "step": 6161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.440088963024743e-05,
+      "loss": 0.2574,
+      "step": 6162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4386989157631362e-05,
+      "loss": 0.2617,
+      "step": 6163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.437308868501529e-05,
+      "loss": 0.281,
+      "step": 6164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4359188212399222e-05,
+      "loss": 0.3175,
+      "step": 6165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4345287739783152e-05,
+      "loss": 0.3415,
+      "step": 6166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4331387267167084e-05,
+      "loss": 0.3042,
+      "step": 6167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4317486794551016e-05,
+      "loss": 0.2978,
+      "step": 6168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4303586321934948e-05,
+      "loss": 0.2761,
+      "step": 6169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4289685849318876e-05,
+      "loss": 0.2949,
+      "step": 6170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4275785376702808e-05,
+      "loss": 0.3099,
+      "step": 6171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.426188490408674e-05,
+      "loss": 0.2898,
+      "step": 6172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4247984431470671e-05,
+      "loss": 0.3048,
+      "step": 6173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4234083958854601e-05,
+      "loss": 0.2874,
+      "step": 6174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4220183486238533e-05,
+      "loss": 0.2536,
+      "step": 6175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4206283013622465e-05,
+      "loss": 0.2958,
+      "step": 6176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4192382541006393e-05,
+      "loss": 0.3038,
+      "step": 6177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4178482068390325e-05,
+      "loss": 0.3552,
+      "step": 6178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4164581595774257e-05,
+      "loss": 0.3183,
+      "step": 6179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4150681123158189e-05,
+      "loss": 0.271,
+      "step": 6180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.413678065054212e-05,
+      "loss": 0.2914,
+      "step": 6181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.412288017792605e-05,
+      "loss": 0.3033,
+      "step": 6182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4108979705309982e-05,
+      "loss": 0.3087,
+      "step": 6183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.409507923269391e-05,
+      "loss": 0.2827,
+      "step": 6184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4081178760077842e-05,
+      "loss": 0.3299,
+      "step": 6185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4067278287461774e-05,
+      "loss": 0.3004,
+      "step": 6186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4053377814845706e-05,
+      "loss": 0.2483,
+      "step": 6187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4039477342229638e-05,
+      "loss": 0.2913,
+      "step": 6188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.402557686961357e-05,
+      "loss": 0.2811,
+      "step": 6189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4011676396997498e-05,
+      "loss": 0.2429,
+      "step": 6190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.399777592438143e-05,
+      "loss": 0.2603,
+      "step": 6191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.398387545176536e-05,
+      "loss": 0.3025,
+      "step": 6192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3969974979149291e-05,
+      "loss": 0.2413,
+      "step": 6193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3956074506533223e-05,
+      "loss": 0.3413,
+      "step": 6194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3942174033917155e-05,
+      "loss": 0.309,
+      "step": 6195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3928273561301087e-05,
+      "loss": 0.3493,
+      "step": 6196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3914373088685015e-05,
+      "loss": 0.3635,
+      "step": 6197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3900472616068947e-05,
+      "loss": 0.3181,
+      "step": 6198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3886572143452878e-05,
+      "loss": 0.287,
+      "step": 6199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3872671670836809e-05,
+      "loss": 0.2812,
+      "step": 6200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.385877119822074e-05,
+      "loss": 0.3215,
+      "step": 6201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3844870725604672e-05,
+      "loss": 0.244,
+      "step": 6202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.38309702529886e-05,
+      "loss": 0.393,
+      "step": 6203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3817069780372532e-05,
+      "loss": 0.3049,
+      "step": 6204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3803169307756464e-05,
+      "loss": 0.2818,
+      "step": 6205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3789268835140396e-05,
+      "loss": 0.2933,
+      "step": 6206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3775368362524327e-05,
+      "loss": 0.2915,
+      "step": 6207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 0.2995,
+      "step": 6208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.374756741729219e-05,
+      "loss": 0.2626,
+      "step": 6209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3733666944676118e-05,
+      "loss": 0.3183,
+      "step": 6210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.371976647206005e-05,
+      "loss": 0.3188,
+      "step": 6211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3705865999443981e-05,
+      "loss": 0.3327,
+      "step": 6212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3691965526827913e-05,
+      "loss": 0.3125,
+      "step": 6213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3678065054211845e-05,
+      "loss": 0.3002,
+      "step": 6214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3664164581595776e-05,
+      "loss": 0.2929,
+      "step": 6215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3650264108979706e-05,
+      "loss": 0.275,
+      "step": 6216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.3862,
+      "step": 6217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3622463163747568e-05,
+      "loss": 0.3428,
+      "step": 6218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3608562691131498e-05,
+      "loss": 0.2897,
+      "step": 6219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.359466221851543e-05,
+      "loss": 0.307,
+      "step": 6220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3580761745899362e-05,
+      "loss": 0.3162,
+      "step": 6221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3566861273283294e-05,
+      "loss": 0.2844,
+      "step": 6222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3552960800667222e-05,
+      "loss": 0.3165,
+      "step": 6223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3539060328051154e-05,
+      "loss": 0.3649,
+      "step": 6224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3525159855435086e-05,
+      "loss": 0.2908,
+      "step": 6225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3511259382819017e-05,
+      "loss": 0.3349,
+      "step": 6226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3497358910202947e-05,
+      "loss": 0.3352,
+      "step": 6227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3483458437586879e-05,
+      "loss": 0.2819,
+      "step": 6228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3469557964970811e-05,
+      "loss": 0.3347,
+      "step": 6229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.345565749235474e-05,
+      "loss": 0.2784,
+      "step": 6230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3441757019738671e-05,
+      "loss": 0.3026,
+      "step": 6231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3427856547122603e-05,
+      "loss": 0.3302,
+      "step": 6232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3413956074506535e-05,
+      "loss": 0.3554,
+      "step": 6233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3400055601890466e-05,
+      "loss": 0.3483,
+      "step": 6234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3386155129274396e-05,
+      "loss": 0.315,
+      "step": 6235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3372254656658326e-05,
+      "loss": 0.3227,
+      "step": 6236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3358354184042256e-05,
+      "loss": 0.2828,
+      "step": 6237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3344453711426188e-05,
+      "loss": 0.2702,
+      "step": 6238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.333055323881012e-05,
+      "loss": 0.2697,
+      "step": 6239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3316652766194052e-05,
+      "loss": 0.3053,
+      "step": 6240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3302752293577984e-05,
+      "loss": 0.3399,
+      "step": 6241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3288851820961915e-05,
+      "loss": 0.3517,
+      "step": 6242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3274951348345844e-05,
+      "loss": 0.2992,
+      "step": 6243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3261050875729775e-05,
+      "loss": 0.3317,
+      "step": 6244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3247150403113705e-05,
+      "loss": 0.3228,
+      "step": 6245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3233249930497637e-05,
+      "loss": 0.3405,
+      "step": 6246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3219349457881569e-05,
+      "loss": 0.2961,
+      "step": 6247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.32054489852655e-05,
+      "loss": 0.3549,
+      "step": 6248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3191548512649432e-05,
+      "loss": 0.3311,
+      "step": 6249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3177648040033361e-05,
+      "loss": 0.2949,
+      "step": 6250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3163747567417293e-05,
+      "loss": 0.3461,
+      "step": 6251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3149847094801224e-05,
+      "loss": 0.3178,
+      "step": 6252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3135946622185154e-05,
+      "loss": 0.3125,
+      "step": 6253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3135946622185154e-05,
+      "loss": 0.3133,
+      "step": 6254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3122046149569086e-05,
+      "loss": 0.3627,
+      "step": 6255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3108145676953018e-05,
+      "loss": 0.3245,
+      "step": 6256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3094245204336946e-05,
+      "loss": 0.2947,
+      "step": 6257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3080344731720878e-05,
+      "loss": 0.3845,
+      "step": 6258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.306644425910481e-05,
+      "loss": 0.3094,
+      "step": 6259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3052543786488742e-05,
+      "loss": 0.2526,
+      "step": 6260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3038643313872673e-05,
+      "loss": 0.3169,
+      "step": 6261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3024742841256603e-05,
+      "loss": 0.3063,
+      "step": 6262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3010842368640535e-05,
+      "loss": 0.3184,
+      "step": 6263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2996941896024464e-05,
+      "loss": 0.3222,
+      "step": 6264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2983041423408395e-05,
+      "loss": 0.266,
+      "step": 6265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2969140950792327e-05,
+      "loss": 0.3191,
+      "step": 6266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2955240478176259e-05,
+      "loss": 0.2916,
+      "step": 6267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.294134000556019e-05,
+      "loss": 0.3249,
+      "step": 6268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2927439532944122e-05,
+      "loss": 0.3138,
+      "step": 6269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.291353906032805e-05,
+      "loss": 0.3235,
+      "step": 6270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2899638587711982e-05,
+      "loss": 0.3351,
+      "step": 6271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2885738115095913e-05,
+      "loss": 0.2744,
+      "step": 6272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2871837642479844e-05,
+      "loss": 0.2871,
+      "step": 6273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2857937169863776e-05,
+      "loss": 0.3064,
+      "step": 6274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2844036697247708e-05,
+      "loss": 0.2852,
+      "step": 6275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.283013622463164e-05,
+      "loss": 0.2551,
+      "step": 6276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2816235752015568e-05,
+      "loss": 0.3184,
+      "step": 6277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.28023352793995e-05,
+      "loss": 0.338,
+      "step": 6278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2788434806783431e-05,
+      "loss": 0.3045,
+      "step": 6279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2774534334167362e-05,
+      "loss": 0.2768,
+      "step": 6280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2760633861551293e-05,
+      "loss": 0.307,
+      "step": 6281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2746733388935225e-05,
+      "loss": 0.3535,
+      "step": 6282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2732832916319157e-05,
+      "loss": 0.3499,
+      "step": 6283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2718932443703085e-05,
+      "loss": 0.3279,
+      "step": 6284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2705031971087017e-05,
+      "loss": 0.3394,
+      "step": 6285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2691131498470949e-05,
+      "loss": 0.3136,
+      "step": 6286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.267723102585488e-05,
+      "loss": 0.3157,
+      "step": 6287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2663330553238812e-05,
+      "loss": 0.2827,
+      "step": 6288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2649430080622742e-05,
+      "loss": 0.3067,
+      "step": 6289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2635529608006672e-05,
+      "loss": 0.3344,
+      "step": 6290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2621629135390602e-05,
+      "loss": 0.2934,
+      "step": 6291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2607728662774534e-05,
+      "loss": 0.3145,
+      "step": 6292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2593828190158466e-05,
+      "loss": 0.2723,
+      "step": 6293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2579927717542398e-05,
+      "loss": 0.3201,
+      "step": 6294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.256602724492633e-05,
+      "loss": 0.3007,
+      "step": 6295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2552126772310261e-05,
+      "loss": 0.2899,
+      "step": 6296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.253822629969419e-05,
+      "loss": 0.288,
+      "step": 6297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2524325827078121e-05,
+      "loss": 0.3819,
+      "step": 6298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2510425354462051e-05,
+      "loss": 0.3427,
+      "step": 6299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2496524881845983e-05,
+      "loss": 0.2696,
+      "step": 6300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2482624409229915e-05,
+      "loss": 0.3192,
+      "step": 6301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2468723936613845e-05,
+      "loss": 0.2818,
+      "step": 6302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2454823463997777e-05,
+      "loss": 0.3296,
+      "step": 6303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2440922991381708e-05,
+      "loss": 0.2752,
+      "step": 6304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2427022518765639e-05,
+      "loss": 0.332,
+      "step": 6305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.241312204614957e-05,
+      "loss": 0.2844,
+      "step": 6306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.23992215735335e-05,
+      "loss": 0.2325,
+      "step": 6307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2385321100917432e-05,
+      "loss": 0.3233,
+      "step": 6308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2371420628301362e-05,
+      "loss": 0.3145,
+      "step": 6309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2357520155685294e-05,
+      "loss": 0.3153,
+      "step": 6310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2343619683069226e-05,
+      "loss": 0.2922,
+      "step": 6311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2329719210453156e-05,
+      "loss": 0.3277,
+      "step": 6312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2315818737837087e-05,
+      "loss": 0.2845,
+      "step": 6313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.230191826522102e-05,
+      "loss": 0.3338,
+      "step": 6314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.228801779260495e-05,
+      "loss": 0.3049,
+      "step": 6315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.227411731998888e-05,
+      "loss": 0.3239,
+      "step": 6316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2260216847372811e-05,
+      "loss": 0.2572,
+      "step": 6317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2246316374756741e-05,
+      "loss": 0.3228,
+      "step": 6318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2232415902140673e-05,
+      "loss": 0.3165,
+      "step": 6319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2218515429524605e-05,
+      "loss": 0.3329,
+      "step": 6320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2204614956908536e-05,
+      "loss": 0.2985,
+      "step": 6321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2190714484292467e-05,
+      "loss": 0.2903,
+      "step": 6322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2176814011676398e-05,
+      "loss": 0.3385,
+      "step": 6323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2162913539060328e-05,
+      "loss": 0.3402,
+      "step": 6324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2149013066444258e-05,
+      "loss": 0.32,
+      "step": 6325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.213511259382819e-05,
+      "loss": 0.2711,
+      "step": 6326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.312,
+      "step": 6327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2107311648596052e-05,
+      "loss": 0.275,
+      "step": 6328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2093411175979984e-05,
+      "loss": 0.3036,
+      "step": 6329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2079510703363916e-05,
+      "loss": 0.3503,
+      "step": 6330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2065610230747847e-05,
+      "loss": 0.2639,
+      "step": 6331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2051709758131777e-05,
+      "loss": 0.2774,
+      "step": 6332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2037809285515707e-05,
+      "loss": 0.3114,
+      "step": 6333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.202390881289964e-05,
+      "loss": 0.3512,
+      "step": 6334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.201000834028357e-05,
+      "loss": 0.2943,
+      "step": 6335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1996107867667501e-05,
+      "loss": 0.3218,
+      "step": 6336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1982207395051433e-05,
+      "loss": 0.298,
+      "step": 6337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1968306922435363e-05,
+      "loss": 0.2984,
+      "step": 6338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1954406449819295e-05,
+      "loss": 0.3109,
+      "step": 6339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1940505977203226e-05,
+      "loss": 0.347,
+      "step": 6340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1926605504587156e-05,
+      "loss": 0.2635,
+      "step": 6341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1912705031971086e-05,
+      "loss": 0.3402,
+      "step": 6342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1898804559355018e-05,
+      "loss": 0.282,
+      "step": 6343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.188490408673895e-05,
+      "loss": 0.2974,
+      "step": 6344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.187100361412288e-05,
+      "loss": 0.3352,
+      "step": 6345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1857103141506812e-05,
+      "loss": 0.3153,
+      "step": 6346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1843202668890744e-05,
+      "loss": 0.3022,
+      "step": 6347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1829302196274674e-05,
+      "loss": 0.3098,
+      "step": 6348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1815401723658605e-05,
+      "loss": 0.2693,
+      "step": 6349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1801501251042537e-05,
+      "loss": 0.3117,
+      "step": 6350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1787600778426467e-05,
+      "loss": 0.2936,
+      "step": 6351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1773700305810397e-05,
+      "loss": 0.2977,
+      "step": 6352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1759799833194329e-05,
+      "loss": 0.3098,
+      "step": 6353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.174589936057826e-05,
+      "loss": 0.3158,
+      "step": 6354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.173199888796219e-05,
+      "loss": 0.2869,
+      "step": 6355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1718098415346123e-05,
+      "loss": 0.3556,
+      "step": 6356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1704197942730054e-05,
+      "loss": 0.3098,
+      "step": 6357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1690297470113984e-05,
+      "loss": 0.318,
+      "step": 6358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1676396997497916e-05,
+      "loss": 0.2614,
+      "step": 6359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1662496524881846e-05,
+      "loss": 0.2872,
+      "step": 6360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1648596052265776e-05,
+      "loss": 0.2776,
+      "step": 6361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1634695579649708e-05,
+      "loss": 0.2508,
+      "step": 6362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.162079510703364e-05,
+      "loss": 0.3037,
+      "step": 6363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1606894634417572e-05,
+      "loss": 0.335,
+      "step": 6364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1592994161801502e-05,
+      "loss": 0.2682,
+      "step": 6365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1579093689185433e-05,
+      "loss": 0.3473,
+      "step": 6366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1565193216569365e-05,
+      "loss": 0.3306,
+      "step": 6367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1551292743953295e-05,
+      "loss": 0.3185,
+      "step": 6368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1537392271337225e-05,
+      "loss": 0.3163,
+      "step": 6369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1523491798721157e-05,
+      "loss": 0.2904,
+      "step": 6370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1509591326105087e-05,
+      "loss": 0.2926,
+      "step": 6371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1495690853489019e-05,
+      "loss": 0.3195,
+      "step": 6372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.148179038087295e-05,
+      "loss": 0.2959,
+      "step": 6373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1467889908256882e-05,
+      "loss": 0.2893,
+      "step": 6374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1453989435640812e-05,
+      "loss": 0.3335,
+      "step": 6375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1440088963024744e-05,
+      "loss": 0.2953,
+      "step": 6376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1426188490408674e-05,
+      "loss": 0.3037,
+      "step": 6377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1412288017792604e-05,
+      "loss": 0.2664,
+      "step": 6378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1398387545176536e-05,
+      "loss": 0.298,
+      "step": 6379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1384487072560468e-05,
+      "loss": 0.3051,
+      "step": 6380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1370586599944398e-05,
+      "loss": 0.3315,
+      "step": 6381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.135668612732833e-05,
+      "loss": 0.3152,
+      "step": 6382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1342785654712261e-05,
+      "loss": 0.2483,
+      "step": 6383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1328885182096193e-05,
+      "loss": 0.3044,
+      "step": 6384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1314984709480123e-05,
+      "loss": 0.3283,
+      "step": 6385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1301084236864053e-05,
+      "loss": 0.3083,
+      "step": 6386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1287183764247985e-05,
+      "loss": 0.3056,
+      "step": 6387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1273283291631915e-05,
+      "loss": 0.2907,
+      "step": 6388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1259382819015847e-05,
+      "loss": 0.3248,
+      "step": 6389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1245482346399779e-05,
+      "loss": 0.2809,
+      "step": 6390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1231581873783709e-05,
+      "loss": 0.2952,
+      "step": 6391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.121768140116764e-05,
+      "loss": 0.2819,
+      "step": 6392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1203780928551572e-05,
+      "loss": 0.2914,
+      "step": 6393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1189880455935502e-05,
+      "loss": 0.2626,
+      "step": 6394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1175979983319432e-05,
+      "loss": 0.3185,
+      "step": 6395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1162079510703364e-05,
+      "loss": 0.3103,
+      "step": 6396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1148179038087296e-05,
+      "loss": 0.2823,
+      "step": 6397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1134278565471226e-05,
+      "loss": 0.3103,
+      "step": 6398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1120378092855158e-05,
+      "loss": 0.3176,
+      "step": 6399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.110647762023909e-05,
+      "loss": 0.3289,
+      "step": 6400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.109257714762302e-05,
+      "loss": 0.3322,
+      "step": 6401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1078676675006951e-05,
+      "loss": 0.3336,
+      "step": 6402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1064776202390881e-05,
+      "loss": 0.3068,
+      "step": 6403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1050875729774811e-05,
+      "loss": 0.3563,
+      "step": 6404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1036975257158743e-05,
+      "loss": 0.284,
+      "step": 6405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1023074784542675e-05,
+      "loss": 0.3074,
+      "step": 6406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 0.3083,
+      "step": 6407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0995273839310537e-05,
+      "loss": 0.2635,
+      "step": 6408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0981373366694469e-05,
+      "loss": 0.3151,
+      "step": 6409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.09674728940784e-05,
+      "loss": 0.285,
+      "step": 6410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.095357242146233e-05,
+      "loss": 0.3528,
+      "step": 6411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.093967194884626e-05,
+      "loss": 0.3127,
+      "step": 6412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0925771476230192e-05,
+      "loss": 0.3002,
+      "step": 6413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0911871003614122e-05,
+      "loss": 0.2796,
+      "step": 6414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0897970530998054e-05,
+      "loss": 0.3165,
+      "step": 6415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0884070058381986e-05,
+      "loss": 0.3068,
+      "step": 6416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0870169585765917e-05,
+      "loss": 0.2996,
+      "step": 6417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0856269113149848e-05,
+      "loss": 0.339,
+      "step": 6418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.084236864053378e-05,
+      "loss": 0.3087,
+      "step": 6419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0828468167917711e-05,
+      "loss": 0.283,
+      "step": 6420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0814567695301641e-05,
+      "loss": 0.3185,
+      "step": 6421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0800667222685571e-05,
+      "loss": 0.2684,
+      "step": 6422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0786766750069503e-05,
+      "loss": 0.3318,
+      "step": 6423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0772866277453433e-05,
+      "loss": 0.3323,
+      "step": 6424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0758965804837365e-05,
+      "loss": 0.3127,
+      "step": 6425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0745065332221297e-05,
+      "loss": 0.3081,
+      "step": 6426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0731164859605228e-05,
+      "loss": 0.2599,
+      "step": 6427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0717264386989158e-05,
+      "loss": 0.3106,
+      "step": 6428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.070336391437309e-05,
+      "loss": 0.3173,
+      "step": 6429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.068946344175702e-05,
+      "loss": 0.2823,
+      "step": 6430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.067556296914095e-05,
+      "loss": 0.2886,
+      "step": 6431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0661662496524882e-05,
+      "loss": 0.2888,
+      "step": 6432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0647762023908814e-05,
+      "loss": 0.3296,
+      "step": 6433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0633861551292744e-05,
+      "loss": 0.2902,
+      "step": 6434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0619961078676676e-05,
+      "loss": 0.3019,
+      "step": 6435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0606060606060607e-05,
+      "loss": 0.2975,
+      "step": 6436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0592160133444537e-05,
+      "loss": 0.2776,
+      "step": 6437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.057825966082847e-05,
+      "loss": 0.3233,
+      "step": 6438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.05643591882124e-05,
+      "loss": 0.29,
+      "step": 6439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0550458715596331e-05,
+      "loss": 0.3151,
+      "step": 6440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0536558242980261e-05,
+      "loss": 0.2974,
+      "step": 6441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0522657770364193e-05,
+      "loss": 0.3149,
+      "step": 6442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0508757297748125e-05,
+      "loss": 0.305,
+      "step": 6443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0494856825132055e-05,
+      "loss": 0.3051,
+      "step": 6444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0480956352515986e-05,
+      "loss": 0.3217,
+      "step": 6445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0467055879899918e-05,
+      "loss": 0.2809,
+      "step": 6446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0453155407283848e-05,
+      "loss": 0.384,
+      "step": 6447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0439254934667778e-05,
+      "loss": 0.2815,
+      "step": 6448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.042535446205171e-05,
+      "loss": 0.3143,
+      "step": 6449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0411453989435642e-05,
+      "loss": 0.3106,
+      "step": 6450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0397553516819572e-05,
+      "loss": 0.2669,
+      "step": 6451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0383653044203504e-05,
+      "loss": 0.3071,
+      "step": 6452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0369752571587435e-05,
+      "loss": 0.3373,
+      "step": 6453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0355852098971365e-05,
+      "loss": 0.3067,
+      "step": 6454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0341951626355297e-05,
+      "loss": 0.3021,
+      "step": 6455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0328051153739227e-05,
+      "loss": 0.2864,
+      "step": 6456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0314150681123157e-05,
+      "loss": 0.2983,
+      "step": 6457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0300250208507089e-05,
+      "loss": 0.3841,
+      "step": 6458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.028634973589102e-05,
+      "loss": 0.2795,
+      "step": 6459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0272449263274953e-05,
+      "loss": 0.3074,
+      "step": 6460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0258548790658883e-05,
+      "loss": 0.2883,
+      "step": 6461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0244648318042814e-05,
+      "loss": 0.2947,
+      "step": 6462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0230747845426746e-05,
+      "loss": 0.2712,
+      "step": 6463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0216847372810676e-05,
+      "loss": 0.2845,
+      "step": 6464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0202946900194606e-05,
+      "loss": 0.368,
+      "step": 6465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0189046427578538e-05,
+      "loss": 0.3068,
+      "step": 6466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0175145954962468e-05,
+      "loss": 0.2769,
+      "step": 6467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.01612454823464e-05,
+      "loss": 0.3048,
+      "step": 6468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0147345009730332e-05,
+      "loss": 0.339,
+      "step": 6469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0133444537114263e-05,
+      "loss": 0.3141,
+      "step": 6470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0119544064498193e-05,
+      "loss": 0.3406,
+      "step": 6471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0105643591882125e-05,
+      "loss": 0.2722,
+      "step": 6472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0091743119266055e-05,
+      "loss": 0.3308,
+      "step": 6473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0077842646649985e-05,
+      "loss": 0.2918,
+      "step": 6474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0063942174033917e-05,
+      "loss": 0.3231,
+      "step": 6475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0050041701417849e-05,
+      "loss": 0.3026,
+      "step": 6476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0036141228801779e-05,
+      "loss": 0.3772,
+      "step": 6477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.002224075618571e-05,
+      "loss": 0.3112,
+      "step": 6478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0008340283569642e-05,
+      "loss": 0.3217,
+      "step": 6479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.994439810953573e-06,
+      "loss": 0.2677,
+      "step": 6480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.980539338337504e-06,
+      "loss": 0.2732,
+      "step": 6481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.966638865721436e-06,
+      "loss": 0.2647,
+      "step": 6482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.952738393105366e-06,
+      "loss": 0.2835,
+      "step": 6483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.938837920489296e-06,
+      "loss": 0.3043,
+      "step": 6484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.924937447873228e-06,
+      "loss": 0.2952,
+      "step": 6485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.91103697525716e-06,
+      "loss": 0.3436,
+      "step": 6486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.89713650264109e-06,
+      "loss": 0.2914,
+      "step": 6487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.883236030025021e-06,
+      "loss": 0.318,
+      "step": 6488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.869335557408953e-06,
+      "loss": 0.2889,
+      "step": 6489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.855435084792883e-06,
+      "loss": 0.361,
+      "step": 6490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.841534612176815e-06,
+      "loss": 0.3037,
+      "step": 6491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.827634139560745e-06,
+      "loss": 0.3135,
+      "step": 6492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.813733666944677e-06,
+      "loss": 0.2642,
+      "step": 6493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.799833194328607e-06,
+      "loss": 0.3704,
+      "step": 6494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.785932721712539e-06,
+      "loss": 0.345,
+      "step": 6495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.77203224909647e-06,
+      "loss": 0.3482,
+      "step": 6496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.7581317764804e-06,
+      "loss": 0.3162,
+      "step": 6497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.744231303864332e-06,
+      "loss": 0.2972,
+      "step": 6498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.730330831248264e-06,
+      "loss": 0.2951,
+      "step": 6499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.716430358632194e-06,
+      "loss": 0.2378,
+      "step": 6500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.702529886016124e-06,
+      "loss": 0.3695,
+      "step": 6501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.688629413400056e-06,
+      "loss": 0.2795,
+      "step": 6502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.674728940783988e-06,
+      "loss": 0.3037,
+      "step": 6503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.660828468167918e-06,
+      "loss": 0.277,
+      "step": 6504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.64692799555185e-06,
+      "loss": 0.2825,
+      "step": 6505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.633027522935781e-06,
+      "loss": 0.2371,
+      "step": 6506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.619127050319711e-06,
+      "loss": 0.2889,
+      "step": 6507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.605226577703643e-06,
+      "loss": 0.2463,
+      "step": 6508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.591326105087573e-06,
+      "loss": 0.3318,
+      "step": 6509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.577425632471503e-06,
+      "loss": 0.2896,
+      "step": 6510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.563525159855435e-06,
+      "loss": 0.2679,
+      "step": 6511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.549624687239367e-06,
+      "loss": 0.2873,
+      "step": 6512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.535724214623297e-06,
+      "loss": 0.2641,
+      "step": 6513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.521823742007229e-06,
+      "loss": 0.3306,
+      "step": 6514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.50792326939116e-06,
+      "loss": 0.2694,
+      "step": 6515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.494022796775092e-06,
+      "loss": 0.287,
+      "step": 6516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.480122324159022e-06,
+      "loss": 0.3596,
+      "step": 6517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.466221851542952e-06,
+      "loss": 0.3067,
+      "step": 6518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.452321378926884e-06,
+      "loss": 0.2958,
+      "step": 6519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.438420906310814e-06,
+      "loss": 0.3261,
+      "step": 6520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.424520433694746e-06,
+      "loss": 0.3281,
+      "step": 6521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.410619961078678e-06,
+      "loss": 0.2747,
+      "step": 6522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.396719488462608e-06,
+      "loss": 0.3118,
+      "step": 6523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.38281901584654e-06,
+      "loss": 0.3635,
+      "step": 6524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.368918543230471e-06,
+      "loss": 0.259,
+      "step": 6525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.355018070614401e-06,
+      "loss": 0.2776,
+      "step": 6526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.341117597998331e-06,
+      "loss": 0.2729,
+      "step": 6527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.327217125382263e-06,
+      "loss": 0.3033,
+      "step": 6528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.313316652766195e-06,
+      "loss": 0.3086,
+      "step": 6529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.299416180150125e-06,
+      "loss": 0.2936,
+      "step": 6530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.285515707534057e-06,
+      "loss": 0.2943,
+      "step": 6531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.271615234917988e-06,
+      "loss": 0.3546,
+      "step": 6532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.257714762301918e-06,
+      "loss": 0.2893,
+      "step": 6533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.24381428968585e-06,
+      "loss": 0.2822,
+      "step": 6534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.22991381706978e-06,
+      "loss": 0.3135,
+      "step": 6535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.216013344453712e-06,
+      "loss": 0.2511,
+      "step": 6536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.202112871837642e-06,
+      "loss": 0.3253,
+      "step": 6537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.188212399221574e-06,
+      "loss": 0.2987,
+      "step": 6538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 0.2874,
+      "step": 6539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.160411453989436e-06,
+      "loss": 0.3121,
+      "step": 6540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.146510981373367e-06,
+      "loss": 0.3222,
+      "step": 6541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.132610508757299e-06,
+      "loss": 0.2561,
+      "step": 6542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.11871003614123e-06,
+      "loss": 0.312,
+      "step": 6543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.10480956352516e-06,
+      "loss": 0.3,
+      "step": 6544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.3017,
+      "step": 6545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.077008618293023e-06,
+      "loss": 0.3097,
+      "step": 6546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.063108145676953e-06,
+      "loss": 0.3001,
+      "step": 6547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.049207673060885e-06,
+      "loss": 0.3104,
+      "step": 6548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.035307200444816e-06,
+      "loss": 0.301,
+      "step": 6549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.021406727828746e-06,
+      "loss": 0.3267,
+      "step": 6550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.007506255212678e-06,
+      "loss": 0.3188,
+      "step": 6551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.99360578259661e-06,
+      "loss": 0.3357,
+      "step": 6552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.97970530998054e-06,
+      "loss": 0.2717,
+      "step": 6553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.96580483736447e-06,
+      "loss": 0.3409,
+      "step": 6554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.951904364748402e-06,
+      "loss": 0.3546,
+      "step": 6555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.938003892132332e-06,
+      "loss": 0.2565,
+      "step": 6556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.924103419516264e-06,
+      "loss": 0.2876,
+      "step": 6557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.910202946900195e-06,
+      "loss": 0.3413,
+      "step": 6558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.896302474284127e-06,
+      "loss": 0.2879,
+      "step": 6559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.882402001668057e-06,
+      "loss": 0.372,
+      "step": 6560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.868501529051989e-06,
+      "loss": 0.3093,
+      "step": 6561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.854601056435919e-06,
+      "loss": 0.3244,
+      "step": 6562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.840700583819849e-06,
+      "loss": 0.2401,
+      "step": 6563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.826800111203781e-06,
+      "loss": 0.273,
+      "step": 6564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.812899638587713e-06,
+      "loss": 0.2957,
+      "step": 6565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.798999165971643e-06,
+      "loss": 0.308,
+      "step": 6566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.785098693355574e-06,
+      "loss": 0.3522,
+      "step": 6567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.771198220739506e-06,
+      "loss": 0.3282,
+      "step": 6568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.757297748123438e-06,
+      "loss": 0.3155,
+      "step": 6569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.743397275507368e-06,
+      "loss": 0.2739,
+      "step": 6570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.729496802891298e-06,
+      "loss": 0.3498,
+      "step": 6571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.71559633027523e-06,
+      "loss": 0.2685,
+      "step": 6572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.70169585765916e-06,
+      "loss": 0.2872,
+      "step": 6573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.687795385043092e-06,
+      "loss": 0.3071,
+      "step": 6574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.673894912427023e-06,
+      "loss": 0.2768,
+      "step": 6575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.659994439810954e-06,
+      "loss": 0.3197,
+      "step": 6576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.646093967194885e-06,
+      "loss": 0.2738,
+      "step": 6577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.632193494578817e-06,
+      "loss": 0.2745,
+      "step": 6578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.618293021962747e-06,
+      "loss": 0.2448,
+      "step": 6579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.604392549346677e-06,
+      "loss": 0.3312,
+      "step": 6580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.590492076730609e-06,
+      "loss": 0.3381,
+      "step": 6581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.57659160411454e-06,
+      "loss": 0.3171,
+      "step": 6582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.56269113149847e-06,
+      "loss": 0.2984,
+      "step": 6583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.548790658882402e-06,
+      "loss": 0.3083,
+      "step": 6584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.534890186266334e-06,
+      "loss": 0.2496,
+      "step": 6585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.520989713650264e-06,
+      "loss": 0.3318,
+      "step": 6586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.507089241034196e-06,
+      "loss": 0.3069,
+      "step": 6587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.493188768418126e-06,
+      "loss": 0.2743,
+      "step": 6588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.479288295802058e-06,
+      "loss": 0.333,
+      "step": 6589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.465387823185988e-06,
+      "loss": 0.3366,
+      "step": 6590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.45148735056992e-06,
+      "loss": 0.3584,
+      "step": 6591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.437586877953851e-06,
+      "loss": 0.3455,
+      "step": 6592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.423686405337782e-06,
+      "loss": 0.3066,
+      "step": 6593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.409785932721713e-06,
+      "loss": 0.2979,
+      "step": 6594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.395885460105645e-06,
+      "loss": 0.2857,
+      "step": 6595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.381984987489575e-06,
+      "loss": 0.2817,
+      "step": 6596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.368084514873505e-06,
+      "loss": 0.3047,
+      "step": 6597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.354184042257437e-06,
+      "loss": 0.3089,
+      "step": 6598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.340283569641367e-06,
+      "loss": 0.3105,
+      "step": 6599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.326383097025299e-06,
+      "loss": 0.3061,
+      "step": 6600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.31248262440923e-06,
+      "loss": 0.3493,
+      "step": 6601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.298582151793162e-06,
+      "loss": 0.3081,
+      "step": 6602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.284681679177092e-06,
+      "loss": 0.3028,
+      "step": 6603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.270781206561024e-06,
+      "loss": 0.3198,
+      "step": 6604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.256880733944954e-06,
+      "loss": 0.2879,
+      "step": 6605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.242980261328884e-06,
+      "loss": 0.3151,
+      "step": 6606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.229079788712816e-06,
+      "loss": 0.3061,
+      "step": 6607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.215179316096748e-06,
+      "loss": 0.2623,
+      "step": 6608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.201278843480678e-06,
+      "loss": 0.321,
+      "step": 6609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.18737837086461e-06,
+      "loss": 0.2909,
+      "step": 6610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.173477898248541e-06,
+      "loss": 0.2866,
+      "step": 6611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.159577425632473e-06,
+      "loss": 0.2996,
+      "step": 6612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.145676953016403e-06,
+      "loss": 0.298,
+      "step": 6613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.131776480400335e-06,
+      "loss": 0.2514,
+      "step": 6614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.117876007784265e-06,
+      "loss": 0.2649,
+      "step": 6615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.103975535168195e-06,
+      "loss": 0.2921,
+      "step": 6616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.090075062552127e-06,
+      "loss": 0.3036,
+      "step": 6617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.076174589936059e-06,
+      "loss": 0.3521,
+      "step": 6618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.062274117319989e-06,
+      "loss": 0.2897,
+      "step": 6619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.04837364470392e-06,
+      "loss": 0.3355,
+      "step": 6620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.034473172087852e-06,
+      "loss": 0.2846,
+      "step": 6621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.020572699471784e-06,
+      "loss": 0.2817,
+      "step": 6622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.006672226855714e-06,
+      "loss": 0.2421,
+      "step": 6623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.992771754239644e-06,
+      "loss": 0.2895,
+      "step": 6624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.978871281623576e-06,
+      "loss": 0.3377,
+      "step": 6625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.964970809007506e-06,
+      "loss": 0.2952,
+      "step": 6626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.951070336391438e-06,
+      "loss": 0.3543,
+      "step": 6627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.93716986377537e-06,
+      "loss": 0.316,
+      "step": 6628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.9232693911593e-06,
+      "loss": 0.3042,
+      "step": 6629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.909368918543231e-06,
+      "loss": 0.2939,
+      "step": 6630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.895468445927163e-06,
+      "loss": 0.3102,
+      "step": 6631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.881567973311093e-06,
+      "loss": 0.2976,
+      "step": 6632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.867667500695023e-06,
+      "loss": 0.3465,
+      "step": 6633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.853767028078955e-06,
+      "loss": 0.3117,
+      "step": 6634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.839866555462887e-06,
+      "loss": 0.3318,
+      "step": 6635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.825966082846817e-06,
+      "loss": 0.2746,
+      "step": 6636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.812065610230748e-06,
+      "loss": 0.3213,
+      "step": 6637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.79816513761468e-06,
+      "loss": 0.3254,
+      "step": 6638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.78426466499861e-06,
+      "loss": 0.3022,
+      "step": 6639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.770364192382542e-06,
+      "loss": 0.3337,
+      "step": 6640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.756463719766472e-06,
+      "loss": 0.2868,
+      "step": 6641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.742563247150402e-06,
+      "loss": 0.297,
+      "step": 6642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.728662774534334e-06,
+      "loss": 0.3248,
+      "step": 6643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.714762301918266e-06,
+      "loss": 0.3179,
+      "step": 6644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.700861829302197e-06,
+      "loss": 0.324,
+      "step": 6645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.686961356686127e-06,
+      "loss": 0.2868,
+      "step": 6646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.67306088407006e-06,
+      "loss": 0.2925,
+      "step": 6647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.659160411453991e-06,
+      "loss": 0.2475,
+      "step": 6648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.645259938837921e-06,
+      "loss": 0.3807,
+      "step": 6649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.631359466221851e-06,
+      "loss": 0.3104,
+      "step": 6650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.617458993605784e-06,
+      "loss": 0.2757,
+      "step": 6651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.603558520989714e-06,
+      "loss": 0.302,
+      "step": 6652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.589658048373645e-06,
+      "loss": 0.2831,
+      "step": 6653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.3098,
+      "step": 6654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.561857103141508e-06,
+      "loss": 0.2653,
+      "step": 6655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.547956630525438e-06,
+      "loss": 0.2942,
+      "step": 6656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.534056157909369e-06,
+      "loss": 0.3086,
+      "step": 6657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.520155685293301e-06,
+      "loss": 0.2959,
+      "step": 6658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.506255212677231e-06,
+      "loss": 0.2965,
+      "step": 6659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.492354740061163e-06,
+      "loss": 0.285,
+      "step": 6660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.478454267445094e-06,
+      "loss": 0.2824,
+      "step": 6661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.464553794829024e-06,
+      "loss": 0.3316,
+      "step": 6662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.4506533222129555e-06,
+      "loss": 0.2772,
+      "step": 6663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.436752849596887e-06,
+      "loss": 0.3331,
+      "step": 6664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.422852376980818e-06,
+      "loss": 0.3236,
+      "step": 6665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.408951904364748e-06,
+      "loss": 0.3092,
+      "step": 6666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.39505143174868e-06,
+      "loss": 0.2852,
+      "step": 6667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.381150959132612e-06,
+      "loss": 0.2567,
+      "step": 6668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.367250486516542e-06,
+      "loss": 0.3209,
+      "step": 6669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.353350013900473e-06,
+      "loss": 0.2637,
+      "step": 6670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.3394495412844045e-06,
+      "loss": 0.336,
+      "step": 6671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.3255490686683345e-06,
+      "loss": 0.2723,
+      "step": 6672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.311648596052266e-06,
+      "loss": 0.3189,
+      "step": 6673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.297748123436197e-06,
+      "loss": 0.3311,
+      "step": 6674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.283847650820127e-06,
+      "loss": 0.3131,
+      "step": 6675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.269947178204059e-06,
+      "loss": 0.2931,
+      "step": 6676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.256046705587991e-06,
+      "loss": 0.2638,
+      "step": 6677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.242146232971922e-06,
+      "loss": 0.3529,
+      "step": 6678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.228245760355852e-06,
+      "loss": 0.2902,
+      "step": 6679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.2143452877397835e-06,
+      "loss": 0.3099,
+      "step": 6680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.200444815123715e-06,
+      "loss": 0.3145,
+      "step": 6681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.186544342507645e-06,
+      "loss": 0.3264,
+      "step": 6682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.172643869891576e-06,
+      "loss": 0.3282,
+      "step": 6683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.158743397275508e-06,
+      "loss": 0.2988,
+      "step": 6684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.144842924659438e-06,
+      "loss": 0.2624,
+      "step": 6685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.13094245204337e-06,
+      "loss": 0.2694,
+      "step": 6686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.117041979427301e-06,
+      "loss": 0.3138,
+      "step": 6687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.1031415068112325e-06,
+      "loss": 0.3173,
+      "step": 6688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.0892410341951625e-06,
+      "loss": 0.2741,
+      "step": 6689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.075340561579094e-06,
+      "loss": 0.2846,
+      "step": 6690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.061440088963025e-06,
+      "loss": 0.3075,
+      "step": 6691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.047539616346955e-06,
+      "loss": 0.3113,
+      "step": 6692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.033639143730887e-06,
+      "loss": 0.3362,
+      "step": 6693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.019738671114819e-06,
+      "loss": 0.3412,
+      "step": 6694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.005838198498749e-06,
+      "loss": 0.3003,
+      "step": 6695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.99193772588268e-06,
+      "loss": 0.2926,
+      "step": 6696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.9780372532666115e-06,
+      "loss": 0.323,
+      "step": 6697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.964136780650543e-06,
+      "loss": 0.2958,
+      "step": 6698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.950236308034473e-06,
+      "loss": 0.2742,
+      "step": 6699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.936335835418404e-06,
+      "loss": 0.3082,
+      "step": 6700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.922435362802336e-06,
+      "loss": 0.2584,
+      "step": 6701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.908534890186266e-06,
+      "loss": 0.3178,
+      "step": 6702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.894634417570198e-06,
+      "loss": 0.2786,
+      "step": 6703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.880733944954129e-06,
+      "loss": 0.3159,
+      "step": 6704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.866833472338059e-06,
+      "loss": 0.3131,
+      "step": 6705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.852932999721991e-06,
+      "loss": 0.3476,
+      "step": 6706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.839032527105922e-06,
+      "loss": 0.3229,
+      "step": 6707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.825132054489853e-06,
+      "loss": 0.3191,
+      "step": 6708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.811231581873784e-06,
+      "loss": 0.3245,
+      "step": 6709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.797331109257715e-06,
+      "loss": 0.321,
+      "step": 6710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.783430636641647e-06,
+      "loss": 0.3098,
+      "step": 6711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.769530164025577e-06,
+      "loss": 0.2776,
+      "step": 6712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.755629691409509e-06,
+      "loss": 0.2613,
+      "step": 6713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.7417292187934396e-06,
+      "loss": 0.3038,
+      "step": 6714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.72782874617737e-06,
+      "loss": 0.2564,
+      "step": 6715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.713928273561301e-06,
+      "loss": 0.2726,
+      "step": 6716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.700027800945233e-06,
+      "loss": 0.3163,
+      "step": 6717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.686127328329163e-06,
+      "loss": 0.3154,
+      "step": 6718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.672226855713094e-06,
+      "loss": 0.2735,
+      "step": 6719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.658326383097026e-06,
+      "loss": 0.3103,
+      "step": 6720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.644425910480958e-06,
+      "loss": 0.3336,
+      "step": 6721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.630525437864888e-06,
+      "loss": 0.3071,
+      "step": 6722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.616624965248819e-06,
+      "loss": 0.3138,
+      "step": 6723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.60272449263275e-06,
+      "loss": 0.3307,
+      "step": 6724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.5888240200166804e-06,
+      "loss": 0.2962,
+      "step": 6725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.574923547400612e-06,
+      "loss": 0.3241,
+      "step": 6726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.561023074784543e-06,
+      "loss": 0.2431,
+      "step": 6727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.547122602168473e-06,
+      "loss": 0.2789,
+      "step": 6728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.533222129552405e-06,
+      "loss": 0.2818,
+      "step": 6729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.519321656936337e-06,
+      "loss": 0.3061,
+      "step": 6730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.505421184320268e-06,
+      "loss": 0.3132,
+      "step": 6731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.491520711704198e-06,
+      "loss": 0.3033,
+      "step": 6732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.477620239088129e-06,
+      "loss": 0.28,
+      "step": 6733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.463719766472061e-06,
+      "loss": 0.3199,
+      "step": 6734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.449819293855991e-06,
+      "loss": 0.3478,
+      "step": 6735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.435918821239922e-06,
+      "loss": 0.2956,
+      "step": 6736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.422018348623854e-06,
+      "loss": 0.3018,
+      "step": 6737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.408117876007784e-06,
+      "loss": 0.2939,
+      "step": 6738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.394217403391716e-06,
+      "loss": 0.2592,
+      "step": 6739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.380316930775647e-06,
+      "loss": 0.3263,
+      "step": 6740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.366416458159578e-06,
+      "loss": 0.3319,
+      "step": 6741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.3525159855435085e-06,
+      "loss": 0.3118,
+      "step": 6742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.33861551292744e-06,
+      "loss": 0.2756,
+      "step": 6743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.324715040311371e-06,
+      "loss": 0.2797,
+      "step": 6744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.310814567695301e-06,
+      "loss": 0.3803,
+      "step": 6745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.296914095079233e-06,
+      "loss": 0.2973,
+      "step": 6746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.283013622463165e-06,
+      "loss": 0.327,
+      "step": 6747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.269113149847095e-06,
+      "loss": 0.2467,
+      "step": 6748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.255212677231026e-06,
+      "loss": 0.3821,
+      "step": 6749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.2413122046149574e-06,
+      "loss": 0.2801,
+      "step": 6750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.227411731998888e-06,
+      "loss": 0.2363,
+      "step": 6751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.213511259382819e-06,
+      "loss": 0.3602,
+      "step": 6752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.19961078676675e-06,
+      "loss": 0.3396,
+      "step": 6753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.185710314150681e-06,
+      "loss": 0.2913,
+      "step": 6754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.171809841534613e-06,
+      "loss": 0.3159,
+      "step": 6755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.157909368918544e-06,
+      "loss": 0.2704,
+      "step": 6756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.144008896302475e-06,
+      "loss": 0.3203,
+      "step": 6757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.1301084236864056e-06,
+      "loss": 0.3494,
+      "step": 6758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.1162079510703365e-06,
+      "loss": 0.2549,
+      "step": 6759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.102307478454268e-06,
+      "loss": 0.3234,
+      "step": 6760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.088407005838199e-06,
+      "loss": 0.2871,
+      "step": 6761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.074506533222129e-06,
+      "loss": 0.2863,
+      "step": 6762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.2829,
+      "step": 6763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.046705587989992e-06,
+      "loss": 0.3029,
+      "step": 6764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.032805115373924e-06,
+      "loss": 0.2658,
+      "step": 6765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.018904642757854e-06,
+      "loss": 0.3224,
+      "step": 6766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.005004170141785e-06,
+      "loss": 0.3223,
+      "step": 6767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.991103697525716e-06,
+      "loss": 0.2698,
+      "step": 6768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.977203224909647e-06,
+      "loss": 0.305,
+      "step": 6769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.963302752293578e-06,
+      "loss": 0.2877,
+      "step": 6770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.949402279677509e-06,
+      "loss": 0.3156,
+      "step": 6771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.93550180706144e-06,
+      "loss": 0.3084,
+      "step": 6772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.921601334445372e-06,
+      "loss": 0.3128,
+      "step": 6773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.907700861829303e-06,
+      "loss": 0.2713,
+      "step": 6774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.893800389213234e-06,
+      "loss": 0.2879,
+      "step": 6775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8798999165971645e-06,
+      "loss": 0.3301,
+      "step": 6776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.865999443981095e-06,
+      "loss": 0.2812,
+      "step": 6777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.852098971365027e-06,
+      "loss": 0.3562,
+      "step": 6778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.838198498748958e-06,
+      "loss": 0.2618,
+      "step": 6779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.824298026132888e-06,
+      "loss": 0.3034,
+      "step": 6780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.81039755351682e-06,
+      "loss": 0.2937,
+      "step": 6781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.796497080900751e-06,
+      "loss": 0.3216,
+      "step": 6782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.782596608284683e-06,
+      "loss": 0.3037,
+      "step": 6783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.768696135668613e-06,
+      "loss": 0.3042,
+      "step": 6784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.7547956630525436e-06,
+      "loss": 0.312,
+      "step": 6785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.740895190436475e-06,
+      "loss": 0.3214,
+      "step": 6786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.726994717820406e-06,
+      "loss": 0.3652,
+      "step": 6787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.713094245204337e-06,
+      "loss": 0.3239,
+      "step": 6788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.699193772588268e-06,
+      "loss": 0.2764,
+      "step": 6789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.685293299972199e-06,
+      "loss": 0.3274,
+      "step": 6790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.671392827356131e-06,
+      "loss": 0.2526,
+      "step": 6791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.657492354740062e-06,
+      "loss": 0.2779,
+      "step": 6792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.6435918821239925e-06,
+      "loss": 0.304,
+      "step": 6793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.6296914095079234e-06,
+      "loss": 0.2498,
+      "step": 6794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.615790936891854e-06,
+      "loss": 0.3492,
+      "step": 6795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.601890464275786e-06,
+      "loss": 0.3451,
+      "step": 6796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.587989991659716e-06,
+      "loss": 0.268,
+      "step": 6797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.574089519043648e-06,
+      "loss": 0.3255,
+      "step": 6798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.560189046427579e-06,
+      "loss": 0.3459,
+      "step": 6799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.54628857381151e-06,
+      "loss": 0.2994,
+      "step": 6800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.532388101195441e-06,
+      "loss": 0.2701,
+      "step": 6801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.518487628579372e-06,
+      "loss": 0.2737,
+      "step": 6802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 0.2768,
+      "step": 6803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.490686683347234e-06,
+      "loss": 0.3105,
+      "step": 6804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.476786210731165e-06,
+      "loss": 0.3385,
+      "step": 6805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.462885738115096e-06,
+      "loss": 0.2619,
+      "step": 6806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.448985265499027e-06,
+      "loss": 0.3372,
+      "step": 6807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.435084792882959e-06,
+      "loss": 0.2834,
+      "step": 6808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.42118432026689e-06,
+      "loss": 0.3276,
+      "step": 6809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4072838476508206e-06,
+      "loss": 0.3016,
+      "step": 6810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3933833750347515e-06,
+      "loss": 0.3569,
+      "step": 6811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.379482902418682e-06,
+      "loss": 0.3027,
+      "step": 6812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.365582429802614e-06,
+      "loss": 0.303,
+      "step": 6813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.351681957186545e-06,
+      "loss": 0.3057,
+      "step": 6814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.337781484570475e-06,
+      "loss": 0.2596,
+      "step": 6815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.323881011954407e-06,
+      "loss": 0.258,
+      "step": 6816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.309980539338338e-06,
+      "loss": 0.2815,
+      "step": 6817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.296080066722269e-06,
+      "loss": 0.3232,
+      "step": 6818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2821795941062e-06,
+      "loss": 0.3094,
+      "step": 6819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2682791214901305e-06,
+      "loss": 0.2995,
+      "step": 6820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.254378648874062e-06,
+      "loss": 0.3053,
+      "step": 6821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.240478176257993e-06,
+      "loss": 0.2652,
+      "step": 6822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.226577703641924e-06,
+      "loss": 0.2609,
+      "step": 6823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.212677231025855e-06,
+      "loss": 0.314,
+      "step": 6824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.198776758409786e-06,
+      "loss": 0.2648,
+      "step": 6825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.184876285793718e-06,
+      "loss": 0.2637,
+      "step": 6826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.170975813177649e-06,
+      "loss": 0.2774,
+      "step": 6827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.157075340561579e-06,
+      "loss": 0.3434,
+      "step": 6828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.14317486794551e-06,
+      "loss": 0.3206,
+      "step": 6829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.129274395329441e-06,
+      "loss": 0.3182,
+      "step": 6830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.115373922713373e-06,
+      "loss": 0.2994,
+      "step": 6831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.101473450097303e-06,
+      "loss": 0.2678,
+      "step": 6832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.087572977481234e-06,
+      "loss": 0.3497,
+      "step": 6833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.073672504865166e-06,
+      "loss": 0.3704,
+      "step": 6834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.059772032249097e-06,
+      "loss": 0.3108,
+      "step": 6835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.045871559633028e-06,
+      "loss": 0.309,
+      "step": 6836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.0319710870169586e-06,
+      "loss": 0.3951,
+      "step": 6837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.0180706144008895e-06,
+      "loss": 0.3038,
+      "step": 6838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.004170141784821e-06,
+      "loss": 0.314,
+      "step": 6839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.990269669168752e-06,
+      "loss": 0.2698,
+      "step": 6840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.976369196552683e-06,
+      "loss": 0.2893,
+      "step": 6841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.962468723936614e-06,
+      "loss": 0.3235,
+      "step": 6842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.948568251320545e-06,
+      "loss": 0.3167,
+      "step": 6843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.934667778704477e-06,
+      "loss": 0.3411,
+      "step": 6844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9207673060884075e-06,
+      "loss": 0.3392,
+      "step": 6845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9068668334723384e-06,
+      "loss": 0.2847,
+      "step": 6846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.892966360856269e-06,
+      "loss": 0.3416,
+      "step": 6847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8790658882402e-06,
+      "loss": 0.316,
+      "step": 6848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.865165415624132e-06,
+      "loss": 0.311,
+      "step": 6849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.851264943008062e-06,
+      "loss": 0.2773,
+      "step": 6850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.837364470391994e-06,
+      "loss": 0.2758,
+      "step": 6851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.823463997775925e-06,
+      "loss": 0.2955,
+      "step": 6852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.809563525159856e-06,
+      "loss": 0.282,
+      "step": 6853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.795663052543787e-06,
+      "loss": 0.2843,
+      "step": 6854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.7817625799277175e-06,
+      "loss": 0.2647,
+      "step": 6855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.767862107311648e-06,
+      "loss": 0.3618,
+      "step": 6856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.75396163469558e-06,
+      "loss": 0.3227,
+      "step": 6857
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.740061162079511e-06,
+      "loss": 0.2333,
+      "step": 6858
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.726160689463442e-06,
+      "loss": 0.3106,
+      "step": 6859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.712260216847373e-06,
+      "loss": 0.3047,
+      "step": 6860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.698359744231304e-06,
+      "loss": 0.2811,
+      "step": 6861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6844592716152356e-06,
+      "loss": 0.3286,
+      "step": 6862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670558798999166e-06,
+      "loss": 0.2862,
+      "step": 6863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.656658326383097e-06,
+      "loss": 0.3233,
+      "step": 6864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.642757853767028e-06,
+      "loss": 0.3198,
+      "step": 6865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.628857381150959e-06,
+      "loss": 0.3128,
+      "step": 6866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.61495690853489e-06,
+      "loss": 0.3889,
+      "step": 6867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.601056435918821e-06,
+      "loss": 0.3408,
+      "step": 6868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 0.293,
+      "step": 6869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.573255490686684e-06,
+      "loss": 0.2783,
+      "step": 6870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.559355018070615e-06,
+      "loss": 0.2555,
+      "step": 6871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.3146,
+      "step": 6872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5315540728384764e-06,
+      "loss": 0.346,
+      "step": 6873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.517653600222408e-06,
+      "loss": 0.3236,
+      "step": 6874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.503753127606339e-06,
+      "loss": 0.2716,
+      "step": 6875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.48985265499027e-06,
+      "loss": 0.2928,
+      "step": 6876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.475952182374201e-06,
+      "loss": 0.2976,
+      "step": 6877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.462051709758132e-06,
+      "loss": 0.3006,
+      "step": 6878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.448151237142064e-06,
+      "loss": 0.3407,
+      "step": 6879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4342507645259945e-06,
+      "loss": 0.2774,
+      "step": 6880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4203502919099246e-06,
+      "loss": 0.3763,
+      "step": 6881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.406449819293856e-06,
+      "loss": 0.2692,
+      "step": 6882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.392549346677787e-06,
+      "loss": 0.3076,
+      "step": 6883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.378648874061719e-06,
+      "loss": 0.3047,
+      "step": 6884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.364748401445649e-06,
+      "loss": 0.2828,
+      "step": 6885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.35084792882958e-06,
+      "loss": 0.2968,
+      "step": 6886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.336947456213512e-06,
+      "loss": 0.3068,
+      "step": 6887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.323046983597443e-06,
+      "loss": 0.2526,
+      "step": 6888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3091465109813735e-06,
+      "loss": 0.3968,
+      "step": 6889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2952460383653045e-06,
+      "loss": 0.2943,
+      "step": 6890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.281345565749235e-06,
+      "loss": 0.3108,
+      "step": 6891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.267445093133167e-06,
+      "loss": 0.2955,
+      "step": 6892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.253544620517098e-06,
+      "loss": 0.2845,
+      "step": 6893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.239644147901029e-06,
+      "loss": 0.2508,
+      "step": 6894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22574367528496e-06,
+      "loss": 0.3551,
+      "step": 6895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.211843202668891e-06,
+      "loss": 0.28,
+      "step": 6896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1979427300528225e-06,
+      "loss": 0.2613,
+      "step": 6897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.184042257436753e-06,
+      "loss": 0.2919,
+      "step": 6898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1701417848206835e-06,
+      "loss": 0.3412,
+      "step": 6899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.156241312204615e-06,
+      "loss": 0.2894,
+      "step": 6900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.142340839588546e-06,
+      "loss": 0.2782,
+      "step": 6901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.128440366972477e-06,
+      "loss": 0.2626,
+      "step": 6902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.114539894356408e-06,
+      "loss": 0.31,
+      "step": 6903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.100639421740339e-06,
+      "loss": 0.3075,
+      "step": 6904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.086738949124271e-06,
+      "loss": 0.3357,
+      "step": 6905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0728384765082016e-06,
+      "loss": 0.3199,
+      "step": 6906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0589380038921325e-06,
+      "loss": 0.3179,
+      "step": 6907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.045037531276063e-06,
+      "loss": 0.3117,
+      "step": 6908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.031137058659994e-06,
+      "loss": 0.3404,
+      "step": 6909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.017236586043926e-06,
+      "loss": 0.329,
+      "step": 6910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.003336113427857e-06,
+      "loss": 0.2761,
+      "step": 6911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.989435640811788e-06,
+      "loss": 0.2945,
+      "step": 6912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.975535168195719e-06,
+      "loss": 0.3057,
+      "step": 6913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.96163469557965e-06,
+      "loss": 0.2729,
+      "step": 6914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9477342229635815e-06,
+      "loss": 0.2612,
+      "step": 6915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9338337503475115e-06,
+      "loss": 0.2771,
+      "step": 6916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.919933277731443e-06,
+      "loss": 0.2872,
+      "step": 6917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.906032805115374e-06,
+      "loss": 0.2786,
+      "step": 6918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.892132332499305e-06,
+      "loss": 0.3197,
+      "step": 6919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.878231859883236e-06,
+      "loss": 0.3013,
+      "step": 6920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.864331387267167e-06,
+      "loss": 0.2719,
+      "step": 6921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.850430914651099e-06,
+      "loss": 0.3147,
+      "step": 6922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.83653044203503e-06,
+      "loss": 0.3002,
+      "step": 6923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8226299694189605e-06,
+      "loss": 0.3155,
+      "step": 6924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.808729496802892e-06,
+      "loss": 0.3132,
+      "step": 6925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7948290241868223e-06,
+      "loss": 0.285,
+      "step": 6926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.780928551570754e-06,
+      "loss": 0.2927,
+      "step": 6927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7670280789546846e-06,
+      "loss": 0.2355,
+      "step": 6928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7531276063386155e-06,
+      "loss": 0.2671,
+      "step": 6929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.739227133722547e-06,
+      "loss": 0.2733,
+      "step": 6930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7253266611064777e-06,
+      "loss": 0.3159,
+      "step": 6931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.711426188490409e-06,
+      "loss": 0.3483,
+      "step": 6932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.69752571587434e-06,
+      "loss": 0.3225,
+      "step": 6933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.683625243258271e-06,
+      "loss": 0.2774,
+      "step": 6934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6697247706422022e-06,
+      "loss": 0.3146,
+      "step": 6935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.655824298026133e-06,
+      "loss": 0.286,
+      "step": 6936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6419238254100636e-06,
+      "loss": 0.2888,
+      "step": 6937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6280233527939954e-06,
+      "loss": 0.3144,
+      "step": 6938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.614122880177926e-06,
+      "loss": 0.2819,
+      "step": 6939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6002224075618576e-06,
+      "loss": 0.2864,
+      "step": 6940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.586321934945788e-06,
+      "loss": 0.3349,
+      "step": 6941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.572421462329719e-06,
+      "loss": 0.3057,
+      "step": 6942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5585209897136504e-06,
+      "loss": 0.327,
+      "step": 6943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5446205170975813e-06,
+      "loss": 0.2952,
+      "step": 6944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5307200444815126e-06,
+      "loss": 0.3335,
+      "step": 6945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5168195718654435e-06,
+      "loss": 0.2949,
+      "step": 6946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5029190992493744e-06,
+      "loss": 0.3468,
+      "step": 6947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4890186266333058e-06,
+      "loss": 0.308,
+      "step": 6948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4751181540172367e-06,
+      "loss": 0.3019,
+      "step": 6949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.461217681401168e-06,
+      "loss": 0.2523,
+      "step": 6950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.447317208785099e-06,
+      "loss": 0.2837,
+      "step": 6951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4334167361690294e-06,
+      "loss": 0.2908,
+      "step": 6952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.419516263552961e-06,
+      "loss": 0.2839,
+      "step": 6953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.405615790936892e-06,
+      "loss": 0.3737,
+      "step": 6954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3917153183208234e-06,
+      "loss": 0.3118,
+      "step": 6955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3778148457047543e-06,
+      "loss": 0.2796,
+      "step": 6956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.363914373088685e-06,
+      "loss": 0.3129,
+      "step": 6957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3500139004726166e-06,
+      "loss": 0.3071,
+      "step": 6958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.336113427856547e-06,
+      "loss": 0.2828,
+      "step": 6959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.322212955240479e-06,
+      "loss": 0.2909,
+      "step": 6960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3083124826244093e-06,
+      "loss": 0.3149,
+      "step": 6961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2944120100083402e-06,
+      "loss": 0.2752,
+      "step": 6962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2805115373922715e-06,
+      "loss": 0.2934,
+      "step": 6963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2666110647762025e-06,
+      "loss": 0.325,
+      "step": 6964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.252710592160134e-06,
+      "loss": 0.3107,
+      "step": 6965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2388101195440647e-06,
+      "loss": 0.2845,
+      "step": 6966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2249096469279956e-06,
+      "loss": 0.2461,
+      "step": 6967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.211009174311927e-06,
+      "loss": 0.328,
+      "step": 6968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.197108701695858e-06,
+      "loss": 0.3204,
+      "step": 6969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.183208229079789e-06,
+      "loss": 0.316,
+      "step": 6970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.16930775646372e-06,
+      "loss": 0.3201,
+      "step": 6971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1554072838476506e-06,
+      "loss": 0.2982,
+      "step": 6972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1415068112315824e-06,
+      "loss": 0.3125,
+      "step": 6973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.127606338615513e-06,
+      "loss": 0.3203,
+      "step": 6974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.113705865999444e-06,
+      "loss": 0.305,
+      "step": 6975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.099805393383375e-06,
+      "loss": 0.2893,
+      "step": 6976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0859049207673064e-06,
+      "loss": 0.2749,
+      "step": 6977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0720044481512373e-06,
+      "loss": 0.3191,
+      "step": 6978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0581039755351682e-06,
+      "loss": 0.3095,
+      "step": 6979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0442035029190996e-06,
+      "loss": 0.2864,
+      "step": 6980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.3808,
+      "step": 6981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.016402557686962e-06,
+      "loss": 0.2472,
+      "step": 6982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0025020850708923e-06,
+      "loss": 0.291,
+      "step": 6983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9886016124548236e-06,
+      "loss": 0.3273,
+      "step": 6984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9747011398387546e-06,
+      "loss": 0.3075,
+      "step": 6985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.960800667222686e-06,
+      "loss": 0.294,
+      "step": 6986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.946900194606617e-06,
+      "loss": 0.2933,
+      "step": 6987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9329997219905477e-06,
+      "loss": 0.2909,
+      "step": 6988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.919099249374479e-06,
+      "loss": 0.2934,
+      "step": 6989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.90519877675841e-06,
+      "loss": 0.3188,
+      "step": 6990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8912983041423413e-06,
+      "loss": 0.321,
+      "step": 6991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8773978315262718e-06,
+      "loss": 0.3072,
+      "step": 6992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.863497358910203e-06,
+      "loss": 0.2848,
+      "step": 6993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.849596886294134e-06,
+      "loss": 0.2804,
+      "step": 6994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8356964136780654e-06,
+      "loss": 0.2611,
+      "step": 6995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8217959410619963e-06,
+      "loss": 0.2951,
+      "step": 6996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.807895468445927e-06,
+      "loss": 0.3693,
+      "step": 6997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.793994995829858e-06,
+      "loss": 0.3133,
+      "step": 6998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7800945232137894e-06,
+      "loss": 0.3078,
+      "step": 6999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7661940505977203e-06,
+      "loss": 0.2764,
+      "step": 7000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7522935779816517e-06,
+      "loss": 0.3599,
+      "step": 7001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7383931053655826e-06,
+      "loss": 0.2951,
+      "step": 7002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7244926327495135e-06,
+      "loss": 0.3416,
+      "step": 7003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.710592160133445e-06,
+      "loss": 0.2851,
+      "step": 7004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6966916875173757e-06,
+      "loss": 0.2839,
+      "step": 7005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.682791214901307e-06,
+      "loss": 0.2758,
+      "step": 7006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6688907422852376e-06,
+      "loss": 0.2833,
+      "step": 7007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.654990269669169e-06,
+      "loss": 0.252,
+      "step": 7008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6410897970531e-06,
+      "loss": 0.283,
+      "step": 7009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.627189324437031e-06,
+      "loss": 0.2844,
+      "step": 7010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.613288851820962e-06,
+      "loss": 0.3171,
+      "step": 7011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.599388379204893e-06,
+      "loss": 0.3346,
+      "step": 7012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5854879065888243e-06,
+      "loss": 0.3223,
+      "step": 7013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.571587433972755e-06,
+      "loss": 0.3239,
+      "step": 7014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5576869613566865e-06,
+      "loss": 0.2625,
+      "step": 7015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.543786488740617e-06,
+      "loss": 0.2665,
+      "step": 7016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5298860161245484e-06,
+      "loss": 0.2607,
+      "step": 7017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5159855435084793e-06,
+      "loss": 0.3338,
+      "step": 7018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5020850708924106e-06,
+      "loss": 0.2817,
+      "step": 7019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4881845982763415e-06,
+      "loss": 0.3594,
+      "step": 7020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4742841256602724e-06,
+      "loss": 0.305,
+      "step": 7021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4603836530442038e-06,
+      "loss": 0.2807,
+      "step": 7022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4464831804281347e-06,
+      "loss": 0.2745,
+      "step": 7023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.432582707812066e-06,
+      "loss": 0.3046,
+      "step": 7024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.418682235195997e-06,
+      "loss": 0.2805,
+      "step": 7025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.404781762579928e-06,
+      "loss": 0.2808,
+      "step": 7026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3908812899638587e-06,
+      "loss": 0.2654,
+      "step": 7027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.37698081734779e-06,
+      "loss": 0.2715,
+      "step": 7028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.363080344731721e-06,
+      "loss": 0.314,
+      "step": 7029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.349179872115652e-06,
+      "loss": 0.3495,
+      "step": 7030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.335279399499583e-06,
+      "loss": 0.2738,
+      "step": 7031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.321378926883514e-06,
+      "loss": 0.3219,
+      "step": 7032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.307478454267445e-06,
+      "loss": 0.2388,
+      "step": 7033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2935779816513764e-06,
+      "loss": 0.3022,
+      "step": 7034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2796775090353073e-06,
+      "loss": 0.2694,
+      "step": 7035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2657770364192382e-06,
+      "loss": 0.2337,
+      "step": 7036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2518765638031695e-06,
+      "loss": 0.2931,
+      "step": 7037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2379760911871005e-06,
+      "loss": 0.2852,
+      "step": 7038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.224075618571032e-06,
+      "loss": 0.2342,
+      "step": 7039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2101751459549623e-06,
+      "loss": 0.336,
+      "step": 7040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1962746733388936e-06,
+      "loss": 0.2604,
+      "step": 7041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1823742007228245e-06,
+      "loss": 0.3102,
+      "step": 7042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.168473728106756e-06,
+      "loss": 0.2961,
+      "step": 7043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1545732554906868e-06,
+      "loss": 0.2896,
+      "step": 7044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1406727828746177e-06,
+      "loss": 0.2902,
+      "step": 7045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.126772310258549e-06,
+      "loss": 0.2853,
+      "step": 7046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.11287183764248e-06,
+      "loss": 0.3095,
+      "step": 7047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0989713650264113e-06,
+      "loss": 0.2838,
+      "step": 7048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0850708924103418e-06,
+      "loss": 0.2394,
+      "step": 7049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.071170419794273e-06,
+      "loss": 0.2789,
+      "step": 7050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.057269947178204e-06,
+      "loss": 0.2877,
+      "step": 7051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0433694745621353e-06,
+      "loss": 0.3187,
+      "step": 7052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0294690019460662e-06,
+      "loss": 0.2853,
+      "step": 7053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015568529329997e-06,
+      "loss": 0.2693,
+      "step": 7054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0016680567139285e-06,
+      "loss": 0.351,
+      "step": 7055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9877675840978594e-06,
+      "loss": 0.3057,
+      "step": 7056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9738671114817907e-06,
+      "loss": 0.2759,
+      "step": 7057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9599666388657216e-06,
+      "loss": 0.2542,
+      "step": 7058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9460661662496526e-06,
+      "loss": 0.286,
+      "step": 7059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9321656936335835e-06,
+      "loss": 0.3182,
+      "step": 7060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.918265221017515e-06,
+      "loss": 0.295,
+      "step": 7061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.904364748401446e-06,
+      "loss": 0.286,
+      "step": 7062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.890464275785377e-06,
+      "loss": 0.2594,
+      "step": 7063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8765638031693077e-06,
+      "loss": 0.3381,
+      "step": 7064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8626633305532389e-06,
+      "loss": 0.2957,
+      "step": 7065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.84876285793717e-06,
+      "loss": 0.3192,
+      "step": 7066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8348623853211011e-06,
+      "loss": 0.2544,
+      "step": 7067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8209619127050318e-06,
+      "loss": 0.3285,
+      "step": 7068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.807061440088963e-06,
+      "loss": 0.2519,
+      "step": 7069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.793160967472894e-06,
+      "loss": 0.2791,
+      "step": 7070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7792604948568252e-06,
+      "loss": 0.2821,
+      "step": 7071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7653600222407563e-06,
+      "loss": 0.2831,
+      "step": 7072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7514595496246872e-06,
+      "loss": 0.385,
+      "step": 7073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7375590770086183e-06,
+      "loss": 0.3004,
+      "step": 7074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7236586043925495e-06,
+      "loss": 0.2895,
+      "step": 7075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7097581317764806e-06,
+      "loss": 0.2632,
+      "step": 7076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6958576591604117e-06,
+      "loss": 0.3105,
+      "step": 7077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6819571865443424e-06,
+      "loss": 0.3114,
+      "step": 7078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6680567139282735e-06,
+      "loss": 0.308,
+      "step": 7079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6541562413122047e-06,
+      "loss": 0.2672,
+      "step": 7080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6402557686961358e-06,
+      "loss": 0.3426,
+      "step": 7081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.626355296080067e-06,
+      "loss": 0.2794,
+      "step": 7082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6124548234639978e-06,
+      "loss": 0.2959,
+      "step": 7083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.598554350847929e-06,
+      "loss": 0.2959,
+      "step": 7084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.58465387823186e-06,
+      "loss": 0.3301,
+      "step": 7085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5707534056157912e-06,
+      "loss": 0.3009,
+      "step": 7086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.556852932999722e-06,
+      "loss": 0.2859,
+      "step": 7087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5429524603836532e-06,
+      "loss": 0.319,
+      "step": 7088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5290519877675841e-06,
+      "loss": 0.3292,
+      "step": 7089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.3034,
+      "step": 7090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5012510425354462e-06,
+      "loss": 0.3137,
+      "step": 7091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4873505699193773e-06,
+      "loss": 0.2955,
+      "step": 7092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4734500973033084e-06,
+      "loss": 0.2714,
+      "step": 7093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4595496246872395e-06,
+      "loss": 0.2618,
+      "step": 7094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4456491520711706e-06,
+      "loss": 0.3016,
+      "step": 7095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4317486794551016e-06,
+      "loss": 0.2504,
+      "step": 7096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4178482068390327e-06,
+      "loss": 0.2924,
+      "step": 7097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4039477342229636e-06,
+      "loss": 0.3167,
+      "step": 7098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3900472616068947e-06,
+      "loss": 0.3072,
+      "step": 7099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3761467889908258e-06,
+      "loss": 0.2975,
+      "step": 7100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3622463163747567e-06,
+      "loss": 0.3066,
+      "step": 7101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3483458437586879e-06,
+      "loss": 0.3018,
+      "step": 7102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3344453711426188e-06,
+      "loss": 0.2961,
+      "step": 7103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.32054489852655e-06,
+      "loss": 0.3161,
+      "step": 7104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.306644425910481e-06,
+      "loss": 0.307,
+      "step": 7105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2927439532944121e-06,
+      "loss": 0.3135,
+      "step": 7106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2788434806783433e-06,
+      "loss": 0.2926,
+      "step": 7107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2649430080622742e-06,
+      "loss": 0.2778,
+      "step": 7108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2510425354462053e-06,
+      "loss": 0.2816,
+      "step": 7109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2371420628301362e-06,
+      "loss": 0.3191,
+      "step": 7110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2232415902140673e-06,
+      "loss": 0.2902,
+      "step": 7111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2093411175979985e-06,
+      "loss": 0.2833,
+      "step": 7112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1954406449819294e-06,
+      "loss": 0.3421,
+      "step": 7113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1815401723658605e-06,
+      "loss": 0.3099,
+      "step": 7114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1676396997497914e-06,
+      "loss": 0.298,
+      "step": 7115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1537392271337225e-06,
+      "loss": 0.3052,
+      "step": 7116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1398387545176537e-06,
+      "loss": 0.3065,
+      "step": 7117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1259382819015848e-06,
+      "loss": 0.2792,
+      "step": 7118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.112037809285516e-06,
+      "loss": 0.3267,
+      "step": 7119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0981373366694468e-06,
+      "loss": 0.2808,
+      "step": 7120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.084236864053378e-06,
+      "loss": 0.3002,
+      "step": 7121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0703363914373088e-06,
+      "loss": 0.3158,
+      "step": 7122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.05643591882124e-06,
+      "loss": 0.3198,
+      "step": 7123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0425354462051709e-06,
+      "loss": 0.2864,
+      "step": 7124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.028634973589102e-06,
+      "loss": 0.292,
+      "step": 7125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0147345009730331e-06,
+      "loss": 0.2804,
+      "step": 7126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0008340283569642e-06,
+      "loss": 0.3039,
+      "step": 7127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.869335557408954e-07,
+      "loss": 0.3132,
+      "step": 7128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.730330831248263e-07,
+      "loss": 0.3274,
+      "step": 7129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.591326105087574e-07,
+      "loss": 0.284,
+      "step": 7130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.452321378926885e-07,
+      "loss": 0.3277,
+      "step": 7131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.313316652766194e-07,
+      "loss": 0.2859,
+      "step": 7132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.174311926605506e-07,
+      "loss": 0.271,
+      "step": 7133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.035307200444815e-07,
+      "loss": 0.2914,
+      "step": 7134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.896302474284126e-07,
+      "loss": 0.2728,
+      "step": 7135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.757297748123436e-07,
+      "loss": 0.2573,
+      "step": 7136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.618293021962747e-07,
+      "loss": 0.2659,
+      "step": 7137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.479288295802059e-07,
+      "loss": 0.2998,
+      "step": 7138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.340283569641368e-07,
+      "loss": 0.2774,
+      "step": 7139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.201278843480679e-07,
+      "loss": 0.2721,
+      "step": 7140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.062274117319989e-07,
+      "loss": 0.3138,
+      "step": 7141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.9232693911593e-07,
+      "loss": 0.3079,
+      "step": 7142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.78426466499861e-07,
+      "loss": 0.2852,
+      "step": 7143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.645259938837921e-07,
+      "loss": 0.276,
+      "step": 7144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.506255212677231e-07,
+      "loss": 0.3106,
+      "step": 7145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.367250486516542e-07,
+      "loss": 0.298,
+      "step": 7146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.228245760355853e-07,
+      "loss": 0.2474,
+      "step": 7147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.089241034195163e-07,
+      "loss": 0.2783,
+      "step": 7148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.950236308034474e-07,
+      "loss": 0.3256,
+      "step": 7149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.811231581873784e-07,
+      "loss": 0.2854,
+      "step": 7150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.672226855713094e-07,
+      "loss": 0.3055,
+      "step": 7151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.533222129552405e-07,
+      "loss": 0.3149,
+      "step": 7152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.394217403391716e-07,
+      "loss": 0.2814,
+      "step": 7153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.255212677231027e-07,
+      "loss": 0.3221,
+      "step": 7154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.116207951070337e-07,
+      "loss": 0.2653,
+      "step": 7155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.977203224909647e-07,
+      "loss": 0.282,
+      "step": 7156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.838198498748957e-07,
+      "loss": 0.3032,
+      "step": 7157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.699193772588268e-07,
+      "loss": 0.3194,
+      "step": 7158
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.56018904642758e-07,
+      "loss": 0.2893,
+      "step": 7159
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.42118432026689e-07,
+      "loss": 0.2852,
+      "step": 7160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2821795941062e-07,
+      "loss": 0.3479,
+      "step": 7161
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.14317486794551e-07,
+      "loss": 0.3047,
+      "step": 7162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.004170141784821e-07,
+      "loss": 0.3211,
+      "step": 7163
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.865165415624131e-07,
+      "loss": 0.2955,
+      "step": 7164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7261606894634426e-07,
+      "loss": 0.2962,
+      "step": 7165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.587155963302753e-07,
+      "loss": 0.2988,
+      "step": 7166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.448151237142063e-07,
+      "loss": 0.2581,
+      "step": 7167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3091465109813737e-07,
+      "loss": 0.3352,
+      "step": 7168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.170141784820684e-07,
+      "loss": 0.309,
+      "step": 7169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0311370586599945e-07,
+      "loss": 0.2911,
+      "step": 7170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.892132332499305e-07,
+      "loss": 0.2494,
+      "step": 7171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7531276063386154e-07,
+      "loss": 0.2772,
+      "step": 7172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6141228801779266e-07,
+      "loss": 0.3541,
+      "step": 7173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.475118154017237e-07,
+      "loss": 0.359,
+      "step": 7174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336113427856547e-07,
+      "loss": 0.3223,
+      "step": 7175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.197108701695858e-07,
+      "loss": 0.3396,
+      "step": 7176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0581039755351683e-07,
+      "loss": 0.2793,
+      "step": 7177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9190992493744785e-07,
+      "loss": 0.3235,
+      "step": 7178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.78009452321379e-07,
+      "loss": 0.3342,
+      "step": 7179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6410897970531e-07,
+      "loss": 0.2882,
+      "step": 7180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5020850708924106e-07,
+      "loss": 0.2583,
+      "step": 7181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3630803447317213e-07,
+      "loss": 0.2981,
+      "step": 7182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2240756185710315e-07,
+      "loss": 0.2857,
+      "step": 7183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.085070892410342e-07,
+      "loss": 0.288,
+      "step": 7184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9460661662496526e-07,
+      "loss": 0.2911,
+      "step": 7185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8070614400889633e-07,
+      "loss": 0.3206,
+      "step": 7186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6680567139282735e-07,
+      "loss": 0.2668,
+      "step": 7187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5290519877675842e-07,
+      "loss": 0.299,
+      "step": 7188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.390047261606895e-07,
+      "loss": 0.2869,
+      "step": 7189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2510425354462053e-07,
+      "loss": 0.3459,
+      "step": 7190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1120378092855157e-07,
+      "loss": 0.2494,
+      "step": 7191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.730330831248263e-08,
+      "loss": 0.2996,
+      "step": 7192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.340283569641367e-08,
+      "loss": 0.3308,
+      "step": 7193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.950236308034474e-08,
+      "loss": 0.317,
+      "step": 7194
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.5577203631401062,
+      "eval_runtime": 1142.9536,
+      "eval_samples_per_second": 14.26,
+      "eval_steps_per_second": 0.446,
+      "eval_wer": 0.38635837078912005,
+      "step": 7194
+    },
+    {
+      "epoch": 1.0,
+      "step": 7194,
+      "total_flos": 1.3853238593561438e+20,
+      "train_loss": 0.5645441397017892,
+      "train_runtime": 66958.2014,
+      "train_samples_per_second": 13.753,
+      "train_steps_per_second": 0.107
+    }
+  ],
+  "max_steps": 7194,
+  "num_train_epochs": 1,
+  "total_flos": 1.3853238593561438e+20,
+  "trial_name": null,
+  "trial_params": null
+}