diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6778 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.995008319467554,
+  "eval_steps": 500,
+  "global_step": 1125,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 0.5904,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.5576,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 0.5976,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.5973,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.5639,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.5819,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.5628,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.618,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.5294,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.5813,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.5424,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.6106,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.5481,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.5409,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.591,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.5591,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.5605,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.531,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 0.5507,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.5224,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.235294117647059e-05,
+      "loss": 0.5045,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.5102,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3529411764705885e-05,
+      "loss": 0.5219,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.5346,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4705882352941179e-05,
+      "loss": 0.5148,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.4681,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5882352941176473e-05,
+      "loss": 0.5132,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.5243,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7058823529411767e-05,
+      "loss": 0.5211,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.4758,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.823529411764706e-05,
+      "loss": 0.5082,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.5046,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 0.4893,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.5332,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999958540892526e-05,
+      "loss": 0.4484,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999834163913866e-05,
+      "loss": 0.4842,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999626870095335e-05,
+      "loss": 0.5139,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999933666115578e-05,
+      "loss": 0.5364,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998963539501557e-05,
+      "loss": 0.5336,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998507508226525e-05,
+      "loss": 0.5511,
+      "step": 40
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997968571112015e-05,
+      "loss": 0.5051,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997346732626797e-05,
+      "loss": 0.4877,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996641997927045e-05,
+      "loss": 0.5369,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999585437285629e-05,
+      "loss": 0.5003,
+      "step": 44
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994983863945387e-05,
+      "loss": 0.4985,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994030478412432e-05,
+      "loss": 0.5352,
+      "step": 46
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999299422416273e-05,
+      "loss": 0.4964,
+      "step": 47
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999187510978872e-05,
+      "loss": 0.5113,
+      "step": 48
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999067314456989e-05,
+      "loss": 0.499,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989388338472732e-05,
+      "loss": 0.4999,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998802070215062e-05,
+      "loss": 0.4777,
+      "step": 51
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986570246943756e-05,
+      "loss": 0.5288,
+      "step": 52
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985036984879053e-05,
+      "loss": 0.528,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983420928670044e-05,
+      "loss": 0.4732,
+      "step": 54
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981722091716784e-05,
+      "loss": 0.4584,
+      "step": 55
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979940488105722e-05,
+      "loss": 0.4659,
+      "step": 56
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99780761326096e-05,
+      "loss": 0.4853,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997612904068732e-05,
+      "loss": 0.4551,
+      "step": 58
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974099228483813e-05,
+      "loss": 0.4559,
+      "step": 59
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997198671282993e-05,
+      "loss": 0.4513,
+      "step": 60
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969791511242272e-05,
+      "loss": 0.485,
+      "step": 61
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967513641923056e-05,
+      "loss": 0.4983,
+      "step": 62
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996515312375997e-05,
+      "loss": 0.4511,
+      "step": 63
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962709976326004e-05,
+      "loss": 0.5124,
+      "step": 64
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960184219879303e-05,
+      "loss": 0.4426,
+      "step": 65
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957575875362993e-05,
+      "loss": 0.4787,
+      "step": 66
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954884964404995e-05,
+      "loss": 0.4828,
+      "step": 67
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995211150931787e-05,
+      "loss": 0.4808,
+      "step": 68
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949255533098602e-05,
+      "loss": 0.448,
+      "step": 69
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946317059428447e-05,
+      "loss": 0.4429,
+      "step": 70
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943296112672697e-05,
+      "loss": 0.5037,
+      "step": 71
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99401927178805e-05,
+      "loss": 0.4844,
+      "step": 72
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937006900784664e-05,
+      "loss": 0.477,
+      "step": 73
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933738687801405e-05,
+      "loss": 0.4675,
+      "step": 74
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930388106030167e-05,
+      "loss": 0.481,
+      "step": 75
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926955183253374e-05,
+      "loss": 0.4965,
+      "step": 76
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923439947936204e-05,
+      "loss": 0.47,
+      "step": 77
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919842429226374e-05,
+      "loss": 0.5255,
+      "step": 78
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991616265695385e-05,
+      "loss": 0.5011,
+      "step": 79
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912400661630658e-05,
+      "loss": 0.474,
+      "step": 80
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990855647445059e-05,
+      "loss": 0.498,
+      "step": 81
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904630127288958e-05,
+      "loss": 0.4869,
+      "step": 82
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900621652702337e-05,
+      "loss": 0.4715,
+      "step": 83
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896531083928273e-05,
+      "loss": 0.4819,
+      "step": 84
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892358454885042e-05,
+      "loss": 0.4892,
+      "step": 85
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888103800171336e-05,
+      "loss": 0.4739,
+      "step": 86
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883767155065987e-05,
+      "loss": 0.4643,
+      "step": 87
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987934855552769e-05,
+      "loss": 0.4706,
+      "step": 88
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874848038194673e-05,
+      "loss": 0.5426,
+      "step": 89
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870265640384436e-05,
+      "loss": 0.4933,
+      "step": 90
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865601400093395e-05,
+      "loss": 0.525,
+      "step": 91
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98608553559966e-05,
+      "loss": 0.493,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98560275474474e-05,
+      "loss": 0.4715,
+      "step": 93
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851118014477128e-05,
+      "loss": 0.5041,
+      "step": 94
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846126797794743e-05,
+      "loss": 0.5204,
+      "step": 95
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841053938786535e-05,
+      "loss": 0.4319,
+      "step": 96
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835899479515738e-05,
+      "loss": 0.4617,
+      "step": 97
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983066346272221e-05,
+      "loss": 0.5126,
+      "step": 98
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825345931822068e-05,
+      "loss": 0.4628,
+      "step": 99
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981994693090733e-05,
+      "loss": 0.4566,
+      "step": 100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981446650474555e-05,
+      "loss": 0.4743,
+      "step": 101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808904698779434e-05,
+      "loss": 0.4863,
+      "step": 102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803261559126493e-05,
+      "loss": 0.4734,
+      "step": 103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797537132578633e-05,
+      "loss": 0.4591,
+      "step": 104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791731466601774e-05,
+      "loss": 0.4906,
+      "step": 105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978584460933546e-05,
+      "loss": 0.4794,
+      "step": 106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779876609592473e-05,
+      "loss": 0.4922,
+      "step": 107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773827516858387e-05,
+      "loss": 0.4898,
+      "step": 108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767697381291203e-05,
+      "loss": 0.4764,
+      "step": 109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761486253720915e-05,
+      "loss": 0.4383,
+      "step": 110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975519418564908e-05,
+      "loss": 0.4845,
+      "step": 111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748821229248404e-05,
+      "loss": 0.467,
+      "step": 112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742367437362307e-05,
+      "loss": 0.4958,
+      "step": 113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735832863504474e-05,
+      "loss": 0.4925,
+      "step": 114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729217561858435e-05,
+      "loss": 0.4825,
+      "step": 115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722521587277077e-05,
+      "loss": 0.4373,
+      "step": 116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715744995282235e-05,
+      "loss": 0.4496,
+      "step": 117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708887842064196e-05,
+      "loss": 0.4494,
+      "step": 118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970195018448125e-05,
+      "loss": 0.4643,
+      "step": 119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969493208005922e-05,
+      "loss": 0.4556,
+      "step": 120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687833586990975e-05,
+      "loss": 0.4781,
+      "step": 121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680654764135944e-05,
+      "loss": 0.4463,
+      "step": 122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673395671019653e-05,
+      "loss": 0.4781,
+      "step": 123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666056367833203e-05,
+      "loss": 0.5488,
+      "step": 124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965863691543279e-05,
+      "loss": 0.4649,
+      "step": 125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965113737533918e-05,
+      "loss": 0.5141,
+      "step": 126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964355780973723e-05,
+      "loss": 0.4839,
+      "step": 127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635898281475344e-05,
+      "loss": 0.4514,
+      "step": 128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962815885406496e-05,
+      "loss": 0.4765,
+      "step": 129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620339591680024e-05,
+      "loss": 0.4466,
+      "step": 130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612440559156472e-05,
+      "loss": 0.4874,
+      "step": 131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604461821991667e-05,
+      "loss": 0.4813,
+      "step": 132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596403446343875e-05,
+      "loss": 0.507,
+      "step": 133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588265499031708e-05,
+      "loss": 0.4626,
+      "step": 134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580048047533578e-05,
+      "loss": 0.4685,
+      "step": 135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571751159987116e-05,
+      "loss": 0.4446,
+      "step": 136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563374905188636e-05,
+      "loss": 0.4674,
+      "step": 137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554919352592552e-05,
+      "loss": 0.4884,
+      "step": 138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954638457231079e-05,
+      "loss": 0.464,
+      "step": 139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537770635112232e-05,
+      "loss": 0.4561,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529077612422103e-05,
+      "loss": 0.4703,
+      "step": 141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520305576321394e-05,
+      "loss": 0.4658,
+      "step": 142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951145459954627e-05,
+      "loss": 0.4397,
+      "step": 143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950252475548744e-05,
+      "loss": 0.4715,
+      "step": 144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493516118189583e-05,
+      "loss": 0.4933,
+      "step": 145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948442876235071e-05,
+      "loss": 0.4527,
+      "step": 146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475262763321554e-05,
+      "loss": 0.4559,
+      "step": 147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466018197104945e-05,
+      "loss": 0.4691,
+      "step": 148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456695140355175e-05,
+      "loss": 0.4315,
+      "step": 149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447293670377362e-05,
+      "loss": 0.4032,
+      "step": 150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943781386512682e-05,
+      "loss": 0.5012,
+      "step": 151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94282558032084e-05,
+      "loss": 0.4351,
+      "step": 152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941861956387585e-05,
+      "loss": 0.4703,
+      "step": 153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408905227031142e-05,
+      "loss": 0.4525,
+      "step": 154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399112873223824e-05,
+      "loss": 0.4723,
+      "step": 155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389242583650348e-05,
+      "loss": 0.4652,
+      "step": 156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379294440153393e-05,
+      "loss": 0.4602,
+      "step": 157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369268525221187e-05,
+      "loss": 0.4548,
+      "step": 158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359164921986826e-05,
+      "loss": 0.4717,
+      "step": 159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348983714227584e-05,
+      "loss": 0.4493,
+      "step": 160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338724986364223e-05,
+      "loss": 0.4913,
+      "step": 161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932838882346028e-05,
+      "loss": 0.4985,
+      "step": 162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317975311221374e-05,
+      "loss": 0.4273,
+      "step": 163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930748453599449e-05,
+      "loss": 0.4683,
+      "step": 164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296916584767263e-05,
+      "loss": 0.4652,
+      "step": 165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286271545167253e-05,
+      "loss": 0.49,
+      "step": 166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927554950546124e-05,
+      "loss": 0.468,
+      "step": 167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926475055455445e-05,
+      "loss": 0.4566,
+      "step": 168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253874781989864e-05,
+      "loss": 0.453,
+      "step": 169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924292227794745e-05,
+      "loss": 0.4719,
+      "step": 170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231893133243408e-05,
+      "loss": 0.4674,
+      "step": 171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922078743932944e-05,
+      "loss": 0.43,
+      "step": 172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920960528829198e-05,
+      "loss": 0.4673,
+      "step": 173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198346772851423e-05,
+      "loss": 0.5261,
+      "step": 174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187011986361374e-05,
+      "loss": 0.4367,
+      "step": 175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175601022807862e-05,
+      "loss": 0.4772,
+      "step": 176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164113976808553e-05,
+      "loss": 0.4825,
+      "step": 177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152550943611986e-05,
+      "loss": 0.4749,
+      "step": 178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140912019096766e-05,
+      "loss": 0.4614,
+      "step": 179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129197299770782e-05,
+      "loss": 0.462,
+      "step": 180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911740688277039e-05,
+      "loss": 0.444,
+      "step": 181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105540865859625e-05,
+      "loss": 0.4612,
+      "step": 182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909359934742938e-05,
+      "loss": 0.4349,
+      "step": 183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081582426496598e-05,
+      "loss": 0.4456,
+      "step": 184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069490202703438e-05,
+      "loss": 0.4648,
+      "step": 185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905732277631647e-05,
+      "loss": 0.4385,
+      "step": 186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045080248225804e-05,
+      "loss": 0.435,
+      "step": 187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032762719944315e-05,
+      "loss": 0.4661,
+      "step": 188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902037029360674e-05,
+      "loss": 0.4373,
+      "step": 189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900790307196887e-05,
+      "loss": 0.4511,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899536115840668e-05,
+      "loss": 0.4598,
+      "step": 191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982744656915475e-05,
+      "loss": 0.459,
+      "step": 192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970053672109037e-05,
+      "loss": 0.4812,
+      "step": 193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957288309218745e-05,
+      "loss": 0.4666,
+      "step": 194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944448674092715e-05,
+      "loss": 0.4457,
+      "step": 195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89315348731949e-05,
+      "loss": 0.4034,
+      "step": 196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891854701360424e-05,
+      "loss": 0.4855,
+      "step": 197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905485203013746e-05,
+      "loss": 0.4421,
+      "step": 198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892349549729617e-05,
+      "loss": 0.4544,
+      "step": 199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879140162670346e-05,
+      "loss": 0.5055,
+      "step": 200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865857151365815e-05,
+      "loss": 0.4556,
+      "step": 201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852500625956382e-05,
+      "loss": 0.4914,
+      "step": 202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883907069719197e-05,
+      "loss": 0.4603,
+      "step": 203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882556747643115e-05,
+      "loss": 0.5052,
+      "step": 204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811991075640223e-05,
+      "loss": 0.4389,
+      "step": 205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798341607392278e-05,
+      "loss": 0.4708,
+      "step": 206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784619184866267e-05,
+      "loss": 0.4118,
+      "step": 207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877082392184607e-05,
+      "loss": 0.5046,
+      "step": 208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756955932719546e-05,
+      "loss": 0.4619,
+      "step": 209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743015332477587e-05,
+      "loss": 0.4588,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729002236713158e-05,
+      "loss": 0.4812,
+      "step": 211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871491676162035e-05,
+      "loss": 0.4636,
+      "step": 212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870075902399341e-05,
+      "loss": 0.4714,
+      "step": 213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686529141225767e-05,
+      "loss": 0.5219,
+      "step": 214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867222723130907e-05,
+      "loss": 0.4643,
+      "step": 215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657853412832204e-05,
+      "loss": 0.4653,
+      "step": 216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643407804980303e-05,
+      "loss": 0.4901,
+      "step": 217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628890527533776e-05,
+      "loss": 0.4314,
+      "step": 218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8614301700867287e-05,
+      "loss": 0.4434,
+      "step": 219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599641445948792e-05,
+      "loss": 0.4869,
+      "step": 220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85849098843385e-05,
+      "loss": 0.4681,
+      "step": 221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570107138187895e-05,
+      "loss": 0.4687,
+      "step": 222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555233330238703e-05,
+      "loss": 0.4753,
+      "step": 223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854028858382188e-05,
+      "loss": 0.441,
+      "step": 224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525273022856608e-05,
+      "loss": 0.5043,
+      "step": 225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851018677184923e-05,
+      "loss": 0.4599,
+      "step": 226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495029955892247e-05,
+      "loss": 0.4627,
+      "step": 227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847980270066327e-05,
+      "loss": 0.4783,
+      "step": 228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464505132423985e-05,
+      "loss": 0.4558,
+      "step": 229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449137378019095e-05,
+      "loss": 0.4461,
+      "step": 230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8433699564875275e-05,
+      "loss": 0.4648,
+      "step": 231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841819182100012e-05,
+      "loss": 0.4398,
+      "step": 232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840261427498107e-05,
+      "loss": 0.4338,
+      "step": 233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838696705598436e-05,
+      "loss": 0.4583,
+      "step": 234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837125029375393e-05,
+      "loss": 0.4284,
+      "step": 235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355464118610375e-05,
+      "loss": 0.457,
+      "step": 236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339608661449833e-05,
+      "loss": 0.4481,
+      "step": 237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832368405374293e-05,
+      "loss": 0.4707,
+      "step": 238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307690427533673e-05,
+      "loss": 0.4477,
+      "step": 239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829162791543835e-05,
+      "loss": 0.5002,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275496650644446e-05,
+      "loss": 0.4989,
+      "step": 241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8259296766909528e-05,
+      "loss": 0.5131,
+      "step": 242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243028398560143e-05,
+      "loss": 0.4341,
+      "step": 243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226691680490696e-05,
+      "loss": 0.4495,
+      "step": 244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210286748162334e-05,
+      "loss": 0.4598,
+      "step": 245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8193813737601834e-05,
+      "loss": 0.5321,
+      "step": 246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177272785400455e-05,
+      "loss": 0.4797,
+      "step": 247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816066402871282e-05,
+      "loss": 0.4519,
+      "step": 248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8143987605255774e-05,
+      "loss": 0.4883,
+      "step": 249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127243653307248e-05,
+      "loss": 0.4784,
+      "step": 250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110432311705095e-05,
+      "loss": 0.5353,
+      "step": 251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093553719845967e-05,
+      "loss": 0.424,
+      "step": 252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076608017684128e-05,
+      "loss": 0.4128,
+      "step": 253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805959534573032e-05,
+      "loss": 0.4379,
+      "step": 254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042515845050578e-05,
+      "loss": 0.4676,
+      "step": 255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025369657265075e-05,
+      "loss": 0.4222,
+      "step": 256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008156924546943e-05,
+      "loss": 0.4756,
+      "step": 257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990877789621082e-05,
+      "loss": 0.4549,
+      "step": 258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973532395763e-05,
+      "loss": 0.4625,
+      "step": 259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956120886797604e-05,
+      "loss": 0.4669,
+      "step": 260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.793864340709802e-05,
+      "loss": 0.4615,
+      "step": 261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921100101584387e-05,
+      "loss": 0.4547,
+      "step": 262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903491115722665e-05,
+      "loss": 0.4783,
+      "step": 263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885816595523425e-05,
+      "loss": 0.4639,
+      "step": 264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868076687540625e-05,
+      "loss": 0.4801,
+      "step": 265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850271538870423e-05,
+      "loss": 0.4445,
+      "step": 266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832401297149927e-05,
+      "loss": 0.4431,
+      "step": 267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7814466110555998e-05,
+      "loss": 0.4525,
+      "step": 268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7796466127804e-05,
+      "loss": 0.4699,
+      "step": 269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777840149814657e-05,
+      "loss": 0.4731,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760272371372404e-05,
+      "loss": 0.4674,
+      "step": 271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7742078897804977e-05,
+      "loss": 0.452,
+      "step": 272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723821228301324e-05,
+      "loss": 0.4288,
+      "step": 273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705499514250784e-05,
+      "loss": 0.4634,
+      "step": 274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687113907573744e-05,
+      "loss": 0.4303,
+      "step": 275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7668664560720362e-05,
+      "loss": 0.4547,
+      "step": 276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765015162666934e-05,
+      "loss": 0.4562,
+      "step": 277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631575258926612e-05,
+      "loss": 0.467,
+      "step": 278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761293561152411e-05,
+      "loss": 0.4708,
+      "step": 279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594232839018463e-05,
+      "loss": 0.4471,
+      "step": 280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757546709648972e-05,
+      "loss": 0.4447,
+      "step": 281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556638539540066e-05,
+      "loss": 0.4682,
+      "step": 282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7537747324292542e-05,
+      "loss": 0.4914,
+      "step": 283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518793607389725e-05,
+      "loss": 0.4371,
+      "step": 284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499777545992452e-05,
+      "loss": 0.4521,
+      "step": 285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748069929777852e-05,
+      "loss": 0.4432,
+      "step": 286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461559020941345e-05,
+      "loss": 0.4478,
+      "step": 287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7442356874188692e-05,
+      "loss": 0.4705,
+      "step": 288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7423093016741332e-05,
+      "loss": 0.4239,
+      "step": 289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403767608331735e-05,
+      "loss": 0.4289,
+      "step": 290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7384380809202736e-05,
+      "loss": 0.4434,
+      "step": 291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364932780106212e-05,
+      "loss": 0.4742,
+      "step": 292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345423682301754e-05,
+      "loss": 0.4567,
+      "step": 293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325853677555313e-05,
+      "loss": 0.4184,
+      "step": 294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306222928137874e-05,
+      "loss": 0.5237,
+      "step": 295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286531596824113e-05,
+      "loss": 0.4759,
+      "step": 296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726677984689103e-05,
+      "loss": 0.4586,
+      "step": 297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246967842116608e-05,
+      "loss": 0.4344,
+      "step": 298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227095746778456e-05,
+      "loss": 0.4498,
+      "step": 299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207163725652444e-05,
+      "loss": 0.4536,
+      "step": 300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718717194401133e-05,
+      "loss": 0.4665,
+      "step": 301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71671205676234e-05,
+      "loss": 0.411,
+      "step": 302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147009762751087e-05,
+      "loss": 0.4953,
+      "step": 303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126839696149596e-05,
+      "loss": 0.4451,
+      "step": 304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710661053506552e-05,
+      "loss": 0.4456,
+      "step": 305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.708632244723545e-05,
+      "loss": 0.4982,
+      "step": 306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065975600884584e-05,
+      "loss": 0.4688,
+      "step": 307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704557016472535e-05,
+      "loss": 0.4629,
+      "step": 308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025106307955976e-05,
+      "loss": 0.4529,
+      "step": 309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004584200259107e-05,
+      "loss": 0.4482,
+      "step": 310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6984004011800402e-05,
+      "loss": 0.4781,
+      "step": 311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696336591322711e-05,
+      "loss": 0.4882,
+      "step": 312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942670075666657e-05,
+      "loss": 0.4569,
+      "step": 313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921916670725233e-05,
+      "loss": 0.4784,
+      "step": 314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690110587048637e-05,
+      "loss": 0.4452,
+      "step": 315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880237847509514e-05,
+      "loss": 0.4762,
+      "step": 316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6859312774828578e-05,
+      "loss": 0.4675,
+      "step": 317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683833082595053e-05,
+      "loss": 0.455,
+      "step": 318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681729217485395e-05,
+      "loss": 0.4497,
+      "step": 319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679619699598757e-05,
+      "loss": 0.4655,
+      "step": 320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6775045464268856e-05,
+      "loss": 0.4772,
+      "step": 321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753837755082527e-05,
+      "loss": 0.4673,
+      "step": 322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673257404427912e-05,
+      "loss": 0.4126,
+      "step": 323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711254508173542e-05,
+      "loss": 0.5155,
+      "step": 324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6689879323543567e-05,
+      "loss": 0.4527,
+      "step": 325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668448667628417e-05,
+      "loss": 0.4565,
+      "step": 326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646962718127263e-05,
+      "loss": 0.4655,
+      "step": 327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6625421653197768e-05,
+      "loss": 0.4525,
+      "step": 328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660382565145459e-05,
+      "loss": 0.4871,
+      "step": 329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658217489196792e-05,
+      "loss": 0.4589,
+      "step": 330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560469554262e-05,
+      "loss": 0.4861,
+      "step": 331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538709818313607e-05,
+      "loss": 0.4608,
+      "step": 332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6516895864550592e-05,
+      "loss": 0.4549,
+      "step": 333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495027873850364e-05,
+      "loss": 0.4655,
+      "step": 334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6473106027538396e-05,
+      "loss": 0.4378,
+      "step": 335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645113050738673e-05,
+      "loss": 0.4438,
+      "step": 336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429101495612453e-05,
+      "loss": 0.4732,
+      "step": 337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64070191748762e-05,
+      "loss": 0.4742,
+      "step": 338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638488372828063e-05,
+      "loss": 0.474,
+      "step": 339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362695339368912e-05,
+      "loss": 0.4575,
+      "step": 340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340454192123215e-05,
+      "loss": 0.4483,
+      "step": 341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631816047096316e-05,
+      "loss": 0.4384,
+      "step": 342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295814360744294e-05,
+      "loss": 0.4839,
+      "step": 343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6273416046756584e-05,
+      "loss": 0.4559,
+      "step": 344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250965714722852e-05,
+      "loss": 0.4595,
+      "step": 345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228463550797236e-05,
+      "loss": 0.4185,
+      "step": 346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6205909741563662e-05,
+      "loss": 0.488,
+      "step": 347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183304474034304e-05,
+      "loss": 0.4688,
+      "step": 348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160647935647987e-05,
+      "loss": 0.451,
+      "step": 349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6137940314268698e-05,
+      "loss": 0.481,
+      "step": 350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.611518179818397e-05,
+      "loss": 0.4441,
+      "step": 351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092372576103362e-05,
+      "loss": 0.4533,
+      "step": 352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069512837156867e-05,
+      "loss": 0.4347,
+      "step": 353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6046602770893364e-05,
+      "loss": 0.4774,
+      "step": 354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602364256727903e-05,
+      "loss": 0.4229,
+      "step": 355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000632416695783e-05,
+      "loss": 0.4723,
+      "step": 356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5977572509939673e-05,
+      "loss": 0.4332,
+      "step": 357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595446303821933e-05,
+      "loss": 0.4329,
+      "step": 358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931304193154377e-05,
+      "loss": 0.4917,
+      "step": 359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5908096166773818e-05,
+      "loss": 0.4061,
+      "step": 360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588483915151447e-05,
+      "loss": 0.4836,
+      "step": 361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586153334021935e-05,
+      "loss": 0.4574,
+      "step": 362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838178926136077e-05,
+      "loss": 0.4775,
+      "step": 363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814776102915293e-05,
+      "loss": 0.4752,
+      "step": 364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579132506460903e-05,
+      "loss": 0.4317,
+      "step": 365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.576782600566911e-05,
+      "loss": 0.4465,
+      "step": 366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744279120945534e-05,
+      "loss": 0.4523,
+      "step": 367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5720684605684868e-05,
+      "loss": 0.4296,
+      "step": 368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5697042655528617e-05,
+      "loss": 0.4438,
+      "step": 369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673353466511618e-05,
+      "loss": 0.4555,
+      "step": 370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5649617235060394e-05,
+      "loss": 0.4515,
+      "step": 371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5625834157991538e-05,
+      "loss": 0.4438,
+      "step": 372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602004432510085e-05,
+      "loss": 0.4272,
+      "step": 373
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557812825620786e-05,
+      "loss": 0.4575,
+      "step": 374
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5554205827061853e-05,
+      "loss": 0.4593,
+      "step": 375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553023734343258e-05,
+      "loss": 0.4487,
+      "step": 376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5506223004062428e-05,
+      "loss": 0.4173,
+      "step": 377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5482163008074017e-05,
+      "loss": 0.4314,
+      "step": 378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458057554968532e-05,
+      "loss": 0.4409,
+      "step": 379
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543390684462409e-05,
+      "loss": 0.419,
+      "step": 380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5409711077294066e-05,
+      "loss": 0.4238,
+      "step": 381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5385470453605457e-05,
+      "loss": 0.4309,
+      "step": 382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536118517455717e-05,
+      "loss": 0.4092,
+      "step": 383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336855441518413e-05,
+      "loss": 0.4221,
+      "step": 384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5312481456226986e-05,
+      "loss": 0.4075,
+      "step": 385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528806342078763e-05,
+      "loss": 0.4009,
+      "step": 386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5263601537670335e-05,
+      "loss": 0.4225,
+      "step": 387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239096009708665e-05,
+      "loss": 0.4454,
+      "step": 388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5214547040098082e-05,
+      "loss": 0.4019,
+      "step": 389
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189954832394267e-05,
+      "loss": 0.4311,
+      "step": 390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165319590511414e-05,
+      "loss": 0.4022,
+      "step": 391
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5140641518720547e-05,
+      "loss": 0.4193,
+      "step": 392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115920821647834e-05,
+      "loss": 0.4131,
+      "step": 393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091157704272886e-05,
+      "loss": 0.3862,
+      "step": 394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066352371927047e-05,
+      "loss": 0.4336,
+      "step": 395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.504150503029171e-05,
+      "loss": 0.3581,
+      "step": 396
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5016615885396591e-05,
+      "loss": 0.3885,
+      "step": 397
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991685143618043e-05,
+      "loss": 0.3908,
+      "step": 398
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4966713011677319e-05,
+      "loss": 0.3858,
+      "step": 399
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4941699696638887e-05,
+      "loss": 0.3836,
+      "step": 400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4916645405908683e-05,
+      "loss": 0.375,
+      "step": 401
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489155034723242e-05,
+      "loss": 0.3368,
+      "step": 402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866414728693839e-05,
+      "loss": 0.3738,
+      "step": 403
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841238758713008e-05,
+      "loss": 0.3942,
+      "step": 404
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4816022646044572e-05,
+      "loss": 0.3718,
+      "step": 405
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790766599776038e-05,
+      "loss": 0.3575,
+      "step": 406
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4765470829326034e-05,
+      "loss": 0.3432,
+      "step": 407
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474013554444257e-05,
+      "loss": 0.3755,
+      "step": 408
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714760955201308e-05,
+      "loss": 0.3403,
+      "step": 409
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689347272003813e-05,
+      "loss": 0.36,
+      "step": 410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.466389470557581e-05,
+      "loss": 0.3481,
+      "step": 411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463840346696543e-05,
+      "loss": 0.3609,
+      "step": 412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4612873767541485e-05,
+      "loss": 0.4068,
+      "step": 413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587305818991673e-05,
+      "loss": 0.3748,
+      "step": 414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4561699833320868e-05,
+      "loss": 0.4032,
+      "step": 415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536056022849328e-05,
+      "loss": 0.3706,
+      "step": 416
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510374600210955e-05,
+      "loss": 0.3459,
+      "step": 417
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448465577835152e-05,
+      "loss": 0.3861,
+      "step": 418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458899770526902e-05,
+      "loss": 0.381,
+      "step": 419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443310679030132e-05,
+      "loss": 0.3499,
+      "step": 420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407277051545565e-05,
+      "loss": 0.3642,
+      "step": 421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381410768435215e-05,
+      "loss": 0.3597,
+      "step": 422
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435550815544888e-05,
+      "loss": 0.3643,
+      "step": 423
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329569427366394e-05,
+      "loss": 0.3647,
+      "step": 424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4303594799267067e-05,
+      "loss": 0.357,
+      "step": 425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4277584486527873e-05,
+      "loss": 0.3767,
+      "step": 426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425153870482169e-05,
+      "loss": 0.3646,
+      "step": 427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225457670115483e-05,
+      "loss": 0.4045,
+      "step": 428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4199341598668539e-05,
+      "loss": 0.3537,
+      "step": 429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.417319070703066e-05,
+      "loss": 0.3381,
+      "step": 430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147005212040374e-05,
+      "loss": 0.3624,
+      "step": 431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120785330823128e-05,
+      "loss": 0.3629,
+      "step": 432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094531280789501e-05,
+      "loss": 0.3435,
+      "step": 433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4068243279633382e-05,
+      "loss": 0.3532,
+      "step": 434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041921545330192e-05,
+      "loss": 0.3551,
+      "step": 435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4015566296135048e-05,
+      "loss": 0.3452,
+      "step": 436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3989177750580975e-05,
+      "loss": 0.3869,
+      "step": 437
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962756127477077e-05,
+      "loss": 0.3562,
+      "step": 438
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936301645906744e-05,
+      "loss": 0.359,
+      "step": 439
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3909814525225812e-05,
+      "loss": 0.3792,
+      "step": 440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3883294985060756e-05,
+      "loss": 0.3399,
+      "step": 441
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3856743245306867e-05,
+      "loss": 0.3487,
+      "step": 442
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383015952612644e-05,
+      "loss": 0.3569,
+      "step": 443
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803544047946918e-05,
+      "loss": 0.3331,
+      "step": 444
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3776897031459104e-05,
+      "loss": 0.3416,
+      "step": 445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3750218697615297e-05,
+      "loss": 0.3528,
+      "step": 446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372350926762748e-05,
+      "loss": 0.3805,
+      "step": 447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696768962965481e-05,
+      "loss": 0.3529,
+      "step": 448
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3669998005355131e-05,
+      "loss": 0.3582,
+      "step": 449
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643196616776433e-05,
+      "loss": 0.3481,
+      "step": 450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3616365019461717e-05,
+      "loss": 0.3622,
+      "step": 451
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3589503435893794e-05,
+      "loss": 0.3737,
+      "step": 452
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3562612088804128e-05,
+      "loss": 0.353,
+      "step": 453
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535691201170959e-05,
+      "loss": 0.3741,
+      "step": 454
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3508740996217493e-05,
+      "loss": 0.3798,
+      "step": 455
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348176169741001e-05,
+      "loss": 0.3374,
+      "step": 456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345475352845604e-05,
+      "loss": 0.3831,
+      "step": 457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427716713302505e-05,
+      "loss": 0.3665,
+      "step": 458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400651476133846e-05,
+      "loss": 0.3636,
+      "step": 459
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3373558041370179e-05,
+      "loss": 0.3915,
+      "step": 460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3346436633665429e-05,
+      "loss": 0.3671,
+      "step": 461
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3319287477905465e-05,
+      "loss": 0.3482,
+      "step": 462
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292110799206243e-05,
+      "loss": 0.369,
+      "step": 463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264906822911931e-05,
+      "loss": 0.3595,
+      "step": 464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237675774593047e-05,
+      "loss": 0.3866,
+      "step": 465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210417880044579e-05,
+      "loss": 0.3772,
+      "step": 466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183133365284122e-05,
+      "loss": 0.3744,
+      "step": 467
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155822456550007e-05,
+      "loss": 0.3496,
+      "step": 468
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312848538029941e-05,
+      "loss": 0.3565,
+      "step": 469
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3101122363206489e-05,
+      "loss": 0.3918,
+      "step": 470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3073733632160494e-05,
+      "loss": 0.3601,
+      "step": 471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3046319414263902e-05,
+      "loss": 0.3498,
+      "step": 472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3018879936830504e-05,
+      "loss": 0.3446,
+      "step": 473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991415427383555e-05,
+      "loss": 0.3852,
+      "step": 474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2963926113653864e-05,
+      "loss": 0.3251,
+      "step": 475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936412223577914e-05,
+      "loss": 0.3487,
+      "step": 476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908873985295969e-05,
+      "loss": 0.3677,
+      "step": 477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2881311627150185e-05,
+      "loss": 0.3517,
+      "step": 478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2853725377682718e-05,
+      "loss": 0.3669,
+      "step": 479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826115465633822e-05,
+      "loss": 0.3528,
+      "step": 480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798482119939957e-05,
+      "loss": 0.3607,
+      "step": 481
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2770825569731895e-05,
+      "loss": 0.387,
+      "step": 482
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2743146044332818e-05,
+      "loss": 0.4097,
+      "step": 483
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2715443773256403e-05,
+      "loss": 0.3662,
+      "step": 484
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268771898620494e-05,
+      "loss": 0.367,
+      "step": 485
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2659971913067415e-05,
+      "loss": 0.3356,
+      "step": 486
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26322027839176e-05,
+      "loss": 0.361,
+      "step": 487
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2604411829012166e-05,
+      "loss": 0.3698,
+      "step": 488
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576599278788741e-05,
+      "loss": 0.3557,
+      "step": 489
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2548765363864038e-05,
+      "loss": 0.3743,
+      "step": 490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2520910315031895e-05,
+      "loss": 0.384,
+      "step": 491
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249303436326142e-05,
+      "loss": 0.3271,
+      "step": 492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2465137739695021e-05,
+      "loss": 0.3559,
+      "step": 493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2437220675646525e-05,
+      "loss": 0.3342,
+      "step": 494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409283402599239e-05,
+      "loss": 0.3518,
+      "step": 495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2381326152204049e-05,
+      "loss": 0.3674,
+      "step": 496
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235334915627748e-05,
+      "loss": 0.347,
+      "step": 497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325352646799796e-05,
+      "loss": 0.3599,
+      "step": 498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2297336855913046e-05,
+      "loss": 0.3667,
+      "step": 499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269302015919173e-05,
+      "loss": 0.3917,
+      "step": 500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2241248359278064e-05,
+      "loss": 0.3539,
+      "step": 501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2213176118605636e-05,
+      "loss": 0.398,
+      "step": 502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2185085526671893e-05,
+      "loss": 0.336,
+      "step": 503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156976816399013e-05,
+      "loss": 0.362,
+      "step": 504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2128850220859399e-05,
+      "loss": 0.3331,
+      "step": 505
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100705973273765e-05,
+      "loss": 0.3294,
+      "step": 506
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072544307009184e-05,
+      "loss": 0.3632,
+      "step": 507
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044365455577172e-05,
+      "loss": 0.3866,
+      "step": 508
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2016169652631725e-05,
+      "loss": 0.3412,
+      "step": 509
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987957131967418e-05,
+      "loss": 0.3603,
+      "step": 510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959728127517434e-05,
+      "loss": 0.3526,
+      "step": 511
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1931482873351631e-05,
+      "loss": 0.3416,
+      "step": 512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903221603674629e-05,
+      "loss": 0.3567,
+      "step": 513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874944552823818e-05,
+      "loss": 0.3783,
+      "step": 514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1846651955267463e-05,
+      "loss": 0.3499,
+      "step": 515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1818344045602728e-05,
+      "loss": 0.3561,
+      "step": 516
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790021058553751e-05,
+      "loss": 0.3491,
+      "step": 517
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1761683228969682e-05,
+      "loss": 0.3572,
+      "step": 518
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173333079182275e-05,
+      "loss": 0.3717,
+      "step": 519
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1704963982206298e-05,
+      "loss": 0.3386,
+      "step": 520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676583035332854e-05,
+      "loss": 0.3471,
+      "step": 521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1648188186532154e-05,
+      "loss": 0.3763,
+      "step": 522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619779671249222e-05,
+      "loss": 0.3417,
+      "step": 523
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591357725042394e-05,
+      "loss": 0.3646,
+      "step": 524
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1562922583581375e-05,
+      "loss": 0.3207,
+      "step": 525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1534474482645282e-05,
+      "loss": 0.3619,
+      "step": 526
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1506013658120687e-05,
+      "loss": 0.3685,
+      "step": 527
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147754034599967e-05,
+      "loss": 0.3485,
+      "step": 528
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449054782377854e-05,
+      "loss": 0.3677,
+      "step": 529
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420557203452445e-05,
+      "loss": 0.3693,
+      "step": 530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1392047845520283e-05,
+      "loss": 0.3598,
+      "step": 531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1363526944975867e-05,
+      "loss": 0.349,
+      "step": 532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1334994738309424e-05,
+      "loss": 0.3619,
+      "step": 533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1306451462104913e-05,
+      "loss": 0.384,
+      "step": 534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1277897353038085e-05,
+      "loss": 0.3546,
+      "step": 535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1249332647874514e-05,
+      "loss": 0.3582,
+      "step": 536
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1220757583467643e-05,
+      "loss": 0.3809,
+      "step": 537
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1192172396756798e-05,
+      "loss": 0.3796,
+      "step": 538
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1163577324765249e-05,
+      "loss": 0.3205,
+      "step": 539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134972604598224e-05,
+      "loss": 0.3704,
+      "step": 540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1106358473440964e-05,
+      "loss": 0.3545,
+      "step": 541
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1077735168556729e-05,
+      "loss": 0.3846,
+      "step": 542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049102927284858e-05,
+      "loss": 0.3524,
+      "step": 543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1020461987038781e-05,
+      "loss": 0.3649,
+      "step": 544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099181258530407e-05,
+      "loss": 0.3825,
+      "step": 545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.096315495963644e-05,
+      "loss": 0.3647,
+      "step": 546
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0934489347659816e-05,
+      "loss": 0.3334,
+      "step": 547
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905815987064328e-05,
+      "loss": 0.3594,
+      "step": 548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0877135115604372e-05,
+      "loss": 0.3864,
+      "step": 549
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0848446971096606e-05,
+      "loss": 0.3959,
+      "step": 550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0819751791418011e-05,
+      "loss": 0.316,
+      "step": 551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0791049814503888e-05,
+      "loss": 0.3843,
+      "step": 552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076234127834591e-05,
+      "loss": 0.3615,
+      "step": 553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0733626420990135e-05,
+      "loss": 0.3497,
+      "step": 554
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070490548053503e-05,
+      "loss": 0.3639,
+      "step": 555
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0676178695129514e-05,
+      "loss": 0.373,
+      "step": 556
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0647446302970954e-05,
+      "loss": 0.335,
+      "step": 557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0618708542303226e-05,
+      "loss": 0.3715,
+      "step": 558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589965651414704e-05,
+      "loss": 0.3475,
+      "step": 559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0561217868636314e-05,
+      "loss": 0.3394,
+      "step": 560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0532465432339539e-05,
+      "loss": 0.3362,
+      "step": 561
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503708580934443e-05,
+      "loss": 0.3487,
+      "step": 562
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0474947552867707e-05,
+      "loss": 0.3599,
+      "step": 563
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0446182586620647e-05,
+      "loss": 0.3511,
+      "step": 564
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0417413920707223e-05,
+      "loss": 0.3641,
+      "step": 565
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388641793672079e-05,
+      "loss": 0.3545,
+      "step": 566
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359866444088555e-05,
+      "loss": 0.3526,
+      "step": 567
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0331088110556718e-05,
+      "loss": 0.3401,
+      "step": 568
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0302307031701364e-05,
+      "loss": 0.3876,
+      "step": 569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027352344617007e-05,
+      "loss": 0.3717,
+      "step": 570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244737592631182e-05,
+      "loss": 0.3404,
+      "step": 571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0215949709771867e-05,
+      "loss": 0.3418,
+      "step": 572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187160036296104e-05,
+      "loss": 0.3805,
+      "step": 573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015836881092273e-05,
+      "loss": 0.368,
+      "step": 574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129576272383446e-05,
+      "loss": 0.3503,
+      "step": 575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.010078265942084e-05,
+      "loss": 0.3781,
+      "step": 576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071988210786412e-05,
+      "loss": 0.3618,
+      "step": 577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043193165238591e-05,
+      "loss": 0.3697,
+      "step": 578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014397761540755e-05,
+      "loss": 0.3701,
+      "step": 579
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.985602238459248e-06,
+      "loss": 0.3829,
+      "step": 580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956806834761412e-06,
+      "loss": 0.3383,
+      "step": 581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928011789213591e-06,
+      "loss": 0.3619,
+      "step": 582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899217340579165e-06,
+      "loss": 0.3566,
+      "step": 583
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.870423727616558e-06,
+      "loss": 0.4108,
+      "step": 584
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84163118907727e-06,
+      "loss": 0.3615,
+      "step": 585
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8128399637039e-06,
+      "loss": 0.3525,
+      "step": 586
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.784050290228135e-06,
+      "loss": 0.3677,
+      "step": 587
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755262407368821e-06,
+      "loss": 0.3812,
+      "step": 588
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.726476553829933e-06,
+      "loss": 0.3659,
+      "step": 589
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69769296829864e-06,
+      "loss": 0.3961,
+      "step": 590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.668911889443285e-06,
+      "loss": 0.3662,
+      "step": 591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64013355591145e-06,
+      "loss": 0.368,
+      "step": 592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611358206327923e-06,
+      "loss": 0.3765,
+      "step": 593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.582586079292779e-06,
+      "loss": 0.3394,
+      "step": 594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.553817413379356e-06,
+      "loss": 0.3745,
+      "step": 595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.525052447132294e-06,
+      "loss": 0.3529,
+      "step": 596
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49629141906556e-06,
+      "loss": 0.3875,
+      "step": 597
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467534567660466e-06,
+      "loss": 0.3686,
+      "step": 598
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.43878213136369e-06,
+      "loss": 0.3617,
+      "step": 599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410034348585299e-06,
+      "loss": 0.3361,
+      "step": 600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381291457696779e-06,
+      "loss": 0.3891,
+      "step": 601
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.352553697029049e-06,
+      "loss": 0.3516,
+      "step": 602
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32382130487049e-06,
+      "loss": 0.3555,
+      "step": 603
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295094519464971e-06,
+      "loss": 0.3973,
+      "step": 604
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266373579009867e-06,
+      "loss": 0.333,
+      "step": 605
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.237658721654092e-06,
+      "loss": 0.3688,
+      "step": 606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208950185496115e-06,
+      "loss": 0.3533,
+      "step": 607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180248208581994e-06,
+      "loss": 0.3495,
+      "step": 608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.151553028903397e-06,
+      "loss": 0.3703,
+      "step": 609
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.122864884395633e-06,
+      "loss": 0.3664,
+      "step": 610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094184012935675e-06,
+      "loss": 0.3618,
+      "step": 611
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065510652340185e-06,
+      "loss": 0.3529,
+      "step": 612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.036845040363564e-06,
+      "loss": 0.3767,
+      "step": 613
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.008187414695932e-06,
+      "loss": 0.3685,
+      "step": 614
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979538012961222e-06,
+      "loss": 0.3732,
+      "step": 615
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950897072715145e-06,
+      "loss": 0.3783,
+      "step": 616
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.922264831443274e-06,
+      "loss": 0.3948,
+      "step": 617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89364152655904e-06,
+      "loss": 0.4019,
+      "step": 618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86502739540178e-06,
+      "loss": 0.3588,
+      "step": 619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836422675234754e-06,
+      "loss": 0.345,
+      "step": 620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.807827603243204e-06,
+      "loss": 0.3887,
+      "step": 621
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77924241653236e-06,
+      "loss": 0.4192,
+      "step": 622
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.750667352125486e-06,
+      "loss": 0.3496,
+      "step": 623
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72210264696192e-06,
+      "loss": 0.3699,
+      "step": 624
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.693548537895092e-06,
+      "loss": 0.3719,
+      "step": 625
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66500526169058e-06,
+      "loss": 0.3846,
+      "step": 626
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636473055024135e-06,
+      "loss": 0.3855,
+      "step": 627
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607952154479724e-06,
+      "loss": 0.3506,
+      "step": 628
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.579442796547559e-06,
+      "loss": 0.3501,
+      "step": 629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550945217622147e-06,
+      "loss": 0.3542,
+      "step": 630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522459654000332e-06,
+      "loss": 0.3578,
+      "step": 631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.493986341879315e-06,
+      "loss": 0.3439,
+      "step": 632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465525517354724e-06,
+      "loss": 0.3969,
+      "step": 633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.437077416418627e-06,
+      "loss": 0.377,
+      "step": 634
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408642274957612e-06,
+      "loss": 0.3705,
+      "step": 635
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.380220328750781e-06,
+      "loss": 0.3564,
+      "step": 636
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35181181346785e-06,
+      "loss": 0.3719,
+      "step": 637
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323416964667151e-06,
+      "loss": 0.3621,
+      "step": 638
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295036017793702e-06,
+      "loss": 0.3879,
+      "step": 639
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.266669208177251e-06,
+      "loss": 0.3928,
+      "step": 640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.238316771030318e-06,
+      "loss": 0.3759,
+      "step": 641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209978941446252e-06,
+      "loss": 0.3604,
+      "step": 642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181655954397275e-06,
+      "loss": 0.3603,
+      "step": 643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.153348044732543e-06,
+      "loss": 0.3574,
+      "step": 644
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125055447176187e-06,
+      "loss": 0.3875,
+      "step": 645
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096778396325378e-06,
+      "loss": 0.3759,
+      "step": 646
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.06851712664837e-06,
+      "loss": 0.3633,
+      "step": 647
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04027187248257e-06,
+      "loss": 0.3665,
+      "step": 648
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012042868032585e-06,
+      "loss": 0.3789,
+      "step": 649
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983830347368276e-06,
+      "loss": 0.3441,
+      "step": 650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.955634544422834e-06,
+      "loss": 0.377,
+      "step": 651
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927455692990819e-06,
+      "loss": 0.36,
+      "step": 652
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89929402672624e-06,
+      "loss": 0.3968,
+      "step": 653
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871149779140604e-06,
+      "loss": 0.345,
+      "step": 654
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.843023183600989e-06,
+      "loss": 0.3758,
+      "step": 655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81491447332811e-06,
+      "loss": 0.3538,
+      "step": 656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.786823881394364e-06,
+      "loss": 0.3591,
+      "step": 657
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.758751640721937e-06,
+      "loss": 0.4011,
+      "step": 658
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.730697984080827e-06,
+      "loss": 0.3575,
+      "step": 659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702663144086957e-06,
+      "loss": 0.3689,
+      "step": 660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674647353200208e-06,
+      "loss": 0.3628,
+      "step": 661
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.646650843722521e-06,
+      "loss": 0.3697,
+      "step": 662
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.618673847795954e-06,
+      "loss": 0.3907,
+      "step": 663
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590716597400761e-06,
+      "loss": 0.3581,
+      "step": 664
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.562779324353478e-06,
+      "loss": 0.3412,
+      "step": 665
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5348622603049805e-06,
+      "loss": 0.3559,
+      "step": 666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506965636738582e-06,
+      "loss": 0.3714,
+      "step": 667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.479089684968106e-06,
+      "loss": 0.4004,
+      "step": 668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451234636135969e-06,
+      "loss": 0.3394,
+      "step": 669
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4234007212112605e-06,
+      "loss": 0.3565,
+      "step": 670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.395588170987839e-06,
+      "loss": 0.4318,
+      "step": 671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367797216082402e-06,
+      "loss": 0.3668,
+      "step": 672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.340028086932587e-06,
+      "loss": 0.3765,
+      "step": 673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.312281013795064e-06,
+      "loss": 0.3583,
+      "step": 674
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2845562267435974e-06,
+      "loss": 0.3845,
+      "step": 675
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2568539556671865e-06,
+      "loss": 0.3638,
+      "step": 676
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2291744302681045e-06,
+      "loss": 0.3596,
+      "step": 677
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.201517880060048e-06,
+      "loss": 0.3734,
+      "step": 678
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1738845343661826e-06,
+      "loss": 0.4091,
+      "step": 679
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1462746223172885e-06,
+      "loss": 0.3682,
+      "step": 680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.118688372849815e-06,
+      "loss": 0.3885,
+      "step": 681
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091126014704032e-06,
+      "loss": 0.3951,
+      "step": 682
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063587776422088e-06,
+      "loss": 0.3679,
+      "step": 683
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036073886346137e-06,
+      "loss": 0.3758,
+      "step": 684
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.008584572616448e-06,
+      "loss": 0.3593,
+      "step": 685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9811200631694985e-06,
+      "loss": 0.3729,
+      "step": 686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.953680585736105e-06,
+      "loss": 0.3955,
+      "step": 687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9262663678395085e-06,
+      "loss": 0.388,
+      "step": 688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.898877636793517e-06,
+      "loss": 0.3882,
+      "step": 689
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8715146197005945e-06,
+      "loss": 0.3757,
+      "step": 690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844177543449997e-06,
+      "loss": 0.3563,
+      "step": 691
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.816866634715881e-06,
+      "loss": 0.3992,
+      "step": 692
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.789582119955425e-06,
+      "loss": 0.3939,
+      "step": 693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7623242254069575e-06,
+      "loss": 0.3638,
+      "step": 694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7350931770880705e-06,
+      "loss": 0.4099,
+      "step": 695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.70788920079376e-06,
+      "loss": 0.3678,
+      "step": 696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680712522094538e-06,
+      "loss": 0.4117,
+      "step": 697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653563366334577e-06,
+      "loss": 0.3731,
+      "step": 698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6264419586298255e-06,
+      "loss": 0.3618,
+      "step": 699
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.599348523866155e-06,
+      "loss": 0.3835,
+      "step": 700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572283286697496e-06,
+      "loss": 0.3751,
+      "step": 701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5452464715439605e-06,
+      "loss": 0.3827,
+      "step": 702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518238302589994e-06,
+      "loss": 0.3789,
+      "step": 703
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.491259003782511e-06,
+      "loss": 0.4002,
+      "step": 704
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464308798829042e-06,
+      "loss": 0.3907,
+      "step": 705
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437387911195876e-06,
+      "loss": 0.3809,
+      "step": 706
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4104965641062074e-06,
+      "loss": 0.404,
+      "step": 707
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.3836349805382865e-06,
+      "loss": 0.358,
+      "step": 708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356803383223568e-06,
+      "loss": 0.3792,
+      "step": 709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.330001994644871e-06,
+      "loss": 0.3623,
+      "step": 710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.303231037034521e-06,
+      "loss": 0.3888,
+      "step": 711
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276490732372522e-06,
+      "loss": 0.3797,
+      "step": 712
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2497813023847054e-06,
+      "loss": 0.4012,
+      "step": 713
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2231029685409e-06,
+      "loss": 0.4031,
+      "step": 714
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.196455952053085e-06,
+      "loss": 0.3845,
+      "step": 715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.169840473873567e-06,
+      "loss": 0.3792,
+      "step": 716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143256754693134e-06,
+      "loss": 0.3564,
+      "step": 717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116705014939246e-06,
+      "loss": 0.3938,
+      "step": 718
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.090185474774192e-06,
+      "loss": 0.3867,
+      "step": 719
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063698354093255e-06,
+      "loss": 0.3824,
+      "step": 720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0372438725229246e-06,
+      "loss": 0.3917,
+      "step": 721
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.010822249419027e-06,
+      "loss": 0.3859,
+      "step": 722
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.984433703864956e-06,
+      "loss": 0.394,
+      "step": 723
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.95807845466981e-06,
+      "loss": 0.3943,
+      "step": 724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9317567203666215e-06,
+      "loss": 0.3872,
+      "step": 725
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.905468719210503e-06,
+      "loss": 0.4103,
+      "step": 726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8792146691768725e-06,
+      "loss": 0.3768,
+      "step": 727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8529947879596286e-06,
+      "loss": 0.3741,
+      "step": 728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.826809292969341e-06,
+      "loss": 0.3668,
+      "step": 729
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.800658401331467e-06,
+      "loss": 0.3932,
+      "step": 730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7745423298845185e-06,
+      "loss": 0.4122,
+      "step": 731
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748461295178315e-06,
+      "loss": 0.388,
+      "step": 732
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.722415513472128e-06,
+      "loss": 0.3617,
+      "step": 733
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.69640520073294e-06,
+      "loss": 0.3927,
+      "step": 734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670430572633607e-06,
+      "loss": 0.3972,
+      "step": 735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.644491844551121e-06,
+      "loss": 0.3812,
+      "step": 736
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.618589231564786e-06,
+      "loss": 0.4102,
+      "step": 737
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.592722948454438e-06,
+      "loss": 0.3912,
+      "step": 738
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5668932096986805e-06,
+      "loss": 0.4056,
+      "step": 739
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5411002294731e-06,
+      "loss": 0.4239,
+      "step": 740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.515344221648484e-06,
+      "loss": 0.3637,
+      "step": 741
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489625399789048e-06,
+      "loss": 0.3889,
+      "step": 742
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4639439771506745e-06,
+      "loss": 0.3952,
+      "step": 743
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438300166679135e-06,
+      "loss": 0.3652,
+      "step": 744
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.412694181008329e-06,
+      "loss": 0.3835,
+      "step": 745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387126232458522e-06,
+      "loss": 0.3762,
+      "step": 746
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361596533034572e-06,
+      "loss": 0.3757,
+      "step": 747
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3361052944241945e-06,
+      "loss": 0.3815,
+      "step": 748
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.310652727996189e-06,
+      "loss": 0.3605,
+      "step": 749
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285239044798696e-06,
+      "loss": 0.3822,
+      "step": 750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.259864455557433e-06,
+      "loss": 0.395,
+      "step": 751
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.234529170673969e-06,
+      "loss": 0.3952,
+      "step": 752
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.209233400223963e-06,
+      "loss": 0.3711,
+      "step": 753
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.183977353955427e-06,
+      "loss": 0.3617,
+      "step": 754
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158761241286996e-06,
+      "loss": 0.3902,
+      "step": 755
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.13358527130616e-06,
+      "loss": 0.3529,
+      "step": 756
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108449652767584e-06,
+      "loss": 0.3755,
+      "step": 757
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.083354594091318e-06,
+      "loss": 0.3682,
+      "step": 758
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058300303361117e-06,
+      "loss": 0.3606,
+      "step": 759
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.033286988322682e-06,
+      "loss": 0.3567,
+      "step": 760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008314856381962e-06,
+      "loss": 0.3396,
+      "step": 761
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.983384114603412e-06,
+      "loss": 0.3463,
+      "step": 762
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.958494969708294e-06,
+      "loss": 0.3861,
+      "step": 763
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.933647628072954e-06,
+      "loss": 0.3678,
+      "step": 764
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908842295727116e-06,
+      "loss": 0.3563,
+      "step": 765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88407917835217e-06,
+      "loss": 0.3506,
+      "step": 766
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.859358481279458e-06,
+      "loss": 0.3471,
+      "step": 767
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.83468040948859e-06,
+      "loss": 0.3471,
+      "step": 768
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.810045167605735e-06,
+      "loss": 0.3468,
+      "step": 769
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.785452959901922e-06,
+      "loss": 0.3487,
+      "step": 770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.76090399029134e-06,
+      "loss": 0.3685,
+      "step": 771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.736398462329664e-06,
+      "loss": 0.3354,
+      "step": 772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711936579212369e-06,
+      "loss": 0.3342,
+      "step": 773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687518543773013e-06,
+      "loss": 0.3391,
+      "step": 774
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.663144558481593e-06,
+      "loss": 0.3413,
+      "step": 775
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6388148254428325e-06,
+      "loss": 0.3022,
+      "step": 776
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6145295463945495e-06,
+      "loss": 0.3165,
+      "step": 777
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590288922705934e-06,
+      "loss": 0.3116,
+      "step": 778
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.566093155375916e-06,
+      "loss": 0.3145,
+      "step": 779
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54194244503147e-06,
+      "loss": 0.3338,
+      "step": 780
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.517836991925983e-06,
+      "loss": 0.3152,
+      "step": 781
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.493776995937573e-06,
+      "loss": 0.299,
+      "step": 782
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.469762656567422e-06,
+      "loss": 0.3142,
+      "step": 783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4457941729381494e-06,
+      "loss": 0.3073,
+      "step": 784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421871743792144e-06,
+      "loss": 0.3062,
+      "step": 785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.397995567489921e-06,
+      "loss": 0.299,
+      "step": 786
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374165842008464e-06,
+      "loss": 0.324,
+      "step": 787
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350382764939609e-06,
+      "loss": 0.3201,
+      "step": 788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326646533488384e-06,
+      "loss": 0.3447,
+      "step": 789
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3029573444713836e-06,
+      "loss": 0.3245,
+      "step": 790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.279315394315138e-06,
+      "loss": 0.3371,
+      "step": 791
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255720879054469e-06,
+      "loss": 0.2931,
+      "step": 792
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.232173994330893e-06,
+      "loss": 0.312,
+      "step": 793
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.20867493539097e-06,
+      "loss": 0.3375,
+      "step": 794
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185223897084709e-06,
+      "loss": 0.3169,
+      "step": 795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1618210738639255e-06,
+      "loss": 0.297,
+      "step": 796
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.138466659780652e-06,
+      "loss": 0.3078,
+      "step": 797
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.115160848485532e-06,
+      "loss": 0.3163,
+      "step": 798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091903833226182e-06,
+      "loss": 0.3183,
+      "step": 799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068695806845624e-06,
+      "loss": 0.3112,
+      "step": 800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.045536961780671e-06,
+      "loss": 0.3102,
+      "step": 801
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.022427490060334e-06,
+      "loss": 0.3317,
+      "step": 802
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999367583304222e-06,
+      "loss": 0.3374,
+      "step": 803
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9763574327209685e-06,
+      "loss": 0.3471,
+      "step": 804
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.953397229106636e-06,
+      "loss": 0.2861,
+      "step": 805
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930487162843132e-06,
+      "loss": 0.3226,
+      "step": 806
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907627423896642e-06,
+      "loss": 0.304,
+      "step": 807
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8848182018160314e-06,
+      "loss": 0.3036,
+      "step": 808
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8620596857313046e-06,
+      "loss": 0.315,
+      "step": 809
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839352064352012e-06,
+      "loss": 0.3069,
+      "step": 810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816695525965702e-06,
+      "loss": 0.3043,
+      "step": 811
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7940902584363383e-06,
+      "loss": 0.3193,
+      "step": 812
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7715364492027685e-06,
+      "loss": 0.3638,
+      "step": 813
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7490342852771523e-06,
+      "loss": 0.3145,
+      "step": 814
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7265839532434154e-06,
+      "loss": 0.3093,
+      "step": 815
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7041856392557084e-06,
+      "loss": 0.3202,
+      "step": 816
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6818395290368415e-06,
+      "loss": 0.3114,
+      "step": 817
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6595458078767864e-06,
+      "loss": 0.3101,
+      "step": 818
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.637304660631088e-06,
+      "loss": 0.3172,
+      "step": 819
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6151162717193776e-06,
+      "loss": 0.2899,
+      "step": 820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.592980825123803e-06,
+      "loss": 0.2858,
+      "step": 821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.570898504387551e-06,
+      "loss": 0.3246,
+      "step": 822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.548869492613273e-06,
+      "loss": 0.343,
+      "step": 823
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5268939724616056e-06,
+      "loss": 0.3068,
+      "step": 824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5049721261496396e-06,
+      "loss": 0.3151,
+      "step": 825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48310413544941e-06,
+      "loss": 0.3007,
+      "step": 826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4612901816863975e-06,
+      "loss": 0.3185,
+      "step": 827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.439530445738004e-06,
+      "loss": 0.3179,
+      "step": 828
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.417825108032081e-06,
+      "loss": 0.3139,
+      "step": 829
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3961743485454133e-06,
+      "loss": 0.332,
+      "step": 830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3745783468022387e-06,
+      "loss": 0.3303,
+      "step": 831
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3530372818727395e-06,
+      "loss": 0.3038,
+      "step": 832
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3315513323715854e-06,
+      "loss": 0.3491,
+      "step": 833
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.310120676456434e-06,
+      "loss": 0.3062,
+      "step": 834
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2887454918264593e-06,
+      "loss": 0.3483,
+      "step": 835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.267425955720881e-06,
+      "loss": 0.3393,
+      "step": 836
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.246162244917478e-06,
+      "loss": 0.2983,
+      "step": 837
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2249545357311486e-06,
+      "loss": 0.3135,
+      "step": 838
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2038030040124324e-06,
+      "loss": 0.3291,
+      "step": 839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1827078251460564e-06,
+      "loss": 0.343,
+      "step": 840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1616691740494743e-06,
+      "loss": 0.325,
+      "step": 841
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1406872251714238e-06,
+      "loss": 0.3412,
+      "step": 842
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1197621524904885e-06,
+      "loss": 0.2988,
+      "step": 843
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.098894129513629e-06,
+      "loss": 0.3147,
+      "step": 844
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078083329274767e-06,
+      "loss": 0.3313,
+      "step": 845
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0573299243333452e-06,
+      "loss": 0.3418,
+      "step": 846
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0366340867728947e-06,
+      "loss": 0.3043,
+      "step": 847
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0159959881996e-06,
+      "loss": 0.304,
+      "step": 848
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9954157997408963e-06,
+      "loss": 0.3172,
+      "step": 849
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9748936920440284e-06,
+      "loss": 0.3181,
+      "step": 850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9544298352746525e-06,
+      "loss": 0.3125,
+      "step": 851
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9340243991154195e-06,
+      "loss": 0.3025,
+      "step": 852
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9136775527645557e-06,
+      "loss": 0.3255,
+      "step": 853
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.893389464934484e-06,
+      "loss": 0.323,
+      "step": 854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8731603038504054e-06,
+      "loss": 0.2961,
+      "step": 855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8529902372489172e-06,
+      "loss": 0.3417,
+      "step": 856
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.832879432376604e-06,
+      "loss": 0.3071,
+      "step": 857
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.812828055988673e-06,
+      "loss": 0.3535,
+      "step": 858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.792836274347559e-06,
+      "loss": 0.3725,
+      "step": 859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7729042532215455e-06,
+      "loss": 0.3285,
+      "step": 860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7530321578833976e-06,
+      "loss": 0.2958,
+      "step": 861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.733220153108973e-06,
+      "loss": 0.3065,
+      "step": 862
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.713468403175891e-06,
+      "loss": 0.3321,
+      "step": 863
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6937770718621282e-06,
+      "loss": 0.3268,
+      "step": 864
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6741463224446928e-06,
+      "loss": 0.3208,
+      "step": 865
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6545763176982477e-06,
+      "loss": 0.3597,
+      "step": 866
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.635067219893789e-06,
+      "loss": 0.3329,
+      "step": 867
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6156191907972663e-06,
+      "loss": 0.2893,
+      "step": 868
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5962323916682676e-06,
+      "loss": 0.3174,
+      "step": 869
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.576906983258669e-06,
+      "loss": 0.2937,
+      "step": 870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.55764312581131e-06,
+      "loss": 0.3285,
+      "step": 871
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5384409790586594e-06,
+      "loss": 0.3368,
+      "step": 872
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5193007022214855e-06,
+      "loss": 0.3104,
+      "step": 873
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.500222454007548e-06,
+      "loss": 0.313,
+      "step": 874
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.481206392610278e-06,
+      "loss": 0.3408,
+      "step": 875
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4622526757074628e-06,
+      "loss": 0.3203,
+      "step": 876
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4433614604599365e-06,
+      "loss": 0.3286,
+      "step": 877
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424532903510284e-06,
+      "loss": 0.3428,
+      "step": 878
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4057671609815403e-06,
+      "loss": 0.2987,
+      "step": 879
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3870643884758915e-06,
+      "loss": 0.312,
+      "step": 880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.368424741073392e-06,
+      "loss": 0.3027,
+      "step": 881
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3498483733306655e-06,
+      "loss": 0.3286,
+      "step": 882
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3313354392796394e-06,
+      "loss": 0.304,
+      "step": 883
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.31288609242626e-06,
+      "loss": 0.3451,
+      "step": 884
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294500485749218e-06,
+      "loss": 0.2977,
+      "step": 885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.276178771698676e-06,
+      "loss": 0.3225,
+      "step": 886
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2579211021950244e-06,
+      "loss": 0.304,
+      "step": 887
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.239727628627599e-06,
+      "loss": 0.3381,
+      "step": 888
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22159850185343e-06,
+      "loss": 0.3395,
+      "step": 889
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2035338721960032e-06,
+      "loss": 0.3172,
+      "step": 890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1855338894440025e-06,
+      "loss": 0.322,
+      "step": 891
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1675987028500746e-06,
+      "loss": 0.3119,
+      "step": 892
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.14972846112958e-06,
+      "loss": 0.3101,
+      "step": 893
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.131923312459376e-06,
+      "loss": 0.3491,
+      "step": 894
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1141834044765775e-06,
+      "loss": 0.3011,
+      "step": 895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.096508884277334e-06,
+      "loss": 0.3209,
+      "step": 896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.078899898415616e-06,
+      "loss": 0.3288,
+      "step": 897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.061356592901984e-06,
+      "loss": 0.327,
+      "step": 898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0438791132023985e-06,
+      "loss": 0.3279,
+      "step": 899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0264676042370026e-06,
+      "loss": 0.3216,
+      "step": 900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009122210378922e-06,
+      "loss": 0.2863,
+      "step": 901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9918430754530625e-06,
+      "loss": 0.3618,
+      "step": 902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.974630342734928e-06,
+      "loss": 0.3196,
+      "step": 903
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9574841549494263e-06,
+      "loss": 0.3463,
+      "step": 904
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.940404654269684e-06,
+      "loss": 0.3143,
+      "step": 905
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9233919823158777e-06,
+      "loss": 0.3457,
+      "step": 906
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9064462801540362e-06,
+      "loss": 0.3251,
+      "step": 907
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8895676882949076e-06,
+      "loss": 0.3347,
+      "step": 908
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8727563466927568e-06,
+      "loss": 0.312,
+      "step": 909
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.85601239474423e-06,
+      "loss": 0.3576,
+      "step": 910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8393359712871817e-06,
+      "loss": 0.316,
+      "step": 911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.822727214599549e-06,
+      "loss": 0.3435,
+      "step": 912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8061862623981685e-06,
+      "loss": 0.3647,
+      "step": 913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.789713251837667e-06,
+      "loss": 0.3082,
+      "step": 914
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.773308319509306e-06,
+      "loss": 0.3001,
+      "step": 915
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7569716014398586e-06,
+      "loss": 0.3531,
+      "step": 916
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7407032330904749e-06,
+      "loss": 0.3461,
+      "step": 917
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7245033493555575e-06,
+      "loss": 0.3254,
+      "step": 918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7083720845616536e-06,
+      "loss": 0.3253,
+      "step": 919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6923095724663296e-06,
+      "loss": 0.3445,
+      "step": 920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.67631594625707e-06,
+      "loss": 0.3411,
+      "step": 921
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6603913385501703e-06,
+      "loss": 0.325,
+      "step": 922
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6445358813896296e-06,
+      "loss": 0.3124,
+      "step": 923
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6287497062460722e-06,
+      "loss": 0.3396,
+      "step": 924
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6130329440156433e-06,
+      "loss": 0.3731,
+      "step": 925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5973857250189328e-06,
+      "loss": 0.3225,
+      "step": 926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5818081789998807e-06,
+      "loss": 0.327,
+      "step": 927
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5663004351247257e-06,
+      "loss": 0.3165,
+      "step": 928
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5508626219809064e-06,
+      "loss": 0.3545,
+      "step": 929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5354948675760161e-06,
+      "loss": 0.3129,
+      "step": 930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5201972993367309e-06,
+      "loss": 0.3221,
+      "step": 931
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5049700441077542e-06,
+      "loss": 0.3398,
+      "step": 932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4898132281507727e-06,
+      "loss": 0.3145,
+      "step": 933
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4747269771433935e-06,
+      "loss": 0.3444,
+      "step": 934
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.459711416178119e-06,
+      "loss": 0.3136,
+      "step": 935
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4447666697613005e-06,
+      "loss": 0.3243,
+      "step": 936
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4298928618121099e-06,
+      "loss": 0.3147,
+      "step": 937
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4150901156615027e-06,
+      "loss": 0.3106,
+      "step": 938
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.400358554051211e-06,
+      "loss": 0.3464,
+      "step": 939
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.385698299132713e-06,
+      "loss": 0.3235,
+      "step": 940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3711094724662255e-06,
+      "loss": 0.3356,
+      "step": 941
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.356592195019698e-06,
+      "loss": 0.3323,
+      "step": 942
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3421465871677997e-06,
+      "loss": 0.2936,
+      "step": 943
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3277727686909325e-06,
+      "loss": 0.3429,
+      "step": 944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3134708587742361e-06,
+      "loss": 0.3546,
+      "step": 945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.299240976006594e-06,
+      "loss": 0.3291,
+      "step": 946
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2850832383796518e-06,
+      "loss": 0.323,
+      "step": 947
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.270997763286841e-06,
+      "loss": 0.3503,
+      "step": 948
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.256984667522414e-06,
+      "loss": 0.3249,
+      "step": 949
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2430440672804544e-06,
+      "loss": 0.3462,
+      "step": 950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.229176078153933e-06,
+      "loss": 0.358,
+      "step": 951
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2153808151337342e-06,
+      "loss": 0.3216,
+      "step": 952
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2016583926077264e-06,
+      "loss": 0.3241,
+      "step": 953
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1880089243597792e-06,
+      "loss": 0.3528,
+      "step": 954
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1744325235688536e-06,
+      "loss": 0.3489,
+      "step": 955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1609293028080325e-06,
+      "loss": 0.3329,
+      "step": 956
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1474993740436202e-06,
+      "loss": 0.3329,
+      "step": 957
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1341428486341876e-06,
+      "loss": 0.3175,
+      "step": 958
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1208598373296564e-06,
+      "loss": 0.3784,
+      "step": 959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1076504502703867e-06,
+      "loss": 0.3414,
+      "step": 960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0945147969862567e-06,
+      "loss": 0.3416,
+      "step": 961
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0814529863957625e-06,
+      "loss": 0.3587,
+      "step": 962
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0684651268051005e-06,
+      "loss": 0.336,
+      "step": 963
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0555513259072869e-06,
+      "loss": 0.3422,
+      "step": 964
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042711690781254e-06,
+      "loss": 0.3609,
+      "step": 965
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0299463278909638e-06,
+      "loss": 0.3378,
+      "step": 966
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.017255343084529e-06,
+      "loss": 0.3576,
+      "step": 967
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0046388415933227e-06,
+      "loss": 0.3501,
+      "step": 968
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.920969280311332e-07,
+      "loss": 0.3347,
+      "step": 969
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.796297063932614e-07,
+      "loss": 0.3037,
+      "step": 970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.672372800556883e-07,
+      "loss": 0.3624,
+      "step": 971
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.549197517741959e-07,
+      "loss": 0.3386,
+      "step": 972
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.426772236835369e-07,
+      "loss": 0.355,
+      "step": 973
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.305097972965626e-07,
+      "loss": 0.3442,
+      "step": 974
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.18417573503404e-07,
+      "loss": 0.321,
+      "step": 975
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.064006525706215e-07,
+      "loss": 0.3338,
+      "step": 976
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.944591341403775e-07,
+      "loss": 0.3494,
+      "step": 977
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.825931172296143e-07,
+      "loss": 0.3363,
+      "step": 978
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.708027002292219e-07,
+      "loss": 0.3397,
+      "step": 979
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.590879809032348e-07,
+      "loss": 0.3389,
+      "step": 980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.474490563880155e-07,
+      "loss": 0.3248,
+      "step": 981
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.358860231914501e-07,
+      "loss": 0.3459,
+      "step": 982
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.243989771921413e-07,
+      "loss": 0.3242,
+      "step": 983
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.129880136386259e-07,
+      "loss": 0.3559,
+      "step": 984
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.016532271485788e-07,
+      "loss": 0.3431,
+      "step": 985
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.903947117080235e-07,
+      "loss": 0.3296,
+      "step": 986
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.792125606705636e-07,
+      "loss": 0.3555,
+      "step": 987
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.681068667565961e-07,
+      "loss": 0.3219,
+      "step": 988
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.570777220525538e-07,
+      "loss": 0.3638,
+      "step": 989
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.461252180101352e-07,
+      "loss": 0.3363,
+      "step": 990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.352494454455517e-07,
+      "loss": 0.3665,
+      "step": 991
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.244504945387631e-07,
+      "loss": 0.3459,
+      "step": 992
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.13728454832745e-07,
+      "loss": 0.3722,
+      "step": 993
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.030834152327393e-07,
+      "loss": 0.3614,
+      "step": 994
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.925154640055109e-07,
+      "loss": 0.3359,
+      "step": 995
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.820246887786275e-07,
+      "loss": 0.3284,
+      "step": 996
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.716111765397193e-07,
+      "loss": 0.3922,
+      "step": 997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.612750136357782e-07,
+      "loss": 0.3705,
+      "step": 998
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.510162857724167e-07,
+      "loss": 0.329,
+      "step": 999
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.408350780131778e-07,
+      "loss": 0.3419,
+      "step": 1000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.307314747788163e-07,
+      "loss": 0.361,
+      "step": 1001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.207055598466094e-07,
+      "loss": 0.379,
+      "step": 1002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.107574163496532e-07,
+      "loss": 0.3538,
+      "step": 1003
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.008871267761762e-07,
+      "loss": 0.3212,
+      "step": 1004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.910947729688598e-07,
+      "loss": 0.3204,
+      "step": 1005
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.813804361241516e-07,
+      "loss": 0.3638,
+      "step": 1006
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.717441967916027e-07,
+      "loss": 0.3182,
+      "step": 1007
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.621861348731838e-07,
+      "loss": 0.3538,
+      "step": 1008
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.527063296226421e-07,
+      "loss": 0.3612,
+      "step": 1009
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.433048596448287e-07,
+      "loss": 0.3494,
+      "step": 1010
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.339818028950561e-07,
+      "loss": 0.3506,
+      "step": 1011
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.247372366784476e-07,
+      "loss": 0.3521,
+      "step": 1012
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.155712376492905e-07,
+      "loss": 0.3154,
+      "step": 1013
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.064838818104201e-07,
+      "loss": 0.3859,
+      "step": 1014
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.97475244512563e-07,
+      "loss": 0.3376,
+      "step": 1015
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.88545400453736e-07,
+      "loss": 0.3795,
+      "step": 1016
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796944236786061e-07,
+      "loss": 0.349,
+      "step": 1017
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.709223875779001e-07,
+      "loss": 0.3469,
+      "step": 1018
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.622293648877696e-07,
+      "loss": 0.3571,
+      "step": 1019
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.536154276892102e-07,
+      "loss": 0.3417,
+      "step": 1020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.450806474074498e-07,
+      "loss": 0.3522,
+      "step": 1021
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.366250948113648e-07,
+      "loss": 0.3513,
+      "step": 1022
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.2824884001288704e-07,
+      "loss": 0.342,
+      "step": 1023
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.199519524664253e-07,
+      "loss": 0.3567,
+      "step": 1024
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.117345009682916e-07,
+      "loss": 0.3635,
+      "step": 1025
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0359655365612483e-07,
+      "loss": 0.343,
+      "step": 1026
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.955381780083356e-07,
+      "loss": 0.349,
+      "step": 1027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8755944084353125e-07,
+      "loss": 0.3496,
+      "step": 1028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.796604083199784e-07,
+      "loss": 0.3666,
+      "step": 1029
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.718411459350446e-07,
+      "loss": 0.3421,
+      "step": 1030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.641017185246576e-07,
+      "loss": 0.3424,
+      "step": 1031
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5644219026276903e-07,
+      "loss": 0.3338,
+      "step": 1032
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.488626246608184e-07,
+      "loss": 0.3629,
+      "step": 1033
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.413630845672122e-07,
+      "loss": 0.3657,
+      "step": 1034
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.339436321667966e-07,
+      "loss": 0.3416,
+      "step": 1035
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.266043289803489e-07,
+      "loss": 0.3396,
+      "step": 1036
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.193452358640581e-07,
+      "loss": 0.3724,
+      "step": 1037
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.121664130090285e-07,
+      "loss": 0.3622,
+      "step": 1038
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.050679199407813e-07,
+      "loss": 0.3592,
+      "step": 1039
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.980498155187506e-07,
+      "loss": 0.345,
+      "step": 1040
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9111215793580894e-07,
+      "loss": 0.3136,
+      "step": 1041
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.842550047177683e-07,
+      "loss": 0.3521,
+      "step": 1042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.774784127229269e-07,
+      "loss": 0.3676,
+      "step": 1043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.707824381415702e-07,
+      "loss": 0.3608,
+      "step": 1044
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.641671364955256e-07,
+      "loss": 0.3364,
+      "step": 1045
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.576325626376952e-07,
+      "loss": 0.369,
+      "step": 1046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.511787707515967e-07,
+      "loss": 0.3717,
+      "step": 1047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4480581435092357e-07,
+      "loss": 0.3683,
+      "step": 1048
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3851374627908875e-07,
+      "loss": 0.3447,
+      "step": 1049
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3230261870879956e-07,
+      "loss": 0.3517,
+      "step": 1050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2617248314161676e-07,
+      "loss": 0.3498,
+      "step": 1051
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2012339040753172e-07,
+      "loss": 0.3658,
+      "step": 1052
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1415539066453884e-07,
+      "loss": 0.3229,
+      "step": 1053
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0826853339822817e-07,
+      "loss": 0.3894,
+      "step": 1054
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.024628674213691e-07,
+      "loss": 0.3829,
+      "step": 1055
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9673844087350736e-07,
+      "loss": 0.347,
+      "step": 1056
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9109530122056742e-07,
+      "loss": 0.3718,
+      "step": 1057
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8553349525445406e-07,
+      "loss": 0.3748,
+      "step": 1058
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8005306909267161e-07,
+      "loss": 0.3746,
+      "step": 1059
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7465406817793296e-07,
+      "loss": 0.3605,
+      "step": 1060
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.693365372777922e-07,
+      "loss": 0.3269,
+      "step": 1061
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6410052048426382e-07,
+      "loss": 0.3733,
+      "step": 1062
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5894606121346745e-07,
+      "loss": 0.3774,
+      "step": 1063
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5387320220525804e-07,
+      "loss": 0.3761,
+      "step": 1064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4888198552287624e-07,
+      "loss": 0.3688,
+      "step": 1065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4397245255260095e-07,
+      "loss": 0.3573,
+      "step": 1066
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3914464400340166e-07,
+      "loss": 0.3554,
+      "step": 1067
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3439859990660775e-07,
+      "loss": 0.3791,
+      "step": 1068
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2973435961556647e-07,
+      "loss": 0.377,
+      "step": 1069
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2515196180532653e-07,
+      "loss": 0.3565,
+      "step": 1070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2065144447231392e-07,
+      "loss": 0.3867,
+      "step": 1071
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1623284493401444e-07,
+      "loss": 0.399,
+      "step": 1072
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1189619982866829e-07,
+      "loss": 0.3667,
+      "step": 1073
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0764154511495928e-07,
+      "loss": 0.3353,
+      "step": 1074
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0346891607172615e-07,
+      "loss": 0.3707,
+      "step": 1075
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.937834729766394e-08,
+      "loss": 0.369,
+      "step": 1076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.536987271104081e-08,
+      "loss": 0.3728,
+      "step": 1077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.144352554940949e-08,
+      "loss": 0.3689,
+      "step": 1078
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.759933836934187e-08,
+      "loss": 0.3647,
+      "step": 1079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.383734304615144e-08,
+      "loss": 0.385,
+      "step": 1080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.015757077363129e-08,
+      "loss": 0.3844,
+      "step": 1081
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.656005206379657e-08,
+      "loss": 0.364,
+      "step": 1082
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.304481674662911e-08,
+      "loss": 0.3736,
+      "step": 1083
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.961189396983537e-08,
+      "loss": 0.3388,
+      "step": 1084
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.626131219859555e-08,
+      "loss": 0.3755,
+      "step": 1085
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.299309921533714e-08,
+      "loss": 0.3521,
+      "step": 1086
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.980728211949838e-08,
+      "loss": 0.3827,
+      "step": 1087
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.670388732730625e-08,
+      "loss": 0.3676,
+      "step": 1088
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3682940571555544e-08,
+      "loss": 0.4033,
+      "step": 1089
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.0744466901397895e-08,
+      "loss": 0.3836,
+      "step": 1090
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.788849068213197e-08,
+      "loss": 0.3629,
+      "step": 1091
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5115035595004744e-08,
+      "loss": 0.3737,
+      "step": 1092
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2424124637009404e-08,
+      "loss": 0.3313,
+      "step": 1093
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9815780120697755e-08,
+      "loss": 0.3986,
+      "step": 1094
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.729002367399925e-08,
+      "loss": 0.395,
+      "step": 1095
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.484687624003447e-08,
+      "loss": 0.3754,
+      "step": 1096
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.248635807694744e-08,
+      "loss": 0.3706,
+      "step": 1097
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.02084887577303e-08,
+      "loss": 0.3856,
+      "step": 1098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8013287170070014e-08,
+      "loss": 0.3858,
+      "step": 1099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5900771516188527e-08,
+      "loss": 0.3794,
+      "step": 1100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3870959312685127e-08,
+      "loss": 0.3939,
+      "step": 1101
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.19238673904032e-08,
+      "loss": 0.376,
+      "step": 1102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.005951189427924e-08,
+      "loss": 0.3799,
+      "step": 1103
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8277908283217404e-08,
+      "loss": 0.3624,
+      "step": 1104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.65790713299574e-08,
+      "loss": 0.3885,
+      "step": 1105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.496301512095011e-08,
+      "loss": 0.3642,
+      "step": 1106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3429753056246608e-08,
+      "loss": 0.4049,
+      "step": 1107
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1979297849381566e-08,
+      "loss": 0.3617,
+      "step": 1108
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0611661527271111e-08,
+      "loss": 0.3731,
+      "step": 1109
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.326855430110693e-09,
+      "loss": 0.4021,
+      "step": 1110
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.124890211284042e-09,
+      "loss": 0.3611,
+      "step": 1111
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.005775837272133e-09,
+      "loss": 0.3858,
+      "step": 1112
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.969521587571025e-09,
+      "loss": 0.3855,
+      "step": 1113
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.016136054616372e-09,
+      "loss": 0.4073,
+      "step": 1114
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.14562714371014e-09,
+      "loss": 0.4058,
+      "step": 1115
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3580020729573338e-09,
+      "loss": 0.3807,
+      "step": 1116
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.6532673732049265e-09,
+      "loss": 0.3794,
+      "step": 1117
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0314288879874635e-09,
+      "loss": 0.3838,
+      "step": 1118
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4924917734771006e-09,
+      "loss": 0.3634,
+      "step": 1119
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.036460498445857e-09,
+      "loss": 0.3849,
+      "step": 1120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.633388442223166e-10,
+      "loss": 0.3814,
+      "step": 1121
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7312990466609187e-10,
+      "loss": 0.3643,
+      "step": 1122
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6583608613562895e-10,
+      "loss": 0.3668,
+      "step": 1123
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1459107477104245e-11,
+      "loss": 0.3527,
+      "step": 1124
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.3751,
+      "step": 1125
+    },
+    {
+      "epoch": 3.0,
+      "step": 1125,
+      "total_flos": 2.878901009252352e+18,
+      "train_loss": 0.3957479126718309,
+      "train_runtime": 37304.899,
+      "train_samples_per_second": 3.866,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1125,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 2.878901009252352e+18,
+  "trial_name": null,
+  "trial_params": null
+}