diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5062 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9991089991089991,
+  "global_step": 841,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.988109393579072e-05,
+      "loss": 2.7158,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.976218787158146e-05,
+      "loss": 2.4151,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.964328180737218e-05,
+      "loss": 2.2672,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.95243757431629e-05,
+      "loss": 2.1708,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.940546967895364e-05,
+      "loss": 2.1738,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.928656361474435e-05,
+      "loss": 2.0842,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.916765755053509e-05,
+      "loss": 2.0694,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.904875148632581e-05,
+      "loss": 2.0431,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.892984542211652e-05,
+      "loss": 2.0017,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.881093935790727e-05,
+      "loss": 1.9447,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.869203329369798e-05,
+      "loss": 2.0282,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.857312722948871e-05,
+      "loss": 1.996,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.845422116527944e-05,
+      "loss": 1.9531,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.833531510107015e-05,
+      "loss": 1.9809,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.821640903686088e-05,
+      "loss": 1.8992,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.809750297265161e-05,
+      "loss": 2.0031,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.797859690844233e-05,
+      "loss": 2.0103,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.785969084423307e-05,
+      "loss": 2.0379,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.774078478002379e-05,
+      "loss": 1.9305,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.762187871581451e-05,
+      "loss": 1.9579,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.750297265160524e-05,
+      "loss": 1.9536,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.738406658739596e-05,
+      "loss": 1.8959,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.726516052318669e-05,
+      "loss": 1.9384,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.714625445897742e-05,
+      "loss": 1.9747,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.702734839476813e-05,
+      "loss": 2.0082,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.690844233055887e-05,
+      "loss": 1.9158,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.678953626634959e-05,
+      "loss": 1.9772,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.667063020214032e-05,
+      "loss": 1.8501,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.655172413793105e-05,
+      "loss": 1.8921,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.643281807372176e-05,
+      "loss": 1.8784,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.631391200951249e-05,
+      "loss": 1.9436,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.619500594530322e-05,
+      "loss": 1.9068,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.607609988109393e-05,
+      "loss": 1.8535,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.595719381688466e-05,
+      "loss": 1.9335,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.583828775267539e-05,
+      "loss": 1.9408,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.571938168846611e-05,
+      "loss": 1.8714,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.560047562425685e-05,
+      "loss": 1.8956,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.548156956004757e-05,
+      "loss": 1.917,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.53626634958383e-05,
+      "loss": 1.9593,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.524375743162902e-05,
+      "loss": 1.9578,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.512485136741974e-05,
+      "loss": 1.8848,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.500594530321047e-05,
+      "loss": 1.8146,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.48870392390012e-05,
+      "loss": 1.931,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.476813317479191e-05,
+      "loss": 1.9622,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.464922711058264e-05,
+      "loss": 1.8196,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.453032104637337e-05,
+      "loss": 1.9061,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.44114149821641e-05,
+      "loss": 1.8412,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.429250891795483e-05,
+      "loss": 1.8545,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.417360285374554e-05,
+      "loss": 1.8734,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.405469678953627e-05,
+      "loss": 1.8747,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3935790725327e-05,
+      "loss": 1.8626,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.381688466111772e-05,
+      "loss": 1.8595,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.369797859690844e-05,
+      "loss": 1.9346,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.357907253269917e-05,
+      "loss": 1.7749,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.34601664684899e-05,
+      "loss": 1.8523,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.334126040428063e-05,
+      "loss": 1.896,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.322235434007135e-05,
+      "loss": 1.8841,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.310344827586207e-05,
+      "loss": 1.8792,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.29845422116528e-05,
+      "loss": 1.8072,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.286563614744352e-05,
+      "loss": 1.8297,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.274673008323425e-05,
+      "loss": 1.8749,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.262782401902498e-05,
+      "loss": 1.8517,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.25089179548157e-05,
+      "loss": 1.9011,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.239001189060642e-05,
+      "loss": 1.9483,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.227110582639715e-05,
+      "loss": 1.9458,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.215219976218788e-05,
+      "loss": 1.8551,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.203329369797861e-05,
+      "loss": 1.9576,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.191438763376932e-05,
+      "loss": 1.8548,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.179548156956005e-05,
+      "loss": 1.8707,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.167657550535078e-05,
+      "loss": 1.8615,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.155766944114151e-05,
+      "loss": 1.8683,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.143876337693222e-05,
+      "loss": 1.8656,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.131985731272295e-05,
+      "loss": 1.8294,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.120095124851368e-05,
+      "loss": 1.8921,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.10820451843044e-05,
+      "loss": 1.7464,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.096313912009513e-05,
+      "loss": 1.8375,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.084423305588586e-05,
+      "loss": 1.8626,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.072532699167658e-05,
+      "loss": 1.8458,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.06064209274673e-05,
+      "loss": 1.8705,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.048751486325803e-05,
+      "loss": 1.8076,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.036860879904876e-05,
+      "loss": 1.814,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.024970273483949e-05,
+      "loss": 1.8143,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.01307966706302e-05,
+      "loss": 1.8467,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.001189060642093e-05,
+      "loss": 1.7998,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.989298454221166e-05,
+      "loss": 1.7384,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.977407847800237e-05,
+      "loss": 1.8921,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.96551724137931e-05,
+      "loss": 1.7481,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.953626634958383e-05,
+      "loss": 1.8534,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.941736028537456e-05,
+      "loss": 1.7691,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.929845422116529e-05,
+      "loss": 1.8355,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.9179548156956e-05,
+      "loss": 1.8045,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.906064209274673e-05,
+      "loss": 1.8272,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.894173602853746e-05,
+      "loss": 1.6901,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.882282996432818e-05,
+      "loss": 1.8515,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.87039239001189e-05,
+      "loss": 1.823,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.858501783590964e-05,
+      "loss": 1.7817,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.846611177170036e-05,
+      "loss": 1.7829,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.83472057074911e-05,
+      "loss": 1.7851,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.822829964328181e-05,
+      "loss": 1.7875,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.810939357907254e-05,
+      "loss": 1.7346,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.799048751486327e-05,
+      "loss": 1.7848,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.787158145065398e-05,
+      "loss": 1.8479,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.775267538644471e-05,
+      "loss": 1.8344,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.763376932223544e-05,
+      "loss": 1.9029,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.751486325802615e-05,
+      "loss": 1.8287,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.73959571938169e-05,
+      "loss": 1.952,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.727705112960761e-05,
+      "loss": 1.8329,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.715814506539834e-05,
+      "loss": 1.7867,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.703923900118907e-05,
+      "loss": 1.6821,
+      "step": 109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.692033293697979e-05,
+      "loss": 1.8331,
+      "step": 110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.680142687277051e-05,
+      "loss": 1.7827,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.668252080856124e-05,
+      "loss": 1.856,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.656361474435196e-05,
+      "loss": 1.7912,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.64447086801427e-05,
+      "loss": 1.8087,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.632580261593342e-05,
+      "loss": 1.8385,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.620689655172413e-05,
+      "loss": 1.8597,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.608799048751487e-05,
+      "loss": 1.7864,
+      "step": 117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.596908442330559e-05,
+      "loss": 1.8338,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.585017835909632e-05,
+      "loss": 1.7652,
+      "step": 119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.573127229488705e-05,
+      "loss": 1.8887,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.561236623067776e-05,
+      "loss": 1.76,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.549346016646849e-05,
+      "loss": 1.7562,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.537455410225922e-05,
+      "loss": 1.7934,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.525564803804994e-05,
+      "loss": 1.7708,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.513674197384068e-05,
+      "loss": 1.833,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.501783590963139e-05,
+      "loss": 1.8027,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.489892984542212e-05,
+      "loss": 1.8631,
+      "step": 127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.478002378121285e-05,
+      "loss": 1.7409,
+      "step": 128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.466111771700357e-05,
+      "loss": 1.7946,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.45422116527943e-05,
+      "loss": 1.8367,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.442330558858502e-05,
+      "loss": 1.7824,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.430439952437574e-05,
+      "loss": 1.7772,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.418549346016648e-05,
+      "loss": 1.8802,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.40665873959572e-05,
+      "loss": 1.8447,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.394768133174791e-05,
+      "loss": 1.7303,
+      "step": 135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.382877526753865e-05,
+      "loss": 1.7148,
+      "step": 136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.370986920332937e-05,
+      "loss": 1.7718,
+      "step": 137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.35909631391201e-05,
+      "loss": 1.7627,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.347205707491083e-05,
+      "loss": 1.7951,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.335315101070154e-05,
+      "loss": 1.7821,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.323424494649228e-05,
+      "loss": 1.7287,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.3115338882283e-05,
+      "loss": 1.724,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.299643281807372e-05,
+      "loss": 1.7938,
+      "step": 143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.287752675386446e-05,
+      "loss": 1.8391,
+      "step": 144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.275862068965517e-05,
+      "loss": 1.8661,
+      "step": 145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.26397146254459e-05,
+      "loss": 1.7528,
+      "step": 146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.252080856123663e-05,
+      "loss": 1.7599,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.240190249702735e-05,
+      "loss": 1.7611,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.228299643281809e-05,
+      "loss": 1.8356,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.21640903686088e-05,
+      "loss": 1.8051,
+      "step": 150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.204518430439952e-05,
+      "loss": 1.7809,
+      "step": 151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.192627824019026e-05,
+      "loss": 1.8646,
+      "step": 152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.180737217598098e-05,
+      "loss": 1.8565,
+      "step": 153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.16884661117717e-05,
+      "loss": 1.8395,
+      "step": 154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.156956004756243e-05,
+      "loss": 1.8218,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.145065398335315e-05,
+      "loss": 1.7845,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.133174791914388e-05,
+      "loss": 1.7718,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.121284185493461e-05,
+      "loss": 1.8222,
+      "step": 158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.109393579072532e-05,
+      "loss": 1.8419,
+      "step": 159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.097502972651607e-05,
+      "loss": 1.7771,
+      "step": 160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.085612366230678e-05,
+      "loss": 1.7965,
+      "step": 161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.073721759809751e-05,
+      "loss": 1.752,
+      "step": 162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.061831153388824e-05,
+      "loss": 1.8606,
+      "step": 163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.049940546967895e-05,
+      "loss": 1.8068,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.038049940546968e-05,
+      "loss": 1.7856,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.026159334126041e-05,
+      "loss": 1.8188,
+      "step": 166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.014268727705113e-05,
+      "loss": 1.774,
+      "step": 167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.002378121284187e-05,
+      "loss": 1.8021,
+      "step": 168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.990487514863258e-05,
+      "loss": 1.7779,
+      "step": 169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.978596908442331e-05,
+      "loss": 1.8481,
+      "step": 170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.966706302021404e-05,
+      "loss": 1.869,
+      "step": 171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.954815695600476e-05,
+      "loss": 1.7668,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.942925089179549e-05,
+      "loss": 1.7863,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.931034482758621e-05,
+      "loss": 1.7641,
+      "step": 174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.919143876337693e-05,
+      "loss": 1.7639,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.907253269916766e-05,
+      "loss": 1.7387,
+      "step": 176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.895362663495839e-05,
+      "loss": 1.8265,
+      "step": 177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.88347205707491e-05,
+      "loss": 1.78,
+      "step": 178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.871581450653985e-05,
+      "loss": 1.7997,
+      "step": 179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.859690844233056e-05,
+      "loss": 1.7667,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.847800237812129e-05,
+      "loss": 1.7949,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.835909631391202e-05,
+      "loss": 1.8046,
+      "step": 182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.824019024970273e-05,
+      "loss": 1.7496,
+      "step": 183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.812128418549346e-05,
+      "loss": 1.8215,
+      "step": 184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.800237812128419e-05,
+      "loss": 1.7899,
+      "step": 185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.788347205707491e-05,
+      "loss": 1.8323,
+      "step": 186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.776456599286564e-05,
+      "loss": 1.7588,
+      "step": 187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.764565992865636e-05,
+      "loss": 1.8187,
+      "step": 188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.75267538644471e-05,
+      "loss": 1.775,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.740784780023782e-05,
+      "loss": 1.7652,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.728894173602854e-05,
+      "loss": 1.8968,
+      "step": 191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.717003567181927e-05,
+      "loss": 1.7757,
+      "step": 192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.705112960761e-05,
+      "loss": 1.7359,
+      "step": 193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.693222354340071e-05,
+      "loss": 1.7144,
+      "step": 194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.681331747919144e-05,
+      "loss": 1.8064,
+      "step": 195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.669441141498217e-05,
+      "loss": 1.7938,
+      "step": 196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.65755053507729e-05,
+      "loss": 1.7412,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.645659928656363e-05,
+      "loss": 1.7161,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.633769322235434e-05,
+      "loss": 1.8272,
+      "step": 199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.621878715814507e-05,
+      "loss": 1.8578,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.60998810939358e-05,
+      "loss": 1.8092,
+      "step": 201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.598097502972651e-05,
+      "loss": 1.7152,
+      "step": 202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.586206896551724e-05,
+      "loss": 1.8046,
+      "step": 203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.574316290130797e-05,
+      "loss": 1.7136,
+      "step": 204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.56242568370987e-05,
+      "loss": 1.8415,
+      "step": 205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.550535077288942e-05,
+      "loss": 1.7565,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.538644470868014e-05,
+      "loss": 1.8995,
+      "step": 207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.526753864447087e-05,
+      "loss": 1.7956,
+      "step": 208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.51486325802616e-05,
+      "loss": 1.8588,
+      "step": 209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.502972651605232e-05,
+      "loss": 1.8185,
+      "step": 210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.491082045184305e-05,
+      "loss": 1.873,
+      "step": 211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.479191438763378e-05,
+      "loss": 1.7106,
+      "step": 212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.46730083234245e-05,
+      "loss": 1.7969,
+      "step": 213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.455410225921522e-05,
+      "loss": 1.8109,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.443519619500595e-05,
+      "loss": 1.6997,
+      "step": 215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.431629013079668e-05,
+      "loss": 1.6892,
+      "step": 216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.419738406658739e-05,
+      "loss": 1.7694,
+      "step": 217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.407847800237812e-05,
+      "loss": 1.8407,
+      "step": 218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.395957193816885e-05,
+      "loss": 1.7592,
+      "step": 219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.384066587395958e-05,
+      "loss": 1.8825,
+      "step": 220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.372175980975031e-05,
+      "loss": 1.7768,
+      "step": 221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.360285374554102e-05,
+      "loss": 1.7464,
+      "step": 222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.348394768133175e-05,
+      "loss": 1.787,
+      "step": 223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.336504161712248e-05,
+      "loss": 1.7042,
+      "step": 224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.32461355529132e-05,
+      "loss": 1.7716,
+      "step": 225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.312722948870393e-05,
+      "loss": 1.8138,
+      "step": 226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.300832342449465e-05,
+      "loss": 1.851,
+      "step": 227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.288941736028538e-05,
+      "loss": 1.7168,
+      "step": 228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.27705112960761e-05,
+      "loss": 1.7311,
+      "step": 229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.265160523186683e-05,
+      "loss": 1.7531,
+      "step": 230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.253269916765756e-05,
+      "loss": 1.7974,
+      "step": 231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.241379310344828e-05,
+      "loss": 1.7964,
+      "step": 232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.2294887039239e-05,
+      "loss": 1.8198,
+      "step": 233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.217598097502973e-05,
+      "loss": 1.7226,
+      "step": 234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.205707491082046e-05,
+      "loss": 1.7325,
+      "step": 235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.193816884661117e-05,
+      "loss": 1.6968,
+      "step": 236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.18192627824019e-05,
+      "loss": 1.8155,
+      "step": 237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.170035671819263e-05,
+      "loss": 1.7248,
+      "step": 238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.158145065398336e-05,
+      "loss": 1.7376,
+      "step": 239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.146254458977409e-05,
+      "loss": 1.7557,
+      "step": 240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.13436385255648e-05,
+      "loss": 1.7545,
+      "step": 241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.122473246135553e-05,
+      "loss": 1.8398,
+      "step": 242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.110582639714626e-05,
+      "loss": 1.9711,
+      "step": 243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.098692033293698e-05,
+      "loss": 1.8022,
+      "step": 244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.08680142687277e-05,
+      "loss": 1.7482,
+      "step": 245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.074910820451843e-05,
+      "loss": 1.6878,
+      "step": 246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.063020214030915e-05,
+      "loss": 1.8446,
+      "step": 247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.051129607609989e-05,
+      "loss": 1.7308,
+      "step": 248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.039239001189061e-05,
+      "loss": 1.7943,
+      "step": 249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.027348394768134e-05,
+      "loss": 1.7377,
+      "step": 250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.015457788347207e-05,
+      "loss": 1.707,
+      "step": 251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.003567181926278e-05,
+      "loss": 1.7998,
+      "step": 252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.991676575505351e-05,
+      "loss": 1.7726,
+      "step": 253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.979785969084424e-05,
+      "loss": 1.721,
+      "step": 254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.967895362663495e-05,
+      "loss": 1.6772,
+      "step": 255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.95600475624257e-05,
+      "loss": 1.7242,
+      "step": 256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.944114149821641e-05,
+      "loss": 1.7214,
+      "step": 257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.932223543400713e-05,
+      "loss": 1.787,
+      "step": 258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.920332936979787e-05,
+      "loss": 1.7535,
+      "step": 259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.908442330558858e-05,
+      "loss": 1.7076,
+      "step": 260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.896551724137931e-05,
+      "loss": 1.7545,
+      "step": 261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.884661117717004e-05,
+      "loss": 1.6848,
+      "step": 262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.872770511296076e-05,
+      "loss": 1.7793,
+      "step": 263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.86087990487515e-05,
+      "loss": 1.7929,
+      "step": 264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.848989298454222e-05,
+      "loss": 1.7828,
+      "step": 265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.837098692033293e-05,
+      "loss": 1.839,
+      "step": 266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.825208085612367e-05,
+      "loss": 1.812,
+      "step": 267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.813317479191439e-05,
+      "loss": 1.654,
+      "step": 268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.801426872770512e-05,
+      "loss": 1.7531,
+      "step": 269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.789536266349585e-05,
+      "loss": 1.7529,
+      "step": 270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.777645659928656e-05,
+      "loss": 1.801,
+      "step": 271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.765755053507729e-05,
+      "loss": 1.8097,
+      "step": 272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.753864447086802e-05,
+      "loss": 1.8078,
+      "step": 273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.741973840665873e-05,
+      "loss": 1.7135,
+      "step": 274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.730083234244948e-05,
+      "loss": 1.7674,
+      "step": 275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.718192627824019e-05,
+      "loss": 1.6498,
+      "step": 276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.706302021403092e-05,
+      "loss": 1.7322,
+      "step": 277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.694411414982165e-05,
+      "loss": 1.7194,
+      "step": 278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.682520808561236e-05,
+      "loss": 1.8677,
+      "step": 279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.67063020214031e-05,
+      "loss": 1.8229,
+      "step": 280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.658739595719382e-05,
+      "loss": 1.7648,
+      "step": 281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.646848989298454e-05,
+      "loss": 1.7758,
+      "step": 282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.634958382877528e-05,
+      "loss": 1.7759,
+      "step": 283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.6230677764566e-05,
+      "loss": 1.7449,
+      "step": 284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.611177170035671e-05,
+      "loss": 1.7185,
+      "step": 285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.599286563614745e-05,
+      "loss": 1.7535,
+      "step": 286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.587395957193817e-05,
+      "loss": 1.7255,
+      "step": 287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.57550535077289e-05,
+      "loss": 1.7131,
+      "step": 288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.563614744351963e-05,
+      "loss": 1.6799,
+      "step": 289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.551724137931034e-05,
+      "loss": 1.7557,
+      "step": 290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.539833531510108e-05,
+      "loss": 1.8136,
+      "step": 291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.52794292508918e-05,
+      "loss": 1.7864,
+      "step": 292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.516052318668251e-05,
+      "loss": 1.7938,
+      "step": 293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.504161712247326e-05,
+      "loss": 1.8176,
+      "step": 294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.492271105826397e-05,
+      "loss": 1.7083,
+      "step": 295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.48038049940547e-05,
+      "loss": 1.8511,
+      "step": 296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.468489892984543e-05,
+      "loss": 1.8187,
+      "step": 297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.456599286563615e-05,
+      "loss": 1.8411,
+      "step": 298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.444708680142689e-05,
+      "loss": 1.7231,
+      "step": 299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.43281807372176e-05,
+      "loss": 1.7731,
+      "step": 300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.420927467300832e-05,
+      "loss": 1.8391,
+      "step": 301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.409036860879906e-05,
+      "loss": 1.6737,
+      "step": 302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.397146254458978e-05,
+      "loss": 1.7448,
+      "step": 303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.38525564803805e-05,
+      "loss": 1.7813,
+      "step": 304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.373365041617123e-05,
+      "loss": 1.7668,
+      "step": 305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.361474435196195e-05,
+      "loss": 1.8137,
+      "step": 306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.349583828775268e-05,
+      "loss": 1.6559,
+      "step": 307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.33769322235434e-05,
+      "loss": 1.7336,
+      "step": 308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.325802615933412e-05,
+      "loss": 1.7671,
+      "step": 309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.313912009512486e-05,
+      "loss": 1.7168,
+      "step": 310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.302021403091558e-05,
+      "loss": 1.7432,
+      "step": 311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.290130796670631e-05,
+      "loss": 1.7132,
+      "step": 312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.278240190249704e-05,
+      "loss": 1.7477,
+      "step": 313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.266349583828775e-05,
+      "loss": 1.6738,
+      "step": 314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.254458977407848e-05,
+      "loss": 1.6969,
+      "step": 315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.242568370986921e-05,
+      "loss": 1.7723,
+      "step": 316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.230677764565993e-05,
+      "loss": 1.7256,
+      "step": 317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.218787158145065e-05,
+      "loss": 1.703,
+      "step": 318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.206896551724138e-05,
+      "loss": 1.8493,
+      "step": 319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.195005945303211e-05,
+      "loss": 1.7068,
+      "step": 320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.183115338882284e-05,
+      "loss": 1.7455,
+      "step": 321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.171224732461356e-05,
+      "loss": 1.7961,
+      "step": 322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.159334126040429e-05,
+      "loss": 1.7037,
+      "step": 323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.147443519619501e-05,
+      "loss": 1.7348,
+      "step": 324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.135552913198573e-05,
+      "loss": 1.7024,
+      "step": 325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.123662306777646e-05,
+      "loss": 1.6221,
+      "step": 326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.111771700356719e-05,
+      "loss": 1.7412,
+      "step": 327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.099881093935791e-05,
+      "loss": 1.7166,
+      "step": 328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.087990487514863e-05,
+      "loss": 1.7569,
+      "step": 329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.076099881093936e-05,
+      "loss": 1.6272,
+      "step": 330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.064209274673008e-05,
+      "loss": 1.6421,
+      "step": 331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.052318668252082e-05,
+      "loss": 1.7958,
+      "step": 332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.040428061831154e-05,
+      "loss": 1.6606,
+      "step": 333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.028537455410226e-05,
+      "loss": 1.6852,
+      "step": 334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.016646848989299e-05,
+      "loss": 1.7697,
+      "step": 335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.004756242568371e-05,
+      "loss": 1.7747,
+      "step": 336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9928656361474435e-05,
+      "loss": 1.8011,
+      "step": 337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9809750297265164e-05,
+      "loss": 1.7967,
+      "step": 338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9690844233055886e-05,
+      "loss": 1.7095,
+      "step": 339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.957193816884662e-05,
+      "loss": 1.6256,
+      "step": 340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9453032104637343e-05,
+      "loss": 1.7116,
+      "step": 341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.933412604042806e-05,
+      "loss": 1.6576,
+      "step": 342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9215219976218794e-05,
+      "loss": 1.7988,
+      "step": 343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9096313912009516e-05,
+      "loss": 1.7584,
+      "step": 344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.897740784780024e-05,
+      "loss": 1.7492,
+      "step": 345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.885850178359097e-05,
+      "loss": 1.7966,
+      "step": 346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.873959571938169e-05,
+      "loss": 1.7573,
+      "step": 347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.862068965517241e-05,
+      "loss": 1.7959,
+      "step": 348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.850178359096315e-05,
+      "loss": 1.728,
+      "step": 349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838287752675386e-05,
+      "loss": 1.7817,
+      "step": 350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.82639714625446e-05,
+      "loss": 1.7165,
+      "step": 351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.814506539833532e-05,
+      "loss": 1.6588,
+      "step": 352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.802615933412604e-05,
+      "loss": 1.7179,
+      "step": 353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.790725326991677e-05,
+      "loss": 1.7937,
+      "step": 354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.778834720570749e-05,
+      "loss": 1.6768,
+      "step": 355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7669441141498215e-05,
+      "loss": 1.8075,
+      "step": 356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7550535077288944e-05,
+      "loss": 1.758,
+      "step": 357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7431629013079666e-05,
+      "loss": 1.7014,
+      "step": 358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.731272294887039e-05,
+      "loss": 1.7249,
+      "step": 359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7193816884661124e-05,
+      "loss": 1.8606,
+      "step": 360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7074910820451846e-05,
+      "loss": 1.7062,
+      "step": 361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6956004756242575e-05,
+      "loss": 1.7216,
+      "step": 362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.68370986920333e-05,
+      "loss": 1.7118,
+      "step": 363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.671819262782402e-05,
+      "loss": 1.7849,
+      "step": 364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.659928656361475e-05,
+      "loss": 1.6746,
+      "step": 365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.648038049940547e-05,
+      "loss": 1.7484,
+      "step": 366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.636147443519619e-05,
+      "loss": 1.7031,
+      "step": 367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.624256837098693e-05,
+      "loss": 1.7489,
+      "step": 368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.612366230677765e-05,
+      "loss": 1.7066,
+      "step": 369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.600475624256838e-05,
+      "loss": 1.7141,
+      "step": 370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.58858501783591e-05,
+      "loss": 1.7295,
+      "step": 371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.576694411414982e-05,
+      "loss": 1.7428,
+      "step": 372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.564803804994055e-05,
+      "loss": 1.8843,
+      "step": 373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.5529131985731273e-05,
+      "loss": 1.6657,
+      "step": 374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5410225921521996e-05,
+      "loss": 1.7918,
+      "step": 375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.529131985731273e-05,
+      "loss": 1.8119,
+      "step": 376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.517241379310345e-05,
+      "loss": 1.7182,
+      "step": 377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.505350772889417e-05,
+      "loss": 1.6241,
+      "step": 378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4934601664684904e-05,
+      "loss": 1.8406,
+      "step": 379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4815695600475626e-05,
+      "loss": 1.7557,
+      "step": 380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4696789536266355e-05,
+      "loss": 1.7313,
+      "step": 381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.457788347205708e-05,
+      "loss": 1.8014,
+      "step": 382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.44589774078478e-05,
+      "loss": 1.7754,
+      "step": 383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4340071343638535e-05,
+      "loss": 1.8043,
+      "step": 384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.422116527942925e-05,
+      "loss": 1.7172,
+      "step": 385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.410225921521997e-05,
+      "loss": 1.8322,
+      "step": 386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.398335315101071e-05,
+      "loss": 1.6688,
+      "step": 387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.386444708680143e-05,
+      "loss": 1.7057,
+      "step": 388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.374554102259215e-05,
+      "loss": 1.7585,
+      "step": 389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.362663495838288e-05,
+      "loss": 1.6733,
+      "step": 390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.35077288941736e-05,
+      "loss": 1.7244,
+      "step": 391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.338882282996434e-05,
+      "loss": 1.6851,
+      "step": 392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3269916765755054e-05,
+      "loss": 1.7308,
+      "step": 393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.3151010701545776e-05,
+      "loss": 1.7408,
+      "step": 394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.303210463733651e-05,
+      "loss": 1.7066,
+      "step": 395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2913198573127234e-05,
+      "loss": 1.7651,
+      "step": 396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2794292508917956e-05,
+      "loss": 1.7379,
+      "step": 397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2675386444708685e-05,
+      "loss": 1.7564,
+      "step": 398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.255648038049941e-05,
+      "loss": 1.639,
+      "step": 399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.243757431629013e-05,
+      "loss": 1.718,
+      "step": 400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.231866825208086e-05,
+      "loss": 1.6737,
+      "step": 401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.219976218787158e-05,
+      "loss": 1.7218,
+      "step": 402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2080856123662315e-05,
+      "loss": 1.8091,
+      "step": 403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.196195005945304e-05,
+      "loss": 1.6999,
+      "step": 404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.184304399524376e-05,
+      "loss": 1.5945,
+      "step": 405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.172413793103449e-05,
+      "loss": 1.7723,
+      "step": 406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.160523186682521e-05,
+      "loss": 1.6238,
+      "step": 407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.148632580261593e-05,
+      "loss": 1.7872,
+      "step": 408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.136741973840666e-05,
+      "loss": 1.6822,
+      "step": 409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.124851367419738e-05,
+      "loss": 1.6825,
+      "step": 410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.112960760998812e-05,
+      "loss": 1.6745,
+      "step": 411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.101070154577884e-05,
+      "loss": 1.7777,
+      "step": 412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0891795481569556e-05,
+      "loss": 1.6885,
+      "step": 413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.077288941736029e-05,
+      "loss": 1.6952,
+      "step": 414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0653983353151014e-05,
+      "loss": 1.7158,
+      "step": 415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.0535077288941736e-05,
+      "loss": 1.6974,
+      "step": 416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0416171224732465e-05,
+      "loss": 1.7148,
+      "step": 417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.029726516052319e-05,
+      "loss": 1.7103,
+      "step": 418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.017835909631391e-05,
+      "loss": 1.7076,
+      "step": 419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0059453032104645e-05,
+      "loss": 1.744,
+      "step": 420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.994054696789536e-05,
+      "loss": 1.7262,
+      "step": 421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.982164090368609e-05,
+      "loss": 1.7663,
+      "step": 422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.970273483947682e-05,
+      "loss": 1.7074,
+      "step": 423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9583828775267547e-05,
+      "loss": 1.727,
+      "step": 424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.946492271105826e-05,
+      "loss": 1.7705,
+      "step": 425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934601664684899e-05,
+      "loss": 1.7091,
+      "step": 426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.922711058263972e-05,
+      "loss": 1.7337,
+      "step": 427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.910820451843044e-05,
+      "loss": 1.8065,
+      "step": 428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8989298454221164e-05,
+      "loss": 1.8079,
+      "step": 429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.887039239001189e-05,
+      "loss": 1.8365,
+      "step": 430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.875148632580262e-05,
+      "loss": 1.6413,
+      "step": 431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8632580261593344e-05,
+      "loss": 1.8578,
+      "step": 432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8513674197384066e-05,
+      "loss": 1.6945,
+      "step": 433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8394768133174794e-05,
+      "loss": 1.6684,
+      "step": 434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.827586206896552e-05,
+      "loss": 1.6958,
+      "step": 435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8156956004756245e-05,
+      "loss": 1.8266,
+      "step": 436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.803804994054697e-05,
+      "loss": 1.6614,
+      "step": 437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7919143876337696e-05,
+      "loss": 1.709,
+      "step": 438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7800237812128425e-05,
+      "loss": 1.7156,
+      "step": 439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.768133174791915e-05,
+      "loss": 1.6307,
+      "step": 440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.756242568370987e-05,
+      "loss": 1.7491,
+      "step": 441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74435196195006e-05,
+      "loss": 1.8355,
+      "step": 442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.732461355529132e-05,
+      "loss": 1.8129,
+      "step": 443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720570749108205e-05,
+      "loss": 1.7334,
+      "step": 444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.708680142687277e-05,
+      "loss": 1.7265,
+      "step": 445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.69678953626635e-05,
+      "loss": 1.8617,
+      "step": 446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684898929845422e-05,
+      "loss": 1.6402,
+      "step": 447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.673008323424495e-05,
+      "loss": 1.8096,
+      "step": 448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.661117717003567e-05,
+      "loss": 1.7994,
+      "step": 449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.64922711058264e-05,
+      "loss": 1.7368,
+      "step": 450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6373365041617124e-05,
+      "loss": 1.7132,
+      "step": 451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.625445897740785e-05,
+      "loss": 1.7676,
+      "step": 452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6135552913198575e-05,
+      "loss": 1.6674,
+      "step": 453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6016646848989304e-05,
+      "loss": 1.6637,
+      "step": 454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5897740784780026e-05,
+      "loss": 1.8421,
+      "step": 455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5778834720570755e-05,
+      "loss": 1.7415,
+      "step": 456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565992865636148e-05,
+      "loss": 1.7677,
+      "step": 457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55410225921522e-05,
+      "loss": 1.7437,
+      "step": 458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.542211652794293e-05,
+      "loss": 1.7128,
+      "step": 459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.530321046373365e-05,
+      "loss": 1.62,
+      "step": 460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.518430439952438e-05,
+      "loss": 1.747,
+      "step": 461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.50653983353151e-05,
+      "loss": 1.6181,
+      "step": 462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.494649227110583e-05,
+      "loss": 1.7439,
+      "step": 463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482758620689655e-05,
+      "loss": 1.7278,
+      "step": 464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470868014268728e-05,
+      "loss": 1.7219,
+      "step": 465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4589774078478e-05,
+      "loss": 1.7518,
+      "step": 466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.447086801426873e-05,
+      "loss": 1.7196,
+      "step": 467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.435196195005945e-05,
+      "loss": 1.7109,
+      "step": 468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.423305588585018e-05,
+      "loss": 1.8058,
+      "step": 469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4114149821640904e-05,
+      "loss": 1.761,
+      "step": 470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.399524375743163e-05,
+      "loss": 1.6968,
+      "step": 471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3876337693222355e-05,
+      "loss": 1.7334,
+      "step": 472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.375743162901308e-05,
+      "loss": 1.7534,
+      "step": 473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3638525564803806e-05,
+      "loss": 1.6206,
+      "step": 474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3519619500594535e-05,
+      "loss": 1.8294,
+      "step": 475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.340071343638526e-05,
+      "loss": 1.7374,
+      "step": 476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.328180737217598e-05,
+      "loss": 1.607,
+      "step": 477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.316290130796671e-05,
+      "loss": 1.7038,
+      "step": 478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.304399524375744e-05,
+      "loss": 1.7262,
+      "step": 479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.292508917954816e-05,
+      "loss": 1.7198,
+      "step": 480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.280618311533888e-05,
+      "loss": 1.7586,
+      "step": 481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.268727705112961e-05,
+      "loss": 1.7254,
+      "step": 482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.256837098692034e-05,
+      "loss": 1.7653,
+      "step": 483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.244946492271106e-05,
+      "loss": 1.6971,
+      "step": 484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.233055885850178e-05,
+      "loss": 1.8036,
+      "step": 485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.221165279429251e-05,
+      "loss": 1.6539,
+      "step": 486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.209274673008324e-05,
+      "loss": 1.7074,
+      "step": 487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1973840665873956e-05,
+      "loss": 1.7434,
+      "step": 488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1854934601664685e-05,
+      "loss": 1.7265,
+      "step": 489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1736028537455414e-05,
+      "loss": 1.8158,
+      "step": 490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.161712247324614e-05,
+      "loss": 1.6312,
+      "step": 491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.149821640903686e-05,
+      "loss": 1.9174,
+      "step": 492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1379310344827587e-05,
+      "loss": 1.6678,
+      "step": 493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1260404280618315e-05,
+      "loss": 1.732,
+      "step": 494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1141498216409044e-05,
+      "loss": 1.72,
+      "step": 495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.102259215219976e-05,
+      "loss": 1.7456,
+      "step": 496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.090368608799049e-05,
+      "loss": 1.678,
+      "step": 497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.078478002378122e-05,
+      "loss": 1.6737,
+      "step": 498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.066587395957194e-05,
+      "loss": 1.7998,
+      "step": 499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.054696789536266e-05,
+      "loss": 1.8074,
+      "step": 500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.042806183115339e-05,
+      "loss": 1.7956,
+      "step": 501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.030915576694412e-05,
+      "loss": 1.7845,
+      "step": 502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.019024970273484e-05,
+      "loss": 1.6547,
+      "step": 503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007134363852556e-05,
+      "loss": 1.7467,
+      "step": 504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995243757431629e-05,
+      "loss": 1.7029,
+      "step": 505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.983353151010702e-05,
+      "loss": 1.6846,
+      "step": 506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.971462544589774e-05,
+      "loss": 1.7089,
+      "step": 507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9595719381688465e-05,
+      "loss": 1.7329,
+      "step": 508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9476813317479194e-05,
+      "loss": 1.7055,
+      "step": 509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.935790725326992e-05,
+      "loss": 1.799,
+      "step": 510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9239001189060645e-05,
+      "loss": 1.7176,
+      "step": 511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.912009512485137e-05,
+      "loss": 1.7907,
+      "step": 512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9001189060642096e-05,
+      "loss": 1.7471,
+      "step": 513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.888228299643282e-05,
+      "loss": 1.8493,
+      "step": 514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.876337693222355e-05,
+      "loss": 1.7391,
+      "step": 515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.864447086801427e-05,
+      "loss": 1.6979,
+      "step": 516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8525564803805e-05,
+      "loss": 1.6477,
+      "step": 517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.840665873959572e-05,
+      "loss": 1.7371,
+      "step": 518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.828775267538645e-05,
+      "loss": 1.6983,
+      "step": 519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.816884661117717e-05,
+      "loss": 1.6594,
+      "step": 520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.80499405469679e-05,
+      "loss": 1.7818,
+      "step": 521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.793103448275862e-05,
+      "loss": 1.7133,
+      "step": 522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.781212841854935e-05,
+      "loss": 1.636,
+      "step": 523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.769322235434007e-05,
+      "loss": 1.7091,
+      "step": 524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.75743162901308e-05,
+      "loss": 1.8906,
+      "step": 525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7455410225921523e-05,
+      "loss": 1.7421,
+      "step": 526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.733650416171225e-05,
+      "loss": 1.6599,
+      "step": 527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7217598097502974e-05,
+      "loss": 1.6795,
+      "step": 528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7098692033293696e-05,
+      "loss": 1.7275,
+      "step": 529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6979785969084425e-05,
+      "loss": 1.7678,
+      "step": 530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6860879904875154e-05,
+      "loss": 1.6144,
+      "step": 531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6741973840665876e-05,
+      "loss": 1.7542,
+      "step": 532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.66230677764566e-05,
+      "loss": 1.8074,
+      "step": 533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.650416171224733e-05,
+      "loss": 1.6957,
+      "step": 534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.638525564803805e-05,
+      "loss": 1.7772,
+      "step": 535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.626634958382878e-05,
+      "loss": 1.7557,
+      "step": 536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.61474435196195e-05,
+      "loss": 1.7522,
+      "step": 537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.602853745541023e-05,
+      "loss": 1.6485,
+      "step": 538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.590963139120095e-05,
+      "loss": 1.7184,
+      "step": 539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.579072532699168e-05,
+      "loss": 1.7057,
+      "step": 540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.56718192627824e-05,
+      "loss": 1.7619,
+      "step": 541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.555291319857313e-05,
+      "loss": 1.6884,
+      "step": 542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.543400713436385e-05,
+      "loss": 1.7536,
+      "step": 543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5315101070154575e-05,
+      "loss": 1.7243,
+      "step": 544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5196195005945304e-05,
+      "loss": 1.7216,
+      "step": 545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.507728894173603e-05,
+      "loss": 1.7615,
+      "step": 546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4958382877526755e-05,
+      "loss": 1.7241,
+      "step": 547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.483947681331748e-05,
+      "loss": 1.7117,
+      "step": 548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4720570749108206e-05,
+      "loss": 1.7089,
+      "step": 549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4601664684898935e-05,
+      "loss": 1.7156,
+      "step": 550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4482758620689657e-05,
+      "loss": 1.7517,
+      "step": 551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.436385255648038e-05,
+      "loss": 1.6341,
+      "step": 552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.424494649227111e-05,
+      "loss": 1.6773,
+      "step": 553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4126040428061836e-05,
+      "loss": 1.7078,
+      "step": 554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.400713436385256e-05,
+      "loss": 1.7233,
+      "step": 555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.388822829964328e-05,
+      "loss": 1.6806,
+      "step": 556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.376932223543401e-05,
+      "loss": 1.6341,
+      "step": 557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.365041617122474e-05,
+      "loss": 1.7141,
+      "step": 558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.353151010701546e-05,
+      "loss": 1.6068,
+      "step": 559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.341260404280618e-05,
+      "loss": 1.7645,
+      "step": 560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.329369797859691e-05,
+      "loss": 1.777,
+      "step": 561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.317479191438764e-05,
+      "loss": 1.705,
+      "step": 562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3055885850178355e-05,
+      "loss": 1.7017,
+      "step": 563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2936979785969084e-05,
+      "loss": 1.7343,
+      "step": 564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.281807372175981e-05,
+      "loss": 1.7232,
+      "step": 565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.269916765755054e-05,
+      "loss": 1.6901,
+      "step": 566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.258026159334126e-05,
+      "loss": 1.6531,
+      "step": 567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.2461355529131986e-05,
+      "loss": 1.6593,
+      "step": 568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2342449464922715e-05,
+      "loss": 1.7687,
+      "step": 569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2223543400713444e-05,
+      "loss": 1.6852,
+      "step": 570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.210463733650416e-05,
+      "loss": 1.7326,
+      "step": 571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.198573127229489e-05,
+      "loss": 1.6662,
+      "step": 572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.186682520808562e-05,
+      "loss": 1.7075,
+      "step": 573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.174791914387634e-05,
+      "loss": 1.7528,
+      "step": 574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.162901307966706e-05,
+      "loss": 1.7275,
+      "step": 575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.151010701545779e-05,
+      "loss": 1.7226,
+      "step": 576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.139120095124852e-05,
+      "loss": 1.7396,
+      "step": 577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.127229488703924e-05,
+      "loss": 1.715,
+      "step": 578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.115338882282996e-05,
+      "loss": 1.5859,
+      "step": 579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.103448275862069e-05,
+      "loss": 1.6918,
+      "step": 580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.091557669441142e-05,
+      "loss": 1.6266,
+      "step": 581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.079667063020214e-05,
+      "loss": 1.7796,
+      "step": 582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0677764565992865e-05,
+      "loss": 1.6694,
+      "step": 583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0558858501783593e-05,
+      "loss": 1.7192,
+      "step": 584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0439952437574316e-05,
+      "loss": 1.6945,
+      "step": 585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.032104637336504e-05,
+      "loss": 1.7392,
+      "step": 586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.020214030915577e-05,
+      "loss": 1.6582,
+      "step": 587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0083234244946495e-05,
+      "loss": 1.6592,
+      "step": 588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9964328180737217e-05,
+      "loss": 1.6325,
+      "step": 589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9845422116527943e-05,
+      "loss": 1.7332,
+      "step": 590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9726516052318672e-05,
+      "loss": 1.7304,
+      "step": 591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9607609988109397e-05,
+      "loss": 1.7202,
+      "step": 592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.948870392390012e-05,
+      "loss": 1.6673,
+      "step": 593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9369797859690845e-05,
+      "loss": 1.7825,
+      "step": 594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9250891795481574e-05,
+      "loss": 1.768,
+      "step": 595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.91319857312723e-05,
+      "loss": 1.7782,
+      "step": 596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.901307966706302e-05,
+      "loss": 1.7824,
+      "step": 597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8894173602853747e-05,
+      "loss": 1.6703,
+      "step": 598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8775267538644472e-05,
+      "loss": 1.7553,
+      "step": 599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8656361474435194e-05,
+      "loss": 1.6247,
+      "step": 600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8537455410225923e-05,
+      "loss": 1.8252,
+      "step": 601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.841854934601665e-05,
+      "loss": 1.6936,
+      "step": 602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8299643281807374e-05,
+      "loss": 1.6764,
+      "step": 603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8180737217598096e-05,
+      "loss": 1.7406,
+      "step": 604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8061831153388825e-05,
+      "loss": 1.8286,
+      "step": 605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.794292508917955e-05,
+      "loss": 1.7745,
+      "step": 606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7824019024970276e-05,
+      "loss": 1.8143,
+      "step": 607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7705112960760998e-05,
+      "loss": 1.7166,
+      "step": 608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7586206896551727e-05,
+      "loss": 1.7409,
+      "step": 609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7467300832342452e-05,
+      "loss": 1.7777,
+      "step": 610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7348394768133178e-05,
+      "loss": 1.6752,
+      "step": 611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.72294887039239e-05,
+      "loss": 1.7319,
+      "step": 612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7110582639714625e-05,
+      "loss": 1.6935,
+      "step": 613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6991676575505354e-05,
+      "loss": 1.7737,
+      "step": 614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6872770511296076e-05,
+      "loss": 1.7669,
+      "step": 615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.67538644470868e-05,
+      "loss": 1.7402,
+      "step": 616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6634958382877527e-05,
+      "loss": 1.7329,
+      "step": 617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6516052318668256e-05,
+      "loss": 1.7957,
+      "step": 618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6397146254458978e-05,
+      "loss": 1.7284,
+      "step": 619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6278240190249703e-05,
+      "loss": 1.693,
+      "step": 620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.615933412604043e-05,
+      "loss": 1.7125,
+      "step": 621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6040428061831158e-05,
+      "loss": 1.7522,
+      "step": 622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.592152199762188e-05,
+      "loss": 1.8251,
+      "step": 623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5802615933412605e-05,
+      "loss": 1.7819,
+      "step": 624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.568370986920333e-05,
+      "loss": 1.85,
+      "step": 625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.556480380499406e-05,
+      "loss": 1.6207,
+      "step": 626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5445897740784778e-05,
+      "loss": 1.7541,
+      "step": 627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5326991676575507e-05,
+      "loss": 1.7714,
+      "step": 628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5208085612366232e-05,
+      "loss": 1.7611,
+      "step": 629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5089179548156955e-05,
+      "loss": 1.7122,
+      "step": 630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.497027348394768e-05,
+      "loss": 1.6719,
+      "step": 631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.485136741973841e-05,
+      "loss": 1.7228,
+      "step": 632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.473246135552913e-05,
+      "loss": 1.744,
+      "step": 633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.461355529131986e-05,
+      "loss": 1.684,
+      "step": 634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4494649227110582e-05,
+      "loss": 1.6769,
+      "step": 635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.437574316290131e-05,
+      "loss": 1.6754,
+      "step": 636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4256837098692033e-05,
+      "loss": 1.7128,
+      "step": 637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.413793103448276e-05,
+      "loss": 1.8231,
+      "step": 638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4019024970273484e-05,
+      "loss": 1.6334,
+      "step": 639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3900118906064213e-05,
+      "loss": 1.8426,
+      "step": 640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3781212841854935e-05,
+      "loss": 1.6816,
+      "step": 641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.366230677764566e-05,
+      "loss": 1.6672,
+      "step": 642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3543400713436386e-05,
+      "loss": 1.777,
+      "step": 643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.342449464922711e-05,
+      "loss": 1.7193,
+      "step": 644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3305588585017836e-05,
+      "loss": 1.7789,
+      "step": 645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3186682520808562e-05,
+      "loss": 1.6801,
+      "step": 646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3067776456599287e-05,
+      "loss": 1.7853,
+      "step": 647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2948870392390013e-05,
+      "loss": 1.7196,
+      "step": 648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.282996432818074e-05,
+      "loss": 1.7626,
+      "step": 649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2711058263971464e-05,
+      "loss": 1.5877,
+      "step": 650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.259215219976219e-05,
+      "loss": 1.6909,
+      "step": 651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2473246135552915e-05,
+      "loss": 1.7194,
+      "step": 652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.235434007134364e-05,
+      "loss": 1.7266,
+      "step": 653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2235434007134366e-05,
+      "loss": 1.8016,
+      "step": 654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.211652794292509e-05,
+      "loss": 1.7172,
+      "step": 655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1997621878715817e-05,
+      "loss": 1.7527,
+      "step": 656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.187871581450654e-05,
+      "loss": 1.8011,
+      "step": 657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1759809750297268e-05,
+      "loss": 1.6741,
+      "step": 658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.164090368608799e-05,
+      "loss": 1.7341,
+      "step": 659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.152199762187872e-05,
+      "loss": 1.7072,
+      "step": 660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140309155766944e-05,
+      "loss": 1.7499,
+      "step": 661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128418549346017e-05,
+      "loss": 1.8448,
+      "step": 662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.116527942925089e-05,
+      "loss": 1.7597,
+      "step": 663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.104637336504162e-05,
+      "loss": 1.7258,
+      "step": 664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0927467300832342e-05,
+      "loss": 1.7689,
+      "step": 665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.080856123662307e-05,
+      "loss": 1.7795,
+      "step": 666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0689655172413793e-05,
+      "loss": 1.6662,
+      "step": 667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0570749108204522e-05,
+      "loss": 1.6029,
+      "step": 668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0451843043995244e-05,
+      "loss": 1.6668,
+      "step": 669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033293697978597e-05,
+      "loss": 1.7547,
+      "step": 670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0214030915576695e-05,
+      "loss": 1.6178,
+      "step": 671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009512485136742e-05,
+      "loss": 1.6927,
+      "step": 672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9976218787158146e-05,
+      "loss": 1.6911,
+      "step": 673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985731272294887e-05,
+      "loss": 1.6884,
+      "step": 674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9738406658739597e-05,
+      "loss": 1.7416,
+      "step": 675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9619500594530322e-05,
+      "loss": 1.7161,
+      "step": 676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9500594530321048e-05,
+      "loss": 1.8103,
+      "step": 677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9381688466111773e-05,
+      "loss": 1.6993,
+      "step": 678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92627824019025e-05,
+      "loss": 1.7343,
+      "step": 679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9143876337693224e-05,
+      "loss": 1.733,
+      "step": 680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902497027348395e-05,
+      "loss": 1.7764,
+      "step": 681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8906064209274675e-05,
+      "loss": 1.7767,
+      "step": 682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.87871581450654e-05,
+      "loss": 1.6872,
+      "step": 683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8668252080856126e-05,
+      "loss": 1.7413,
+      "step": 684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8549346016646848e-05,
+      "loss": 1.8013,
+      "step": 685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8430439952437577e-05,
+      "loss": 1.7289,
+      "step": 686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.83115338882283e-05,
+      "loss": 1.7307,
+      "step": 687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8192627824019025e-05,
+      "loss": 1.7933,
+      "step": 688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.807372175980975e-05,
+      "loss": 1.706,
+      "step": 689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7954815695600476e-05,
+      "loss": 1.8917,
+      "step": 690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.78359096313912e-05,
+      "loss": 1.7024,
+      "step": 691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7717003567181926e-05,
+      "loss": 1.7541,
+      "step": 692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7598097502972652e-05,
+      "loss": 1.6181,
+      "step": 693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7479191438763377e-05,
+      "loss": 1.6308,
+      "step": 694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7360285374554103e-05,
+      "loss": 1.7445,
+      "step": 695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 1.7265,
+      "step": 696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7122473246135554e-05,
+      "loss": 1.6639,
+      "step": 697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700356718192628e-05,
+      "loss": 1.728,
+      "step": 698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6884661117717005e-05,
+      "loss": 1.6708,
+      "step": 699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.676575505350773e-05,
+      "loss": 1.6943,
+      "step": 700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6646848989298456e-05,
+      "loss": 1.715,
+      "step": 701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6527942925089178e-05,
+      "loss": 1.754,
+      "step": 702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6409036860879907e-05,
+      "loss": 1.7819,
+      "step": 703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.629013079667063e-05,
+      "loss": 1.7606,
+      "step": 704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6171224732461357e-05,
+      "loss": 1.7331,
+      "step": 705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.605231866825208e-05,
+      "loss": 1.7167,
+      "step": 706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.593341260404281e-05,
+      "loss": 1.6439,
+      "step": 707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.581450653983353e-05,
+      "loss": 1.7538,
+      "step": 708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.569560047562426e-05,
+      "loss": 1.6926,
+      "step": 709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.557669441141498e-05,
+      "loss": 1.7003,
+      "step": 710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.545778834720571e-05,
+      "loss": 1.6388,
+      "step": 711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5338882282996432e-05,
+      "loss": 1.7999,
+      "step": 712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5219976218787158e-05,
+      "loss": 1.8244,
+      "step": 713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5101070154577885e-05,
+      "loss": 1.7612,
+      "step": 714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4982164090368609e-05,
+      "loss": 1.6714,
+      "step": 715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4863258026159336e-05,
+      "loss": 1.7071,
+      "step": 716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.474435196195006e-05,
+      "loss": 1.774,
+      "step": 717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4625445897740787e-05,
+      "loss": 1.669,
+      "step": 718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.450653983353151e-05,
+      "loss": 1.6811,
+      "step": 719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4387633769322236e-05,
+      "loss": 1.7823,
+      "step": 720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4268727705112961e-05,
+      "loss": 1.7448,
+      "step": 721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4149821640903687e-05,
+      "loss": 1.685,
+      "step": 722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4030915576694412e-05,
+      "loss": 1.7258,
+      "step": 723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3912009512485138e-05,
+      "loss": 1.7228,
+      "step": 724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 1.6932,
+      "step": 725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3674197384066589e-05,
+      "loss": 1.7379,
+      "step": 726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3555291319857313e-05,
+      "loss": 1.6923,
+      "step": 727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3436385255648038e-05,
+      "loss": 1.6931,
+      "step": 728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3317479191438763e-05,
+      "loss": 1.7949,
+      "step": 729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3198573127229489e-05,
+      "loss": 1.6087,
+      "step": 730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3079667063020214e-05,
+      "loss": 1.677,
+      "step": 731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.296076099881094e-05,
+      "loss": 1.7484,
+      "step": 732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2841854934601665e-05,
+      "loss": 1.7916,
+      "step": 733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2722948870392389e-05,
+      "loss": 1.7321,
+      "step": 734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2604042806183116e-05,
+      "loss": 1.7227,
+      "step": 735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.248513674197384e-05,
+      "loss": 1.7293,
+      "step": 736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2366230677764565e-05,
+      "loss": 1.7587,
+      "step": 737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2247324613555291e-05,
+      "loss": 1.7297,
+      "step": 738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2128418549346016e-05,
+      "loss": 1.6164,
+      "step": 739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2009512485136742e-05,
+      "loss": 1.6858,
+      "step": 740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1890606420927467e-05,
+      "loss": 1.6656,
+      "step": 741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1771700356718193e-05,
+      "loss": 1.6742,
+      "step": 742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1652794292508918e-05,
+      "loss": 1.7152,
+      "step": 743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1533888228299644e-05,
+      "loss": 1.7866,
+      "step": 744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.141498216409037e-05,
+      "loss": 1.7228,
+      "step": 745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1296076099881095e-05,
+      "loss": 1.7881,
+      "step": 746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.117717003567182e-05,
+      "loss": 1.6299,
+      "step": 747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1058263971462546e-05,
+      "loss": 1.7067,
+      "step": 748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.093935790725327e-05,
+      "loss": 1.613,
+      "step": 749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0820451843043995e-05,
+      "loss": 1.7521,
+      "step": 750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.070154577883472e-05,
+      "loss": 1.6697,
+      "step": 751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0582639714625446e-05,
+      "loss": 1.6753,
+      "step": 752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0463733650416171e-05,
+      "loss": 1.6678,
+      "step": 753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0344827586206897e-05,
+      "loss": 1.6927,
+      "step": 754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0225921521997622e-05,
+      "loss": 1.75,
+      "step": 755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0107015457788348e-05,
+      "loss": 1.7282,
+      "step": 756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.988109393579073e-06,
+      "loss": 1.7415,
+      "step": 757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.869203329369798e-06,
+      "loss": 1.8399,
+      "step": 758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.750297265160524e-06,
+      "loss": 1.7497,
+      "step": 759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.63139120095125e-06,
+      "loss": 1.8685,
+      "step": 760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.512485136741975e-06,
+      "loss": 1.6439,
+      "step": 761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.3935790725327e-06,
+      "loss": 1.7027,
+      "step": 762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.274673008323424e-06,
+      "loss": 1.7368,
+      "step": 763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.15576694411415e-06,
+      "loss": 1.7389,
+      "step": 764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.036860879904875e-06,
+      "loss": 1.7409,
+      "step": 765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.9179548156956e-06,
+      "loss": 1.7213,
+      "step": 766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.799048751486326e-06,
+      "loss": 1.6725,
+      "step": 767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.680142687277051e-06,
+      "loss": 1.6685,
+      "step": 768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.561236623067777e-06,
+      "loss": 1.7212,
+      "step": 769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.442330558858502e-06,
+      "loss": 1.64,
+      "step": 770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.323424494649228e-06,
+      "loss": 1.827,
+      "step": 771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.204518430439953e-06,
+      "loss": 1.73,
+      "step": 772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.085612366230679e-06,
+      "loss": 1.754,
+      "step": 773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.966706302021404e-06,
+      "loss": 1.7209,
+      "step": 774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.84780023781213e-06,
+      "loss": 1.6482,
+      "step": 775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.728894173602855e-06,
+      "loss": 1.7035,
+      "step": 776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.609988109393579e-06,
+      "loss": 1.6127,
+      "step": 777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.491082045184304e-06,
+      "loss": 1.8336,
+      "step": 778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.37217598097503e-06,
+      "loss": 1.7032,
+      "step": 779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.253269916765755e-06,
+      "loss": 1.6551,
+      "step": 780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.134363852556481e-06,
+      "loss": 1.7107,
+      "step": 781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.015457788347206e-06,
+      "loss": 1.6169,
+      "step": 782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 1.5955,
+      "step": 783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.777645659928656e-06,
+      "loss": 1.722,
+      "step": 784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.658739595719382e-06,
+      "loss": 1.7515,
+      "step": 785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.539833531510107e-06,
+      "loss": 1.756,
+      "step": 786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.420927467300833e-06,
+      "loss": 1.6729,
+      "step": 787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.302021403091558e-06,
+      "loss": 1.7191,
+      "step": 788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.183115338882283e-06,
+      "loss": 1.78,
+      "step": 789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.064209274673008e-06,
+      "loss": 1.7046,
+      "step": 790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.945303210463734e-06,
+      "loss": 1.6924,
+      "step": 791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.826397146254459e-06,
+      "loss": 1.6932,
+      "step": 792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.707491082045185e-06,
+      "loss": 1.7385,
+      "step": 793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.58858501783591e-06,
+      "loss": 1.7397,
+      "step": 794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.469678953626635e-06,
+      "loss": 1.6176,
+      "step": 795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.35077288941736e-06,
+      "loss": 1.7231,
+      "step": 796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.231866825208086e-06,
+      "loss": 1.7129,
+      "step": 797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.112960760998811e-06,
+      "loss": 1.701,
+      "step": 798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9940546967895365e-06,
+      "loss": 1.7318,
+      "step": 799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.875148632580262e-06,
+      "loss": 1.7185,
+      "step": 800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.7562425683709874e-06,
+      "loss": 1.7562,
+      "step": 801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.637336504161712e-06,
+      "loss": 1.7011,
+      "step": 802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5184304399524375e-06,
+      "loss": 1.6884,
+      "step": 803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399524375743163e-06,
+      "loss": 1.6938,
+      "step": 804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2806183115338884e-06,
+      "loss": 1.7102,
+      "step": 805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.161712247324614e-06,
+      "loss": 1.8204,
+      "step": 806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.042806183115339e-06,
+      "loss": 1.6964,
+      "step": 807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.923900118906065e-06,
+      "loss": 1.7789,
+      "step": 808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8049940546967894e-06,
+      "loss": 1.6964,
+      "step": 809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.686087990487515e-06,
+      "loss": 1.7336,
+      "step": 810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5671819262782404e-06,
+      "loss": 1.6915,
+      "step": 811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 1.7074,
+      "step": 812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.329369797859691e-06,
+      "loss": 1.7323,
+      "step": 813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2104637336504163e-06,
+      "loss": 1.7204,
+      "step": 814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0915576694411414e-06,
+      "loss": 1.7736,
+      "step": 815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.972651605231867e-06,
+      "loss": 1.7981,
+      "step": 816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8537455410225923e-06,
+      "loss": 1.7996,
+      "step": 817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7348394768133173e-06,
+      "loss": 1.6551,
+      "step": 818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.615933412604043e-06,
+      "loss": 1.5696,
+      "step": 819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4970273483947683e-06,
+      "loss": 1.8418,
+      "step": 820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3781212841854937e-06,
+      "loss": 1.7655,
+      "step": 821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2592152199762188e-06,
+      "loss": 1.6867,
+      "step": 822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1403091557669442e-06,
+      "loss": 1.6559,
+      "step": 823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0214030915576697e-06,
+      "loss": 1.7492,
+      "step": 824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9024970273483947e-06,
+      "loss": 1.6673,
+      "step": 825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7835909631391202e-06,
+      "loss": 1.682,
+      "step": 826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6646848989298454e-06,
+      "loss": 1.7276,
+      "step": 827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5457788347205707e-06,
+      "loss": 1.7947,
+      "step": 828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4268727705112961e-06,
+      "loss": 1.8449,
+      "step": 829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3079667063020214e-06,
+      "loss": 1.6148,
+      "step": 830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1890606420927469e-06,
+      "loss": 1.845,
+      "step": 831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0701545778834721e-06,
+      "loss": 1.6853,
+      "step": 832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.512485136741974e-07,
+      "loss": 1.7254,
+      "step": 833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.323424494649227e-07,
+      "loss": 1.696,
+      "step": 834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.134363852556481e-07,
+      "loss": 1.8302,
+      "step": 835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.945303210463734e-07,
+      "loss": 1.771,
+      "step": 836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.756242568370987e-07,
+      "loss": 1.7109,
+      "step": 837
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5671819262782404e-07,
+      "loss": 1.6868,
+      "step": 838
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3781212841854934e-07,
+      "loss": 1.7072,
+      "step": 839
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1890606420927467e-07,
+      "loss": 1.674,
+      "step": 840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.7679,
+      "step": 841
+    }
+  ],
+  "max_steps": 841,
+  "num_train_epochs": 1,
+  "total_flos": 9.501901140996915e+16,
+  "trial_name": null,
+  "trial_params": null
+}