diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13290 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.997172745264349,
+  "eval_steps": 500,
+  "global_step": 4420,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 6.8903,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5037593984962404e-08,
+      "loss": 7.056,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5112781954887216e-08,
+      "loss": 7.1163,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.518796992481202e-08,
+      "loss": 7.0162,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0526315789473683e-07,
+      "loss": 7.1292,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3533834586466163e-07,
+      "loss": 7.0274,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6541353383458646e-07,
+      "loss": 7.0955,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9548872180451126e-07,
+      "loss": 7.0135,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2556390977443606e-07,
+      "loss": 7.191,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.556390977443609e-07,
+      "loss": 6.8574,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.857142857142857e-07,
+      "loss": 7.0637,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.157894736842105e-07,
+      "loss": 6.6183,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.458646616541353e-07,
+      "loss": 6.7347,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.759398496240601e-07,
+      "loss": 6.6015,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.060150375939849e-07,
+      "loss": 6.4449,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.3609022556390975e-07,
+      "loss": 6.1891,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6616541353383456e-07,
+      "loss": 6.3982,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962406015037593e-07,
+      "loss": 6.1609,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.263157894736842e-07,
+      "loss": 6.0694,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.56390977443609e-07,
+      "loss": 6.3096,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.864661654135338e-07,
+      "loss": 5.8542,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.165413533834586e-07,
+      "loss": 5.871,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.466165413533834e-07,
+      "loss": 5.7045,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.766917293233082e-07,
+      "loss": 5.9885,
+      "step": 48
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.06766917293233e-07,
+      "loss": 5.7822,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.368421052631578e-07,
+      "loss": 5.5907,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.669172932330827e-07,
+      "loss": 5.3985,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.969924812030074e-07,
+      "loss": 5.5724,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.270676691729323e-07,
+      "loss": 5.0061,
+      "step": 58
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.57142857142857e-07,
+      "loss": 4.6891,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.872180451127819e-07,
+      "loss": 5.3034,
+      "step": 62
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.172932330827066e-07,
+      "loss": 4.9166,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.473684210526315e-07,
+      "loss": 4.9849,
+      "step": 66
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774436090225563e-07,
+      "loss": 4.9612,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0075187969924813e-06,
+      "loss": 4.8004,
+      "step": 70
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.037593984962406e-06,
+      "loss": 4.7059,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.0676691729323308e-06,
+      "loss": 4.3691,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.0977443609022555e-06,
+      "loss": 4.5099,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1278195488721805e-06,
+      "loss": 4.739,
+      "step": 78
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1578947368421053e-06,
+      "loss": 4.6763,
+      "step": 80
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.18796992481203e-06,
+      "loss": 4.3841,
+      "step": 82
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.218045112781955e-06,
+      "loss": 4.0368,
+      "step": 84
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2481203007518797e-06,
+      "loss": 4.1215,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2781954887218045e-06,
+      "loss": 4.5899,
+      "step": 88
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3082706766917292e-06,
+      "loss": 4.3187,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3383458646616542e-06,
+      "loss": 4.0256,
+      "step": 92
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.368421052631579e-06,
+      "loss": 4.127,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3984962406015037e-06,
+      "loss": 4.4956,
+      "step": 96
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 3.9016,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.4586466165413534e-06,
+      "loss": 3.762,
+      "step": 100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.4887218045112781e-06,
+      "loss": 4.0754,
+      "step": 102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5187969924812029e-06,
+      "loss": 4.006,
+      "step": 104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5488721804511278e-06,
+      "loss": 3.9874,
+      "step": 106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 3.9949,
+      "step": 108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6090225563909773e-06,
+      "loss": 3.7541,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6390977443609023e-06,
+      "loss": 4.2143,
+      "step": 112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.669172932330827e-06,
+      "loss": 3.8826,
+      "step": 114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6992481203007518e-06,
+      "loss": 4.4338,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7293233082706765e-06,
+      "loss": 3.6651,
+      "step": 118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7593984962406015e-06,
+      "loss": 3.9413,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7894736842105262e-06,
+      "loss": 4.2412,
+      "step": 122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.819548872180451e-06,
+      "loss": 4.0753,
+      "step": 124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.849624060150376e-06,
+      "loss": 3.5147,
+      "step": 126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8796992481203007e-06,
+      "loss": 3.5877,
+      "step": 128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9097744360902255e-06,
+      "loss": 3.9202,
+      "step": 130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9398496240601504e-06,
+      "loss": 4.2452,
+      "step": 132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.969924812030075e-06,
+      "loss": 4.2966,
+      "step": 134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-06,
+      "loss": 4.0258,
+      "step": 136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999989259544593e-06,
+      "loss": 3.2049,
+      "step": 138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9999957038201444e-06,
+      "loss": 3.8649,
+      "step": 140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.999990333603977e-06,
+      "loss": 3.9265,
+      "step": 142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999982815317493e-06,
+      "loss": 3.3749,
+      "step": 144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999973148976841e-06,
+      "loss": 3.6495,
+      "step": 146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999961334602787e-06,
+      "loss": 3.789,
+      "step": 148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999947372220708e-06,
+      "loss": 3.4248,
+      "step": 150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9999312618605972e-06,
+      "loss": 3.3628,
+      "step": 152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999913003557061e-06,
+      "loss": 3.669,
+      "step": 154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998925973493196e-06,
+      "loss": 3.5372,
+      "step": 156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9998700432812073e-06,
+      "loss": 3.2437,
+      "step": 158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.999845341401173e-06,
+      "loss": 3.9666,
+      "step": 160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.999818491762278e-06,
+      "loss": 3.5522,
+      "step": 162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9997894944221986e-06,
+      "loss": 3.765,
+      "step": 164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9997583494432226e-06,
+      "loss": 3.489,
+      "step": 166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.999725056892252e-06,
+      "loss": 3.7294,
+      "step": 168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.999689616840804e-06,
+      "loss": 3.3408,
+      "step": 170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9996520293650056e-06,
+      "loss": 3.5929,
+      "step": 172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9996122945455987e-06,
+      "loss": 3.5767,
+      "step": 174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9995704124679367e-06,
+      "loss": 3.167,
+      "step": 176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.999526383221987e-06,
+      "loss": 3.651,
+      "step": 178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.999480206902328e-06,
+      "loss": 3.3083,
+      "step": 180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9994318836081506e-06,
+      "loss": 3.3985,
+      "step": 182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9993814134432575e-06,
+      "loss": 3.2612,
+      "step": 184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9993287965160636e-06,
+      "loss": 3.7523,
+      "step": 186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9992740329395945e-06,
+      "loss": 3.9041,
+      "step": 188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9992171228314873e-06,
+      "loss": 3.5194,
+      "step": 190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9991580663139904e-06,
+      "loss": 3.3409,
+      "step": 192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.999096863513963e-06,
+      "loss": 3.3884,
+      "step": 194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.999033514562873e-06,
+      "loss": 3.6586,
+      "step": 196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9989680195968008e-06,
+      "loss": 3.4453,
+      "step": 198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.998900378756435e-06,
+      "loss": 3.7887,
+      "step": 200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9988305921870747e-06,
+      "loss": 3.2259,
+      "step": 202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.998758660038628e-06,
+      "loss": 3.926,
+      "step": 204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.998684582465611e-06,
+      "loss": 3.5587,
+      "step": 206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.998608359627149e-06,
+      "loss": 3.2248,
+      "step": 208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.998529991686976e-06,
+      "loss": 3.4838,
+      "step": 210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.998449478813434e-06,
+      "loss": 3.4091,
+      "step": 212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.998366821179471e-06,
+      "loss": 3.2509,
+      "step": 214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.998282018962644e-06,
+      "loss": 3.3953,
+      "step": 216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9981950723451152e-06,
+      "loss": 3.3451,
+      "step": 218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.998105981513654e-06,
+      "loss": 3.3634,
+      "step": 220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.998014746659636e-06,
+      "loss": 3.2317,
+      "step": 222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9979213679790414e-06,
+      "loss": 3.1458,
+      "step": 224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9978258456724567e-06,
+      "loss": 3.6836,
+      "step": 226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997728179945072e-06,
+      "loss": 3.2641,
+      "step": 228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997628371006683e-06,
+      "loss": 3.4994,
+      "step": 230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997526419071687e-06,
+      "loss": 3.314,
+      "step": 232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.997422324359087e-06,
+      "loss": 3.4686,
+      "step": 234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9973160870924883e-06,
+      "loss": 3.5634,
+      "step": 236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9972077075000974e-06,
+      "loss": 3.3327,
+      "step": 238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9970971858147235e-06,
+      "loss": 3.5988,
+      "step": 240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.996984522273778e-06,
+      "loss": 3.6566,
+      "step": 242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.996869717119272e-06,
+      "loss": 3.5185,
+      "step": 244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9967527705978177e-06,
+      "loss": 3.1404,
+      "step": 246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.996633682960626e-06,
+      "loss": 3.1059,
+      "step": 248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.996512454463509e-06,
+      "loss": 3.5391,
+      "step": 250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9963890853668764e-06,
+      "loss": 3.5021,
+      "step": 252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9962635759357355e-06,
+      "loss": 2.8194,
+      "step": 254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.996135926439693e-06,
+      "loss": 3.0764,
+      "step": 256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9960061371529506e-06,
+      "loss": 3.1193,
+      "step": 258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9958742083543086e-06,
+      "loss": 3.2474,
+      "step": 260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9957401403271615e-06,
+      "loss": 2.932,
+      "step": 262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.995603933359499e-06,
+      "loss": 2.8882,
+      "step": 264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.995465587743908e-06,
+      "loss": 3.2823,
+      "step": 266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9953251037775655e-06,
+      "loss": 3.1695,
+      "step": 268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9951824817622444e-06,
+      "loss": 3.1922,
+      "step": 270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9950377220043103e-06,
+      "loss": 3.1263,
+      "step": 272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9948908248147202e-06,
+      "loss": 2.8974,
+      "step": 274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.994741790509022e-06,
+      "loss": 3.3318,
+      "step": 276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.994590619407356e-06,
+      "loss": 3.1373,
+      "step": 278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9944373118344505e-06,
+      "loss": 3.0683,
+      "step": 280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9942818681196243e-06,
+      "loss": 3.1446,
+      "step": 282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9941242885967856e-06,
+      "loss": 3.1867,
+      "step": 284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9939645736044283e-06,
+      "loss": 3.2701,
+      "step": 286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.993802723485636e-06,
+      "loss": 3.0124,
+      "step": 288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9936387385880763e-06,
+      "loss": 3.1993,
+      "step": 290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993472619264004e-06,
+      "loss": 2.841,
+      "step": 292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993304365870259e-06,
+      "loss": 3.2096,
+      "step": 294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993133978768265e-06,
+      "loss": 2.9372,
+      "step": 296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929614583240286e-06,
+      "loss": 3.1033,
+      "step": 298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9927868049081394e-06,
+      "loss": 2.9037,
+      "step": 300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992610018895769e-06,
+      "loss": 2.9596,
+      "step": 302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9924311006666695e-06,
+      "loss": 2.8743,
+      "step": 304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.992250050605174e-06,
+      "loss": 2.7332,
+      "step": 306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9920668691001946e-06,
+      "loss": 3.3899,
+      "step": 308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9918815565452215e-06,
+      "loss": 2.8847,
+      "step": 310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9916941133383232e-06,
+      "loss": 3.1096,
+      "step": 312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9915045398821445e-06,
+      "loss": 2.9963,
+      "step": 314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.991312836583907e-06,
+      "loss": 3.1046,
+      "step": 316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.991119003855406e-06,
+      "loss": 3.2598,
+      "step": 318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9909230421130126e-06,
+      "loss": 2.7556,
+      "step": 320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.99072495177767e-06,
+      "loss": 2.9834,
+      "step": 322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.990524733274895e-06,
+      "loss": 3.0658,
+      "step": 324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9903223870347743e-06,
+      "loss": 3.1869,
+      "step": 326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9901179134919664e-06,
+      "loss": 2.9748,
+      "step": 328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.989911313085699e-06,
+      "loss": 2.9126,
+      "step": 330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.989702586259769e-06,
+      "loss": 3.136,
+      "step": 332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.98949173346254e-06,
+      "loss": 3.0548,
+      "step": 334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9892787551469436e-06,
+      "loss": 3.0525,
+      "step": 336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9890636517704765e-06,
+      "loss": 3.0175,
+      "step": 338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9888464237952e-06,
+      "loss": 3.0134,
+      "step": 340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9886270716877395e-06,
+      "loss": 3.0958,
+      "step": 342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.988405595919284e-06,
+      "loss": 3.0671,
+      "step": 344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.988181996965583e-06,
+      "loss": 3.3007,
+      "step": 346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9879562753069475e-06,
+      "loss": 3.0881,
+      "step": 348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.987728431428249e-06,
+      "loss": 2.901,
+      "step": 350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987498465818915e-06,
+      "loss": 3.126,
+      "step": 352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9872663789729353e-06,
+      "loss": 2.8135,
+      "step": 354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9870321713888513e-06,
+      "loss": 2.8591,
+      "step": 356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9867958435697627e-06,
+      "loss": 3.0549,
+      "step": 358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9865573960233237e-06,
+      "loss": 3.0025,
+      "step": 360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9863168292617415e-06,
+      "loss": 2.8908,
+      "step": 362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9860741438017745e-06,
+      "loss": 2.753,
+      "step": 364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.985829340164734e-06,
+      "loss": 3.3022,
+      "step": 366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9855824188764805e-06,
+      "loss": 2.6739,
+      "step": 368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9853333804674227e-06,
+      "loss": 2.9021,
+      "step": 370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9850822254725187e-06,
+      "loss": 2.796,
+      "step": 372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9848289544312713e-06,
+      "loss": 2.9392,
+      "step": 374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9845735678877307e-06,
+      "loss": 3.1737,
+      "step": 376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9843160663904896e-06,
+      "loss": 2.7822,
+      "step": 378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9840564504926856e-06,
+      "loss": 3.1038,
+      "step": 380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9837947207519966e-06,
+      "loss": 2.881,
+      "step": 382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.983530877730642e-06,
+      "loss": 2.5971,
+      "step": 384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.983264921995381e-06,
+      "loss": 2.7485,
+      "step": 386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9829968541175103e-06,
+      "loss": 2.6596,
+      "step": 388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9827266746728644e-06,
+      "loss": 2.7252,
+      "step": 390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9824543842418137e-06,
+      "loss": 2.8186,
+      "step": 392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.982179983409262e-06,
+      "loss": 2.7891,
+      "step": 394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.981903472764648e-06,
+      "loss": 2.6887,
+      "step": 396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.981624852901941e-06,
+      "loss": 2.5659,
+      "step": 398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.981344124419643e-06,
+      "loss": 2.8682,
+      "step": 400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9810612879207835e-06,
+      "loss": 2.423,
+      "step": 402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.980776344012921e-06,
+      "loss": 2.8891,
+      "step": 404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9804892933081412e-06,
+      "loss": 2.84,
+      "step": 406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.980200136423055e-06,
+      "loss": 2.7778,
+      "step": 408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.979908873978798e-06,
+      "loss": 3.15,
+      "step": 410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9796155066010285e-06,
+      "loss": 2.8711,
+      "step": 412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9793200349199264e-06,
+      "loss": 2.9754,
+      "step": 414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.979022459570191e-06,
+      "loss": 2.6299,
+      "step": 416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.978722781191042e-06,
+      "loss": 2.8453,
+      "step": 418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.978421000426216e-06,
+      "loss": 2.7152,
+      "step": 420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.978117117923965e-06,
+      "loss": 3.0451,
+      "step": 422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9778111343370563e-06,
+      "loss": 2.8192,
+      "step": 424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.977503050322771e-06,
+      "loss": 3.0503,
+      "step": 426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9771928665429016e-06,
+      "loss": 2.803,
+      "step": 428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9768805836637507e-06,
+      "loss": 3.0309,
+      "step": 430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.97656620235613e-06,
+      "loss": 3.2594,
+      "step": 432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9762497232953607e-06,
+      "loss": 2.7724,
+      "step": 434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.975931147161267e-06,
+      "loss": 2.9467,
+      "step": 436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9756104746381803e-06,
+      "loss": 2.4082,
+      "step": 438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9752877064149344e-06,
+      "loss": 2.8044,
+      "step": 440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9749628431848647e-06,
+      "loss": 2.7142,
+      "step": 442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9746358856458065e-06,
+      "loss": 2.9282,
+      "step": 444
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9743068345000954e-06,
+      "loss": 2.8878,
+      "step": 446
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9739756904545625e-06,
+      "loss": 2.6279,
+      "step": 448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9736424542205353e-06,
+      "loss": 3.3394,
+      "step": 450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.973307126513836e-06,
+      "loss": 3.0321,
+      "step": 452
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.972969708054779e-06,
+      "loss": 2.6647,
+      "step": 454
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9726301995681697e-06,
+      "loss": 2.8719,
+      "step": 456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9722886017833032e-06,
+      "loss": 2.6796,
+      "step": 458
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.971944915433963e-06,
+      "loss": 2.7943,
+      "step": 460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9715991412584184e-06,
+      "loss": 2.5229,
+      "step": 462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9712512799994236e-06,
+      "loss": 2.6882,
+      "step": 464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.970901332404217e-06,
+      "loss": 3.1826,
+      "step": 466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9705492992245173e-06,
+      "loss": 2.8889,
+      "step": 468
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9701951812165236e-06,
+      "loss": 2.9288,
+      "step": 470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9698389791409147e-06,
+      "loss": 3.2185,
+      "step": 472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.969480693762844e-06,
+      "loss": 2.8005,
+      "step": 474
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9691203258519414e-06,
+      "loss": 2.5454,
+      "step": 476
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9687578761823105e-06,
+      "loss": 2.9237,
+      "step": 478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9683933455325258e-06,
+      "loss": 2.5245,
+      "step": 480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9680267346856323e-06,
+      "loss": 2.8414,
+      "step": 482
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.967842649434193e-06,
+      "loss": 2.732,
+      "step": 484
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9674729197696204e-06,
+      "loss": 3.1263,
+      "step": 486
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.967101111885097e-06,
+      "loss": 2.9437,
+      "step": 488
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.966727226579301e-06,
+      "loss": 2.3668,
+      "step": 490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.966351264655371e-06,
+      "loss": 2.9765,
+      "step": 492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9659732269209086e-06,
+      "loss": 2.9363,
+      "step": 494
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.965593114187972e-06,
+      "loss": 3.0973,
+      "step": 496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.965210927273079e-06,
+      "loss": 2.8855,
+      "step": 498
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.964826666997202e-06,
+      "loss": 2.6448,
+      "step": 500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.964440334185766e-06,
+      "loss": 2.9767,
+      "step": 502
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.96405192966865e-06,
+      "loss": 2.8073,
+      "step": 504
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9636614542801816e-06,
+      "loss": 2.8848,
+      "step": 506
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9632689088591385e-06,
+      "loss": 2.9247,
+      "step": 508
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9628742942487427e-06,
+      "loss": 2.8988,
+      "step": 510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9624776112966636e-06,
+      "loss": 2.7298,
+      "step": 512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9620788608550118e-06,
+      "loss": 2.3712,
+      "step": 514
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961678043780339e-06,
+      "loss": 2.802,
+      "step": 516
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9612751609336377e-06,
+      "loss": 2.6516,
+      "step": 518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608702131803363e-06,
+      "loss": 2.5732,
+      "step": 520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9604632013902996e-06,
+      "loss": 2.5408,
+      "step": 522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9600541264378266e-06,
+      "loss": 2.9325,
+      "step": 524
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9596429892016466e-06,
+      "loss": 2.7673,
+      "step": 526
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9592297905649202e-06,
+      "loss": 3.0604,
+      "step": 528
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9588145314152364e-06,
+      "loss": 2.6299,
+      "step": 530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.958397212644609e-06,
+      "loss": 2.8273,
+      "step": 532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9579778351494762e-06,
+      "loss": 2.5243,
+      "step": 534
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9575563998306996e-06,
+      "loss": 2.51,
+      "step": 536
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.957132907593561e-06,
+      "loss": 2.7396,
+      "step": 538
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.95670735934776e-06,
+      "loss": 2.5031,
+      "step": 540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9562797560074133e-06,
+      "loss": 2.6223,
+      "step": 542
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.955850098491051e-06,
+      "loss": 2.6593,
+      "step": 544
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9554183877216173e-06,
+      "loss": 2.7128,
+      "step": 546
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.954984624626466e-06,
+      "loss": 2.6417,
+      "step": 548
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.95454881013736e-06,
+      "loss": 2.3611,
+      "step": 550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.954110945190468e-06,
+      "loss": 2.5021,
+      "step": 552
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9536710307263644e-06,
+      "loss": 2.7342,
+      "step": 554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9532290676900247e-06,
+      "loss": 2.5659,
+      "step": 556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9527850570308266e-06,
+      "loss": 2.546,
+      "step": 558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9523389997025453e-06,
+      "loss": 2.7346,
+      "step": 560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.951890896663352e-06,
+      "loss": 2.8895,
+      "step": 562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9514407488758135e-06,
+      "loss": 2.5559,
+      "step": 564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.950988557306888e-06,
+      "loss": 2.5673,
+      "step": 566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9505343229279237e-06,
+      "loss": 2.4996,
+      "step": 568
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.950078046714658e-06,
+      "loss": 3.0601,
+      "step": 570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9496197296472143e-06,
+      "loss": 2.5457,
+      "step": 572
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.949159372710098e-06,
+      "loss": 2.5823,
+      "step": 574
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9486969768921986e-06,
+      "loss": 2.6923,
+      "step": 576
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9482325431867846e-06,
+      "loss": 2.5386,
+      "step": 578
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9477660725915013e-06,
+      "loss": 2.483,
+      "step": 580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9472975661083705e-06,
+      "loss": 2.4662,
+      "step": 582
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9468270247437868e-06,
+      "loss": 2.5708,
+      "step": 584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9463544495085153e-06,
+      "loss": 2.5796,
+      "step": 586
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9458798414176913e-06,
+      "loss": 2.5566,
+      "step": 588
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.945403201490816e-06,
+      "loss": 2.5595,
+      "step": 590
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9449245307517553e-06,
+      "loss": 2.4607,
+      "step": 592
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9444438302287374e-06,
+      "loss": 2.2646,
+      "step": 594
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9439611009543515e-06,
+      "loss": 2.5673,
+      "step": 596
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.943476343965543e-06,
+      "loss": 2.371,
+      "step": 598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9429895603036153e-06,
+      "loss": 2.5064,
+      "step": 600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9425007510142224e-06,
+      "loss": 2.7005,
+      "step": 602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9420099171473732e-06,
+      "loss": 2.5411,
+      "step": 604
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9415170597574216e-06,
+      "loss": 2.5514,
+      "step": 606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9410221799030716e-06,
+      "loss": 2.5828,
+      "step": 608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9405252786473694e-06,
+      "loss": 2.3784,
+      "step": 610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.940026357057705e-06,
+      "loss": 2.7243,
+      "step": 612
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9395254162058062e-06,
+      "loss": 2.5178,
+      "step": 614
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9390224571677405e-06,
+      "loss": 2.5705,
+      "step": 616
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9385174810239096e-06,
+      "loss": 2.9351,
+      "step": 618
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9380104888590475e-06,
+      "loss": 2.5937,
+      "step": 620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9375014817622207e-06,
+      "loss": 2.7249,
+      "step": 622
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.936990460826822e-06,
+      "loss": 2.7076,
+      "step": 624
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.936477427150571e-06,
+      "loss": 2.7452,
+      "step": 626
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.935962381835511e-06,
+      "loss": 2.8691,
+      "step": 628
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9354453259880065e-06,
+      "loss": 2.5158,
+      "step": 630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.93492626071874e-06,
+      "loss": 2.89,
+      "step": 632
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.934405187142711e-06,
+      "loss": 2.2936,
+      "step": 634
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.933882106379233e-06,
+      "loss": 2.8166,
+      "step": 636
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9333570195519318e-06,
+      "loss": 2.4871,
+      "step": 638
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.932829927788741e-06,
+      "loss": 2.6609,
+      "step": 640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9323008322219025e-06,
+      "loss": 2.5998,
+      "step": 642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9317697339879607e-06,
+      "loss": 2.4645,
+      "step": 644
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9312366342277637e-06,
+      "loss": 2.7153,
+      "step": 646
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9307015340864578e-06,
+      "loss": 2.7173,
+      "step": 648
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9301644347134873e-06,
+      "loss": 2.4206,
+      "step": 650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.929625337262591e-06,
+      "loss": 2.6368,
+      "step": 652
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9290842428917983e-06,
+      "loss": 2.6202,
+      "step": 654
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.92854115276343e-06,
+      "loss": 2.729,
+      "step": 656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9279960680440924e-06,
+      "loss": 2.3452,
+      "step": 658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.927448989904678e-06,
+      "loss": 2.7506,
+      "step": 660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9268999195203594e-06,
+      "loss": 2.5662,
+      "step": 662
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.926348858070591e-06,
+      "loss": 2.4619,
+      "step": 664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9257958067391024e-06,
+      "loss": 2.5436,
+      "step": 666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.925240766713898e-06,
+      "loss": 2.1889,
+      "step": 668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.924683739187255e-06,
+      "loss": 2.8072,
+      "step": 670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9241247253557184e-06,
+      "loss": 2.6656,
+      "step": 672
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9235637264201013e-06,
+      "loss": 2.6298,
+      "step": 674
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.923000743585481e-06,
+      "loss": 2.5295,
+      "step": 676
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.922435778061195e-06,
+      "loss": 2.5569,
+      "step": 678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.921868831060841e-06,
+      "loss": 2.4206,
+      "step": 680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.921299903802273e-06,
+      "loss": 2.3958,
+      "step": 682
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9207289975075986e-06,
+      "loss": 2.4561,
+      "step": 684
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.920156113403176e-06,
+      "loss": 2.7795,
+      "step": 686
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9195812527196133e-06,
+      "loss": 2.9221,
+      "step": 688
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.919004416691763e-06,
+      "loss": 2.7184,
+      "step": 690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.918425606558721e-06,
+      "loss": 2.7425,
+      "step": 692
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9178448235638255e-06,
+      "loss": 2.7597,
+      "step": 694
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.91726206895465e-06,
+      "loss": 2.512,
+      "step": 696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.916677343983005e-06,
+      "loss": 2.4675,
+      "step": 698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.916090649904933e-06,
+      "loss": 2.6547,
+      "step": 700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9155019879807064e-06,
+      "loss": 2.7396,
+      "step": 702
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9149113594748245e-06,
+      "loss": 2.8,
+      "step": 704
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.914318765656011e-06,
+      "loss": 2.4821,
+      "step": 706
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.913724207797212e-06,
+      "loss": 2.5723,
+      "step": 708
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9131276871755913e-06,
+      "loss": 2.7257,
+      "step": 710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9125292050725296e-06,
+      "loss": 2.6002,
+      "step": 712
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9119287627736212e-06,
+      "loss": 2.3223,
+      "step": 714
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9113263615686704e-06,
+      "loss": 2.6472,
+      "step": 716
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9107220027516902e-06,
+      "loss": 2.4824,
+      "step": 718
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9101156876208984e-06,
+      "loss": 2.9242,
+      "step": 720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9095074174787157e-06,
+      "loss": 2.6182,
+      "step": 722
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9088971936317603e-06,
+      "loss": 2.6885,
+      "step": 724
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.90828501739085e-06,
+      "loss": 2.382,
+      "step": 726
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9076708900709945e-06,
+      "loss": 2.6815,
+      "step": 728
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.907054812991395e-06,
+      "loss": 2.739,
+      "step": 730
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.906436787475442e-06,
+      "loss": 2.8119,
+      "step": 732
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9058168148507097e-06,
+      "loss": 2.492,
+      "step": 734
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9051948964489562e-06,
+      "loss": 2.9267,
+      "step": 736
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9045710336061188e-06,
+      "loss": 2.4951,
+      "step": 738
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9039452276623115e-06,
+      "loss": 2.6114,
+      "step": 740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.903317479961823e-06,
+      "loss": 2.7481,
+      "step": 742
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9026877918531122e-06,
+      "loss": 2.5025,
+      "step": 744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9020561646888065e-06,
+      "loss": 2.5243,
+      "step": 746
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9014225998256986e-06,
+      "loss": 2.7307,
+      "step": 748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9007870986247436e-06,
+      "loss": 2.4616,
+      "step": 750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.900149662451056e-06,
+      "loss": 2.3835,
+      "step": 752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8995102926739064e-06,
+      "loss": 2.4583,
+      "step": 754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8988689906667196e-06,
+      "loss": 2.4495,
+      "step": 756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8982257578070708e-06,
+      "loss": 3.1206,
+      "step": 758
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.897580595476683e-06,
+      "loss": 2.7375,
+      "step": 760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8969335050614231e-06,
+      "loss": 2.6551,
+      "step": 762
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8962844879513003e-06,
+      "loss": 2.3353,
+      "step": 764
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8956335455404626e-06,
+      "loss": 2.7879,
+      "step": 766
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.894980679227194e-06,
+      "loss": 2.5966,
+      "step": 768
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.89432589041391e-06,
+      "loss": 2.5804,
+      "step": 770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8936691805071571e-06,
+      "loss": 2.7343,
+      "step": 772
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8930105509176082e-06,
+      "loss": 2.5984,
+      "step": 774
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8923500030600592e-06,
+      "loss": 2.701,
+      "step": 776
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8916875383534274e-06,
+      "loss": 2.4911,
+      "step": 778
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8910231582207473e-06,
+      "loss": 2.3311,
+      "step": 780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8903568640891673e-06,
+      "loss": 2.6318,
+      "step": 782
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8896886573899487e-06,
+      "loss": 2.2295,
+      "step": 784
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.88901853955846e-06,
+      "loss": 2.7293,
+      "step": 786
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8883465120341756e-06,
+      "loss": 2.5001,
+      "step": 788
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8876725762606714e-06,
+      "loss": 2.2673,
+      "step": 790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.886996733685623e-06,
+      "loss": 2.9348,
+      "step": 792
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.886318985760802e-06,
+      "loss": 2.6025,
+      "step": 794
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8856393339420724e-06,
+      "loss": 2.2533,
+      "step": 796
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.884957779689388e-06,
+      "loss": 2.2767,
+      "step": 798
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8842743244667903e-06,
+      "loss": 2.6349,
+      "step": 800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8835889697424025e-06,
+      "loss": 2.5554,
+      "step": 802
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8829017169884293e-06,
+      "loss": 2.4651,
+      "step": 804
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8822125676811523e-06,
+      "loss": 2.5952,
+      "step": 806
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.881521523300927e-06,
+      "loss": 2.2587,
+      "step": 808
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8808285853321793e-06,
+      "loss": 2.3114,
+      "step": 810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.880133755263403e-06,
+      "loss": 2.7006,
+      "step": 812
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8794370345871574e-06,
+      "loss": 2.4779,
+      "step": 814
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.878738424800061e-06,
+      "loss": 2.304,
+      "step": 816
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8780379274027914e-06,
+      "loss": 2.5801,
+      "step": 818
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8773355439000808e-06,
+      "loss": 2.5185,
+      "step": 820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.876631275800713e-06,
+      "loss": 2.5946,
+      "step": 822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8759251246175202e-06,
+      "loss": 2.7588,
+      "step": 824
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8752170918673794e-06,
+      "loss": 2.7387,
+      "step": 826
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8745071790712097e-06,
+      "loss": 2.331,
+      "step": 828
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8737953877539676e-06,
+      "loss": 2.258,
+      "step": 830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8730817194446465e-06,
+      "loss": 2.9485,
+      "step": 832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8723661756762703e-06,
+      "loss": 2.3587,
+      "step": 834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.871648757985893e-06,
+      "loss": 2.6883,
+      "step": 836
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.870929467914592e-06,
+      "loss": 2.8789,
+      "step": 838
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8702083070074692e-06,
+      "loss": 2.6571,
+      "step": 840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8694852768136431e-06,
+      "loss": 2.6182,
+      "step": 842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8687603788862485e-06,
+      "loss": 2.3781,
+      "step": 844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.868033614782432e-06,
+      "loss": 2.6386,
+      "step": 846
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8673049860633496e-06,
+      "loss": 2.7053,
+      "step": 848
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.866574494294162e-06,
+      "loss": 2.7169,
+      "step": 850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8658421410440316e-06,
+      "loss": 2.5135,
+      "step": 852
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8651079278861203e-06,
+      "loss": 2.5498,
+      "step": 854
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.864371856397585e-06,
+      "loss": 2.6104,
+      "step": 856
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8636339281595739e-06,
+      "loss": 2.5974,
+      "step": 858
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.862894144757224e-06,
+      "loss": 2.5574,
+      "step": 860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8621525077796582e-06,
+      "loss": 2.4691,
+      "step": 862
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8614090188199793e-06,
+      "loss": 2.7892,
+      "step": 864
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8606636794752699e-06,
+      "loss": 2.4791,
+      "step": 866
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8599164913465866e-06,
+      "loss": 2.558,
+      "step": 868
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8591674560389576e-06,
+      "loss": 2.7124,
+      "step": 870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.858416575161379e-06,
+      "loss": 2.4838,
+      "step": 872
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8576638503268111e-06,
+      "loss": 2.6113,
+      "step": 874
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8569092831521757e-06,
+      "loss": 2.7273,
+      "step": 876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8561528752583518e-06,
+      "loss": 2.6104,
+      "step": 878
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8553946282701722e-06,
+      "loss": 2.6673,
+      "step": 880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8546345438164206e-06,
+      "loss": 2.4585,
+      "step": 882
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8538726235298276e-06,
+      "loss": 2.6186,
+      "step": 884
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8531088690470677e-06,
+      "loss": 2.627,
+      "step": 886
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8523432820087548e-06,
+      "loss": 2.4928,
+      "step": 888
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8515758640594394e-06,
+      "loss": 2.538,
+      "step": 890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8508066168476057e-06,
+      "loss": 2.63,
+      "step": 892
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8500355420256662e-06,
+      "loss": 2.3831,
+      "step": 894
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8492626412499602e-06,
+      "loss": 2.5052,
+      "step": 896
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.848487916180749e-06,
+      "loss": 2.3827,
+      "step": 898
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.847711368482212e-06,
+      "loss": 2.4427,
+      "step": 900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8469329998224455e-06,
+      "loss": 2.4639,
+      "step": 902
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8461528118734551e-06,
+      "loss": 2.6087,
+      "step": 904
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8453708063111564e-06,
+      "loss": 2.3654,
+      "step": 906
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8445869848153684e-06,
+      "loss": 2.6367,
+      "step": 908
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8438013490698105e-06,
+      "loss": 2.4576,
+      "step": 910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8430139007621007e-06,
+      "loss": 2.8134,
+      "step": 912
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.842224641583749e-06,
+      "loss": 2.623,
+      "step": 914
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8414335732301564e-06,
+      "loss": 2.3569,
+      "step": 916
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8406406974006097e-06,
+      "loss": 2.3732,
+      "step": 918
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8398460157982783e-06,
+      "loss": 2.3951,
+      "step": 920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.839049530130211e-06,
+      "loss": 2.8091,
+      "step": 922
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.838251242107331e-06,
+      "loss": 2.6456,
+      "step": 924
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8374511534444337e-06,
+      "loss": 2.3516,
+      "step": 926
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8366492658601831e-06,
+      "loss": 2.6604,
+      "step": 928
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8358455810771062e-06,
+      "loss": 2.9681,
+      "step": 930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8350401008215912e-06,
+      "loss": 2.3861,
+      "step": 932
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8342328268238831e-06,
+      "loss": 2.6136,
+      "step": 934
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.83342376081808e-06,
+      "loss": 2.2425,
+      "step": 936
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8326129045421295e-06,
+      "loss": 2.3558,
+      "step": 938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8318002597378243e-06,
+      "loss": 2.5504,
+      "step": 940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8309858281508e-06,
+      "loss": 2.8984,
+      "step": 942
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8301696115305294e-06,
+      "loss": 2.607,
+      "step": 944
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8293516116303205e-06,
+      "loss": 2.6641,
+      "step": 946
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8285318302073113e-06,
+      "loss": 2.3655,
+      "step": 948
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.827710269022467e-06,
+      "loss": 2.649,
+      "step": 950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8268869298405762e-06,
+      "loss": 2.2951,
+      "step": 952
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8260618144302459e-06,
+      "loss": 2.8344,
+      "step": 954
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8252349245638997e-06,
+      "loss": 2.3341,
+      "step": 956
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8244062620177721e-06,
+      "loss": 2.3695,
+      "step": 958
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8235758285719052e-06,
+      "loss": 2.8024,
+      "step": 960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8227436260101468e-06,
+      "loss": 2.5526,
+      "step": 962
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.821909656120143e-06,
+      "loss": 2.47,
+      "step": 964
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.821073920693337e-06,
+      "loss": 2.449,
+      "step": 966
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.820236421524965e-06,
+      "loss": 2.4071,
+      "step": 968
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8193971604140515e-06,
+      "loss": 2.8168,
+      "step": 970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8185561391634056e-06,
+      "loss": 2.561,
+      "step": 972
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8177133595796177e-06,
+      "loss": 2.2887,
+      "step": 974
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.816868823473055e-06,
+      "loss": 2.2514,
+      "step": 976
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.816022532657858e-06,
+      "loss": 2.6151,
+      "step": 978
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8151744889519365e-06,
+      "loss": 2.5533,
+      "step": 980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8143246941769657e-06,
+      "loss": 2.4734,
+      "step": 982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8134731501583817e-06,
+      "loss": 2.2393,
+      "step": 984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8126198587253794e-06,
+      "loss": 2.5585,
+      "step": 986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.811764821710906e-06,
+      "loss": 2.4241,
+      "step": 988
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.810908040951659e-06,
+      "loss": 2.7939,
+      "step": 990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8100495182880813e-06,
+      "loss": 2.5703,
+      "step": 992
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.809189255564358e-06,
+      "loss": 2.7578,
+      "step": 994
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.808327254628412e-06,
+      "loss": 2.458,
+      "step": 996
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8074635173318992e-06,
+      "loss": 2.6176,
+      "step": 998
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8065980455302066e-06,
+      "loss": 2.365,
+      "step": 1000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8057308410824463e-06,
+      "loss": 2.4624,
+      "step": 1002
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8048619058514522e-06,
+      "loss": 2.7063,
+      "step": 1004
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8039912417037763e-06,
+      "loss": 2.4129,
+      "step": 1006
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.803118850509685e-06,
+      "loss": 2.6864,
+      "step": 1008
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8022447341431538e-06,
+      "loss": 3.0802,
+      "step": 1010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8013688944818638e-06,
+      "loss": 2.3165,
+      "step": 1012
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8004913334071992e-06,
+      "loss": 2.4781,
+      "step": 1014
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7996120528042402e-06,
+      "loss": 2.4112,
+      "step": 1016
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7987310545617622e-06,
+      "loss": 2.3141,
+      "step": 1018
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7978483405722296e-06,
+      "loss": 2.2761,
+      "step": 1020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7969639127317925e-06,
+      "loss": 2.5459,
+      "step": 1022
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7960777729402823e-06,
+      "loss": 2.5135,
+      "step": 1024
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.795189923101208e-06,
+      "loss": 2.6812,
+      "step": 1026
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7943003651217522e-06,
+      "loss": 2.4027,
+      "step": 1028
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7934091009127657e-06,
+      "loss": 2.5999,
+      "step": 1030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7925161323887662e-06,
+      "loss": 2.3978,
+      "step": 1032
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7916214614679306e-06,
+      "loss": 2.6241,
+      "step": 1034
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.790725090072094e-06,
+      "loss": 2.7711,
+      "step": 1036
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7898270201267436e-06,
+      "loss": 2.2952,
+      "step": 1038
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7889272535610158e-06,
+      "loss": 2.5796,
+      "step": 1040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7880257923076907e-06,
+      "loss": 2.148,
+      "step": 1042
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7871226383031893e-06,
+      "loss": 2.727,
+      "step": 1044
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7862177934875687e-06,
+      "loss": 2.3907,
+      "step": 1046
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.785311259804518e-06,
+      "loss": 2.6242,
+      "step": 1048
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7844030392013538e-06,
+      "loss": 2.4537,
+      "step": 1050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7834931336290172e-06,
+      "loss": 2.7705,
+      "step": 1052
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.782581545042068e-06,
+      "loss": 2.393,
+      "step": 1054
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7816682753986815e-06,
+      "loss": 2.4858,
+      "step": 1056
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.780753326660644e-06,
+      "loss": 2.3062,
+      "step": 1058
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.779836700793349e-06,
+      "loss": 2.2885,
+      "step": 1060
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7789183997657919e-06,
+      "loss": 2.5484,
+      "step": 1062
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7779984255505675e-06,
+      "loss": 2.5681,
+      "step": 1064
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.777076780123864e-06,
+      "loss": 2.3677,
+      "step": 1066
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7761534654654593e-06,
+      "loss": 2.3538,
+      "step": 1068
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7752284835587182e-06,
+      "loss": 2.6437,
+      "step": 1070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.774301836390585e-06,
+      "loss": 2.6875,
+      "step": 1072
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.773373525951583e-06,
+      "loss": 2.413,
+      "step": 1074
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7724435542358078e-06,
+      "loss": 2.6612,
+      "step": 1076
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7715119232409225e-06,
+      "loss": 2.2541,
+      "step": 1078
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7705786349681562e-06,
+      "loss": 2.7833,
+      "step": 1080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7696436914222965e-06,
+      "loss": 2.692,
+      "step": 1082
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7687070946116874e-06,
+      "loss": 2.5163,
+      "step": 1084
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7677688465482244e-06,
+      "loss": 2.4503,
+      "step": 1086
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7668289492473493e-06,
+      "loss": 2.2691,
+      "step": 1088
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7658874047280476e-06,
+      "loss": 2.6636,
+      "step": 1090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.764944215012843e-06,
+      "loss": 2.2402,
+      "step": 1092
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7639993821277921e-06,
+      "loss": 3.0009,
+      "step": 1094
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.763052908102482e-06,
+      "loss": 2.3759,
+      "step": 1096
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.762104794970026e-06,
+      "loss": 2.4261,
+      "step": 1098
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7611550447670567e-06,
+      "loss": 2.3872,
+      "step": 1100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7602036595337243e-06,
+      "loss": 2.5853,
+      "step": 1102
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7592506413136906e-06,
+      "loss": 2.3782,
+      "step": 1104
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7582959921541264e-06,
+      "loss": 2.3908,
+      "step": 1106
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7573397141057042e-06,
+      "loss": 2.4113,
+      "step": 1108
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7563818092225967e-06,
+      "loss": 2.4058,
+      "step": 1110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7554222795624707e-06,
+      "loss": 2.5073,
+      "step": 1112
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.754461127186483e-06,
+      "loss": 2.4979,
+      "step": 1114
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7534983541592773e-06,
+      "loss": 2.491,
+      "step": 1116
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7525339625489772e-06,
+      "loss": 2.3299,
+      "step": 1118
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7515679544271838e-06,
+      "loss": 2.6209,
+      "step": 1120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7506003318689703e-06,
+      "loss": 2.2329,
+      "step": 1122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7496310969528785e-06,
+      "loss": 2.84,
+      "step": 1124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.748660251760913e-06,
+      "loss": 2.4634,
+      "step": 1126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7476877983785378e-06,
+      "loss": 2.2628,
+      "step": 1128
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7467137388946716e-06,
+      "loss": 2.4435,
+      "step": 1130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7457380754016823e-06,
+      "loss": 2.6127,
+      "step": 1132
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7447608099953842e-06,
+      "loss": 2.6242,
+      "step": 1134
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7437819447750327e-06,
+      "loss": 2.6045,
+      "step": 1136
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.742801481843319e-06,
+      "loss": 2.4415,
+      "step": 1138
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.741819423306367e-06,
+      "loss": 2.3979,
+      "step": 1140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.740835771273728e-06,
+      "loss": 2.6006,
+      "step": 1142
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7398505278583758e-06,
+      "loss": 2.6163,
+      "step": 1144
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7388636951767035e-06,
+      "loss": 2.396,
+      "step": 1146
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7378752753485171e-06,
+      "loss": 2.4038,
+      "step": 1148
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7368852704970328e-06,
+      "loss": 2.3577,
+      "step": 1150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.735893682748871e-06,
+      "loss": 2.8212,
+      "step": 1152
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7349005142340522e-06,
+      "loss": 2.6463,
+      "step": 1154
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7339057670859936e-06,
+      "loss": 2.5239,
+      "step": 1156
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.732909443441502e-06,
+      "loss": 2.5437,
+      "step": 1158
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7319115454407715e-06,
+      "loss": 2.5785,
+      "step": 1160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.730912075227378e-06,
+      "loss": 2.2862,
+      "step": 1162
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7299110349482747e-06,
+      "loss": 2.1492,
+      "step": 1164
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.728908426753787e-06,
+      "loss": 2.428,
+      "step": 1166
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7279042527976085e-06,
+      "loss": 2.4149,
+      "step": 1168
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7268985152367967e-06,
+      "loss": 2.5587,
+      "step": 1170
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7258912162317675e-06,
+      "loss": 2.4955,
+      "step": 1172
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7248823579462904e-06,
+      "loss": 2.4643,
+      "step": 1174
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7238719425474852e-06,
+      "loss": 2.4708,
+      "step": 1176
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7228599722058164e-06,
+      "loss": 2.3787,
+      "step": 1178
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7218464490950884e-06,
+      "loss": 2.6198,
+      "step": 1180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7208313753924408e-06,
+      "loss": 2.3774,
+      "step": 1182
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7198147532783448e-06,
+      "loss": 2.5964,
+      "step": 1184
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.718796584936597e-06,
+      "loss": 2.4021,
+      "step": 1186
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7177768725543158e-06,
+      "loss": 2.5438,
+      "step": 1188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7167556183219362e-06,
+      "loss": 2.5724,
+      "step": 1190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7157328244332057e-06,
+      "loss": 2.6797,
+      "step": 1192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7147084930851783e-06,
+      "loss": 2.4916,
+      "step": 1194
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7136826264782114e-06,
+      "loss": 2.3222,
+      "step": 1196
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7126552268159595e-06,
+      "loss": 2.1401,
+      "step": 1198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7116262963053712e-06,
+      "loss": 2.9593,
+      "step": 1200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7105958371566823e-06,
+      "loss": 2.4008,
+      "step": 1202
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.709563851583413e-06,
+      "loss": 2.6847,
+      "step": 1204
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7085303418023626e-06,
+      "loss": 2.208,
+      "step": 1206
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7074953100336042e-06,
+      "loss": 2.3636,
+      "step": 1208
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7064587585004806e-06,
+      "loss": 2.4801,
+      "step": 1210
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.705420689429598e-06,
+      "loss": 2.4984,
+      "step": 1212
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7043811050508244e-06,
+      "loss": 2.5331,
+      "step": 1214
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.703340007597281e-06,
+      "loss": 2.4208,
+      "step": 1216
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.70229739930534e-06,
+      "loss": 2.3254,
+      "step": 1218
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7012532824146195e-06,
+      "loss": 2.4997,
+      "step": 1220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.700207659167977e-06,
+      "loss": 2.3298,
+      "step": 1222
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.699160531811507e-06,
+      "loss": 2.4146,
+      "step": 1224
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.698111902594534e-06,
+      "loss": 2.3754,
+      "step": 1226
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6970617737696096e-06,
+      "loss": 2.7424,
+      "step": 1228
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.696010147592506e-06,
+      "loss": 2.6457,
+      "step": 1230
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6949570263222116e-06,
+      "loss": 2.5523,
+      "step": 1232
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6939024122209268e-06,
+      "loss": 2.424,
+      "step": 1234
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6928463075540594e-06,
+      "loss": 2.3549,
+      "step": 1236
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6917887145902178e-06,
+      "loss": 2.2031,
+      "step": 1238
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6907296356012085e-06,
+      "loss": 2.4517,
+      "step": 1240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6896690728620296e-06,
+      "loss": 2.4635,
+      "step": 1242
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6886070286508657e-06,
+      "loss": 2.5939,
+      "step": 1244
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6875435052490854e-06,
+      "loss": 2.6403,
+      "step": 1246
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6864785049412334e-06,
+      "loss": 2.3347,
+      "step": 1248
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6854120300150272e-06,
+      "loss": 2.4685,
+      "step": 1250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6843440827613527e-06,
+      "loss": 2.3795,
+      "step": 1252
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6832746654742573e-06,
+      "loss": 2.5659,
+      "step": 1254
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6822037804509471e-06,
+      "loss": 2.4002,
+      "step": 1256
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6811314299917804e-06,
+      "loss": 2.4262,
+      "step": 1258
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6800576164002635e-06,
+      "loss": 2.599,
+      "step": 1260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6789823419830463e-06,
+      "loss": 2.4544,
+      "step": 1262
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6779056090499158e-06,
+      "loss": 2.3587,
+      "step": 1264
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.676827419913793e-06,
+      "loss": 2.428,
+      "step": 1266
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6757477768907254e-06,
+      "loss": 2.4027,
+      "step": 1268
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6746666822998851e-06,
+      "loss": 2.675,
+      "step": 1270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6735841384635619e-06,
+      "loss": 2.3356,
+      "step": 1272
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6725001477071582e-06,
+      "loss": 2.573,
+      "step": 1274
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6714147123591854e-06,
+      "loss": 2.2519,
+      "step": 1276
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.670327834751257e-06,
+      "loss": 2.7857,
+      "step": 1278
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6692395172180849e-06,
+      "loss": 2.6319,
+      "step": 1280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.668149762097475e-06,
+      "loss": 2.6005,
+      "step": 1282
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6670585717303201e-06,
+      "loss": 2.4265,
+      "step": 1284
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6659659484605966e-06,
+      "loss": 2.3283,
+      "step": 1286
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6648718946353584e-06,
+      "loss": 2.684,
+      "step": 1288
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6637764126047335e-06,
+      "loss": 2.4864,
+      "step": 1290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6626795047219168e-06,
+      "loss": 2.5547,
+      "step": 1292
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.661581173343166e-06,
+      "loss": 2.6756,
+      "step": 1294
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6604814208277972e-06,
+      "loss": 2.5402,
+      "step": 1296
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.659380249538179e-06,
+      "loss": 2.602,
+      "step": 1298
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6582776618397277e-06,
+      "loss": 2.4283,
+      "step": 1300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6571736601009021e-06,
+      "loss": 2.4714,
+      "step": 1302
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6560682466931981e-06,
+      "loss": 2.4282,
+      "step": 1304
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.654961423991145e-06,
+      "loss": 2.4965,
+      "step": 1306
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.653853194372298e-06,
+      "loss": 2.4844,
+      "step": 1308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6527435602172361e-06,
+      "loss": 2.5985,
+      "step": 1310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.651632523909554e-06,
+      "loss": 2.2583,
+      "step": 1312
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.650520087835859e-06,
+      "loss": 2.6012,
+      "step": 1314
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6494062543857651e-06,
+      "loss": 2.5839,
+      "step": 1316
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6482910259518881e-06,
+      "loss": 2.719,
+      "step": 1318
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.64717440492984e-06,
+      "loss": 2.7217,
+      "step": 1320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6460563937182248e-06,
+      "loss": 2.4131,
+      "step": 1322
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6449369947186321e-06,
+      "loss": 2.4515,
+      "step": 1324
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6438162103356332e-06,
+      "loss": 2.412,
+      "step": 1326
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.642694042976775e-06,
+      "loss": 2.2864,
+      "step": 1328
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.641570495052575e-06,
+      "loss": 2.315,
+      "step": 1330
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.6404455689765165e-06,
+      "loss": 2.4863,
+      "step": 1332
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.6393192671650433e-06,
+      "loss": 2.7534,
+      "step": 1334
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.6381915920375539e-06,
+      "loss": 2.4737,
+      "step": 1336
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.637062546016398e-06,
+      "loss": 2.3454,
+      "step": 1338
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.6359321315268683e-06,
+      "loss": 2.5099,
+      "step": 1340
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.634800350997199e-06,
+      "loss": 2.3486,
+      "step": 1342
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.633667206858557e-06,
+      "loss": 2.2076,
+      "step": 1344
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.6325327015450396e-06,
+      "loss": 2.2752,
+      "step": 1346
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.6313968374936673e-06,
+      "loss": 2.3892,
+      "step": 1348
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.6302596171443795e-06,
+      "loss": 2.3764,
+      "step": 1350
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.6291210429400296e-06,
+      "loss": 2.2785,
+      "step": 1352
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.627981117326378e-06,
+      "loss": 2.5743,
+      "step": 1354
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.6268398427520894e-06,
+      "loss": 2.7498,
+      "step": 1356
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.6256972216687248e-06,
+      "loss": 2.1805,
+      "step": 1358
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.624553256530739e-06,
+      "loss": 2.5828,
+      "step": 1360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6234079497954731e-06,
+      "loss": 2.6088,
+      "step": 1362
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6222613039231501e-06,
+      "loss": 2.4188,
+      "step": 1364
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.62111332137687e-06,
+      "loss": 2.2557,
+      "step": 1366
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6199640046226035e-06,
+      "loss": 2.6659,
+      "step": 1368
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6188133561291884e-06,
+      "loss": 2.0596,
+      "step": 1370
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6176613783683218e-06,
+      "loss": 2.5042,
+      "step": 1372
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.616508073814557e-06,
+      "loss": 2.3132,
+      "step": 1374
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6153534449452972e-06,
+      "loss": 2.413,
+      "step": 1376
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6141974942407907e-06,
+      "loss": 2.596,
+      "step": 1378
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6130402241841247e-06,
+      "loss": 2.3572,
+      "step": 1380
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6118816372612207e-06,
+      "loss": 2.3489,
+      "step": 1382
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6107217359608287e-06,
+      "loss": 2.5228,
+      "step": 1384
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6095605227745229e-06,
+      "loss": 2.3761,
+      "step": 1386
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.608398000196694e-06,
+      "loss": 2.5952,
+      "step": 1388
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6072341707245474e-06,
+      "loss": 2.1189,
+      "step": 1390
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.6060690368580944e-06,
+      "loss": 2.2626,
+      "step": 1392
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.6049026011001488e-06,
+      "loss": 2.5117,
+      "step": 1394
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.6037348659563205e-06,
+      "loss": 2.3999,
+      "step": 1396
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.6025658339350112e-06,
+      "loss": 2.2231,
+      "step": 1398
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.6013955075474083e-06,
+      "loss": 2.3908,
+      "step": 1400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.6002238893074794e-06,
+      "loss": 2.1988,
+      "step": 1402
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.599050981731967e-06,
+      "loss": 2.3146,
+      "step": 1404
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.597876787340383e-06,
+      "loss": 2.6744,
+      "step": 1406
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.596701308655005e-06,
+      "loss": 2.405,
+      "step": 1408
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5955245482008674e-06,
+      "loss": 2.3743,
+      "step": 1410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5943465085057594e-06,
+      "loss": 2.6347,
+      "step": 1412
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5931671921002172e-06,
+      "loss": 2.4252,
+      "step": 1414
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5919866015175197e-06,
+      "loss": 2.6307,
+      "step": 1416
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5908047392936835e-06,
+      "loss": 2.1272,
+      "step": 1418
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.589621607967456e-06,
+      "loss": 2.6618,
+      "step": 1420
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5884372100803112e-06,
+      "loss": 2.6405,
+      "step": 1422
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5872515481764436e-06,
+      "loss": 2.3396,
+      "step": 1424
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5860646248027623e-06,
+      "loss": 2.7622,
+      "step": 1426
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5848764425088878e-06,
+      "loss": 2.3926,
+      "step": 1428
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5836870038471436e-06,
+      "loss": 2.4872,
+      "step": 1430
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.582496311372552e-06,
+      "loss": 2.3206,
+      "step": 1432
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5813043676428293e-06,
+      "loss": 2.3901,
+      "step": 1434
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5801111752183788e-06,
+      "loss": 2.4461,
+      "step": 1436
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5789167366622866e-06,
+      "loss": 2.5871,
+      "step": 1438
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5777210545403154e-06,
+      "loss": 2.6168,
+      "step": 1440
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5765241314208998e-06,
+      "loss": 2.4094,
+      "step": 1442
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.575325969875139e-06,
+      "loss": 2.3476,
+      "step": 1444
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5741265724767941e-06,
+      "loss": 2.2706,
+      "step": 1446
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5729259418022789e-06,
+      "loss": 2.2549,
+      "step": 1448
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5717240804306578e-06,
+      "loss": 2.3836,
+      "step": 1450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5705209909436387e-06,
+      "loss": 2.273,
+      "step": 1452
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.569316675925567e-06,
+      "loss": 2.2781,
+      "step": 1454
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.568111137963422e-06,
+      "loss": 2.4964,
+      "step": 1456
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5669043796468078e-06,
+      "loss": 2.452,
+      "step": 1458
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5656964035679518e-06,
+      "loss": 2.2103,
+      "step": 1460
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5644872123216968e-06,
+      "loss": 2.3263,
+      "step": 1462
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5632768085054956e-06,
+      "loss": 2.3993,
+      "step": 1464
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5620651947194054e-06,
+      "loss": 2.3741,
+      "step": 1466
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5608523735660834e-06,
+      "loss": 2.7935,
+      "step": 1468
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.55963834765078e-06,
+      "loss": 2.5123,
+      "step": 1470
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5584231195813332e-06,
+      "loss": 2.5983,
+      "step": 1472
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5572066919681634e-06,
+      "loss": 2.4809,
+      "step": 1474
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.5559890674242687e-06,
+      "loss": 2.3429,
+      "step": 1476
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.5547702485652164e-06,
+      "loss": 2.5719,
+      "step": 1478
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5535502380091411e-06,
+      "loss": 2.4216,
+      "step": 1480
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5523290383767366e-06,
+      "loss": 2.6257,
+      "step": 1482
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.551106652291251e-06,
+      "loss": 2.5949,
+      "step": 1484
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5498830823784808e-06,
+      "loss": 2.5236,
+      "step": 1486
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5486583312667652e-06,
+      "loss": 2.2841,
+      "step": 1488
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5474324015869819e-06,
+      "loss": 2.6608,
+      "step": 1490
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5462052959725388e-06,
+      "loss": 2.6327,
+      "step": 1492
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5449770170593715e-06,
+      "loss": 2.3065,
+      "step": 1494
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5437475674859335e-06,
+      "loss": 2.4162,
+      "step": 1496
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.542516949893196e-06,
+      "loss": 2.5158,
+      "step": 1498
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.5412851669246368e-06,
+      "loss": 2.4369,
+      "step": 1500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.540052221226238e-06,
+      "loss": 2.6495,
+      "step": 1502
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5388181154464795e-06,
+      "loss": 2.3576,
+      "step": 1504
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5375828522363325e-06,
+      "loss": 2.6325,
+      "step": 1506
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5363464342492552e-06,
+      "loss": 2.4328,
+      "step": 1508
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5351088641411863e-06,
+      "loss": 2.5354,
+      "step": 1510
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5338701445705385e-06,
+      "loss": 2.7207,
+      "step": 1512
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5326302781981948e-06,
+      "loss": 2.3158,
+      "step": 1514
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5313892676875008e-06,
+      "loss": 2.6291,
+      "step": 1516
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5301471157042603e-06,
+      "loss": 2.4181,
+      "step": 1518
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5289038249167285e-06,
+      "loss": 2.4438,
+      "step": 1520
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5276593979956078e-06,
+      "loss": 2.4456,
+      "step": 1522
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5264138376140405e-06,
+      "loss": 2.3465,
+      "step": 1524
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5251671464476034e-06,
+      "loss": 2.6112,
+      "step": 1526
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5239193271743025e-06,
+      "loss": 2.7099,
+      "step": 1528
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5226703824745682e-06,
+      "loss": 2.5169,
+      "step": 1530
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5214203150312463e-06,
+      "loss": 2.6179,
+      "step": 1532
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5201691275295963e-06,
+      "loss": 2.6206,
+      "step": 1534
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5189168226572825e-06,
+      "loss": 2.4832,
+      "step": 1536
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5176634031043706e-06,
+      "loss": 2.4303,
+      "step": 1538
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5164088715633185e-06,
+      "loss": 2.3131,
+      "step": 1540
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.515153230728976e-06,
+      "loss": 2.4736,
+      "step": 1542
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5138964832985724e-06,
+      "loss": 2.3829,
+      "step": 1544
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5126386319717166e-06,
+      "loss": 2.3293,
+      "step": 1546
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5113796794503875e-06,
+      "loss": 2.5245,
+      "step": 1548
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5101196284389297e-06,
+      "loss": 2.6711,
+      "step": 1550
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5088584816440482e-06,
+      "loss": 2.2194,
+      "step": 1552
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5075962417748e-06,
+      "loss": 2.6407,
+      "step": 1554
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.5063329115425923e-06,
+      "loss": 2.5811,
+      "step": 1556
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.505068493661173e-06,
+      "loss": 2.533,
+      "step": 1558
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.503802990846627e-06,
+      "loss": 2.3742,
+      "step": 1560
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.50253640581737e-06,
+      "loss": 2.4339,
+      "step": 1562
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.5012687412941412e-06,
+      "loss": 2.1867,
+      "step": 1564
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.5e-06,
+      "loss": 2.3181,
+      "step": 1566
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4987301846603183e-06,
+      "loss": 2.5327,
+      "step": 1568
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4974592980027749e-06,
+      "loss": 2.4644,
+      "step": 1570
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.49618734275735e-06,
+      "loss": 2.3521,
+      "step": 1572
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4949143216563195e-06,
+      "loss": 2.323,
+      "step": 1574
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.493640237434249e-06,
+      "loss": 2.4037,
+      "step": 1576
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4923650928279867e-06,
+      "loss": 2.2227,
+      "step": 1578
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4910888905766602e-06,
+      "loss": 2.5474,
+      "step": 1580
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4898116334216673e-06,
+      "loss": 2.165,
+      "step": 1582
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4885333241066734e-06,
+      "loss": 2.5061,
+      "step": 1584
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4872539653776028e-06,
+      "loss": 2.3724,
+      "step": 1586
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4859735599826352e-06,
+      "loss": 2.6099,
+      "step": 1588
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4846921106721974e-06,
+      "loss": 2.5928,
+      "step": 1590
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4834096201989596e-06,
+      "loss": 2.2369,
+      "step": 1592
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4821260913178282e-06,
+      "loss": 2.3638,
+      "step": 1594
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4808415267859398e-06,
+      "loss": 2.3177,
+      "step": 1596
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4795559293626562e-06,
+      "loss": 2.2792,
+      "step": 1598
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4782693018095577e-06,
+      "loss": 2.3882,
+      "step": 1600
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4769816468904378e-06,
+      "loss": 2.3006,
+      "step": 1602
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4756929673712962e-06,
+      "loss": 2.2557,
+      "step": 1604
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4744032660203339e-06,
+      "loss": 2.5488,
+      "step": 1606
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4731125456079467e-06,
+      "loss": 2.3487,
+      "step": 1608
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.47182080890672e-06,
+      "loss": 2.641,
+      "step": 1610
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4705280586914218e-06,
+      "loss": 2.1686,
+      "step": 1612
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4692342977389976e-06,
+      "loss": 2.3053,
+      "step": 1614
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4679395288285627e-06,
+      "loss": 2.247,
+      "step": 1616
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4666437547413999e-06,
+      "loss": 2.6372,
+      "step": 1618
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.465346978260949e-06,
+      "loss": 2.6255,
+      "step": 1620
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4640492021728043e-06,
+      "loss": 2.1904,
+      "step": 1622
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4627504292647074e-06,
+      "loss": 2.5925,
+      "step": 1624
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4614506623265406e-06,
+      "loss": 2.6503,
+      "step": 1626
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4601499041503216e-06,
+      "loss": 2.3066,
+      "step": 1628
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.458848157530197e-06,
+      "loss": 2.4649,
+      "step": 1630
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4575454252624383e-06,
+      "loss": 2.2839,
+      "step": 1632
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4562417101454316e-06,
+      "loss": 2.5123,
+      "step": 1634
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4549370149796769e-06,
+      "loss": 2.3119,
+      "step": 1636
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4536313425677775e-06,
+      "loss": 2.4356,
+      "step": 1638
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.452324695714437e-06,
+      "loss": 2.5462,
+      "step": 1640
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.451017077226452e-06,
+      "loss": 2.4099,
+      "step": 1642
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4497084899127056e-06,
+      "loss": 2.4389,
+      "step": 1644
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4483989365841632e-06,
+      "loss": 2.7143,
+      "step": 1646
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.447088420053864e-06,
+      "loss": 2.5341,
+      "step": 1648
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4457769431369173e-06,
+      "loss": 2.6033,
+      "step": 1650
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4444645086504947e-06,
+      "loss": 2.2592,
+      "step": 1652
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.443151119413825e-06,
+      "loss": 2.3131,
+      "step": 1654
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4418367782481882e-06,
+      "loss": 2.5731,
+      "step": 1656
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4405214879769084e-06,
+      "loss": 2.6759,
+      "step": 1658
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4392052514253497e-06,
+      "loss": 2.5662,
+      "step": 1660
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.437888071420907e-06,
+      "loss": 2.365,
+      "step": 1662
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4365699507930039e-06,
+      "loss": 2.3041,
+      "step": 1664
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4352508923730829e-06,
+      "loss": 2.2466,
+      "step": 1666
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.433930898994602e-06,
+      "loss": 2.6212,
+      "step": 1668
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.432609973493027e-06,
+      "loss": 2.3997,
+      "step": 1670
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.431288118705826e-06,
+      "loss": 2.4061,
+      "step": 1672
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4299653374724642e-06,
+      "loss": 2.6446,
+      "step": 1674
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4286416326343957e-06,
+      "loss": 2.3368,
+      "step": 1676
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4273170070350587e-06,
+      "loss": 2.2844,
+      "step": 1678
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4259914635198701e-06,
+      "loss": 2.3913,
+      "step": 1680
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4246650049362185e-06,
+      "loss": 2.271,
+      "step": 1682
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4233376341334567e-06,
+      "loss": 2.2982,
+      "step": 1684
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.422009353962899e-06,
+      "loss": 2.1357,
+      "step": 1686
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4206801672778117e-06,
+      "loss": 2.1893,
+      "step": 1688
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4193500769334088e-06,
+      "loss": 2.5499,
+      "step": 1690
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4180190857868459e-06,
+      "loss": 2.4208,
+      "step": 1692
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4166871966972132e-06,
+      "loss": 2.5682,
+      "step": 1694
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4153544125255291e-06,
+      "loss": 2.5289,
+      "step": 1696
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4140207361347364e-06,
+      "loss": 2.622,
+      "step": 1698
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4126861703896926e-06,
+      "loss": 2.3856,
+      "step": 1700
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4113507181571669e-06,
+      "loss": 2.3768,
+      "step": 1702
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4100143823058324e-06,
+      "loss": 2.3536,
+      "step": 1704
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4086771657062597e-06,
+      "loss": 2.4242,
+      "step": 1706
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4073390712309126e-06,
+      "loss": 2.5667,
+      "step": 1708
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4060001017541395e-06,
+      "loss": 2.2116,
+      "step": 1710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4046602601521688e-06,
+      "loss": 2.3444,
+      "step": 1712
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4033195493031021e-06,
+      "loss": 2.4935,
+      "step": 1714
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4019779720869088e-06,
+      "loss": 2.4833,
+      "step": 1716
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4006355313854189e-06,
+      "loss": 2.7645,
+      "step": 1718
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3992922300823166e-06,
+      "loss": 2.5373,
+      "step": 1720
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3979480710631366e-06,
+      "loss": 2.3545,
+      "step": 1722
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3966030572152542e-06,
+      "loss": 2.5239,
+      "step": 1724
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.395257191427882e-06,
+      "loss": 2.5872,
+      "step": 1726
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3939104765920616e-06,
+      "loss": 2.3978,
+      "step": 1728
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.39256291560066e-06,
+      "loss": 2.5142,
+      "step": 1730
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3912145113483606e-06,
+      "loss": 2.7417,
+      "step": 1732
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3898652667316582e-06,
+      "loss": 2.3602,
+      "step": 1734
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3885151846488536e-06,
+      "loss": 2.3813,
+      "step": 1736
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.387164268000046e-06,
+      "loss": 2.5534,
+      "step": 1738
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.385812519687127e-06,
+      "loss": 2.5258,
+      "step": 1740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3844599426137755e-06,
+      "loss": 2.6035,
+      "step": 1742
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3831065396854507e-06,
+      "loss": 2.5751,
+      "step": 1744
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3817523138093842e-06,
+      "loss": 2.5558,
+      "step": 1746
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3803972678945774e-06,
+      "loss": 2.0873,
+      "step": 1748
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.379041404851792e-06,
+      "loss": 2.1898,
+      "step": 1750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3776847275935456e-06,
+      "loss": 2.5692,
+      "step": 1752
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3763272390341042e-06,
+      "loss": 2.4294,
+      "step": 1754
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.374968942089477e-06,
+      "loss": 2.429,
+      "step": 1756
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3736098396774094e-06,
+      "loss": 2.4356,
+      "step": 1758
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3722499347173772e-06,
+      "loss": 2.2805,
+      "step": 1760
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3708892301305807e-06,
+      "loss": 2.4228,
+      "step": 1762
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3695277288399365e-06,
+      "loss": 2.2363,
+      "step": 1764
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3681654337700736e-06,
+      "loss": 2.426,
+      "step": 1766
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3668023478473264e-06,
+      "loss": 2.7234,
+      "step": 1768
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.365438473999727e-06,
+      "loss": 2.5752,
+      "step": 1770
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3640738151570005e-06,
+      "loss": 2.4224,
+      "step": 1772
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.362708374250559e-06,
+      "loss": 2.4265,
+      "step": 1774
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3613421542134936e-06,
+      "loss": 2.4323,
+      "step": 1776
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3599751579805697e-06,
+      "loss": 2.2774,
+      "step": 1778
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.358607388488219e-06,
+      "loss": 2.5681,
+      "step": 1780
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.357238848674536e-06,
+      "loss": 2.2788,
+      "step": 1782
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3558695414792677e-06,
+      "loss": 2.5053,
+      "step": 1784
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3544994698438115e-06,
+      "loss": 2.4669,
+      "step": 1786
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3531286367112058e-06,
+      "loss": 2.2683,
+      "step": 1788
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.351757045026125e-06,
+      "loss": 2.2351,
+      "step": 1790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.350384697734873e-06,
+      "loss": 2.3426,
+      "step": 1792
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3490115977853772e-06,
+      "loss": 2.3011,
+      "step": 1794
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.34763774812718e-06,
+      "loss": 2.3215,
+      "step": 1796
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3462631517114374e-06,
+      "loss": 2.4746,
+      "step": 1798
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3448878114909064e-06,
+      "loss": 2.4723,
+      "step": 1800
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3435117304199434e-06,
+      "loss": 2.5015,
+      "step": 1802
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3421349114544962e-06,
+      "loss": 2.3998,
+      "step": 1804
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3407573575520973e-06,
+      "loss": 2.2199,
+      "step": 1806
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3393790716718577e-06,
+      "loss": 2.1974,
+      "step": 1808
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3380000567744608e-06,
+      "loss": 2.4168,
+      "step": 1810
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3366203158221566e-06,
+      "loss": 2.4152,
+      "step": 1812
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.335239851778754e-06,
+      "loss": 2.4792,
+      "step": 1814
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.3338586676096157e-06,
+      "loss": 2.267,
+      "step": 1816
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.3324767662816513e-06,
+      "loss": 2.2245,
+      "step": 1818
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3310941507633107e-06,
+      "loss": 2.1719,
+      "step": 1820
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3297108240245775e-06,
+      "loss": 2.5786,
+      "step": 1822
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3283267890369638e-06,
+      "loss": 2.4891,
+      "step": 1824
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.326942048773503e-06,
+      "loss": 2.232,
+      "step": 1826
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3255566062087428e-06,
+      "loss": 2.2262,
+      "step": 1828
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3241704643187408e-06,
+      "loss": 2.1527,
+      "step": 1830
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3227836260810555e-06,
+      "loss": 2.5635,
+      "step": 1832
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3213960944747416e-06,
+      "loss": 2.6371,
+      "step": 1834
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3200078724803437e-06,
+      "loss": 2.1861,
+      "step": 1836
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3186189630798892e-06,
+      "loss": 2.2795,
+      "step": 1838
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3172293692568816e-06,
+      "loss": 2.363,
+      "step": 1840
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3158390939962952e-06,
+      "loss": 2.8042,
+      "step": 1842
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3144481402845677e-06,
+      "loss": 2.4579,
+      "step": 1844
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3130565111095947e-06,
+      "loss": 2.486,
+      "step": 1846
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3116642094607221e-06,
+      "loss": 2.3256,
+      "step": 1848
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3102712383287411e-06,
+      "loss": 2.5683,
+      "step": 1850
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.30887760070588e-06,
+      "loss": 2.3407,
+      "step": 1852
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3074832995857996e-06,
+      "loss": 2.2962,
+      "step": 1854
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3060883379635855e-06,
+      "loss": 2.7058,
+      "step": 1856
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.304692718835743e-06,
+      "loss": 2.3485,
+      "step": 1858
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3032964452001886e-06,
+      "loss": 2.6635,
+      "step": 1860
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.301899520056245e-06,
+      "loss": 2.1123,
+      "step": 1862
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3005019464046352e-06,
+      "loss": 2.4141,
+      "step": 1864
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2991037272474743e-06,
+      "loss": 2.5197,
+      "step": 1866
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2977048655882644e-06,
+      "loss": 2.2944,
+      "step": 1868
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.296305364431888e-06,
+      "loss": 2.6415,
+      "step": 1870
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.294905226784601e-06,
+      "loss": 2.3808,
+      "step": 1872
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2935044556540265e-06,
+      "loss": 2.4877,
+      "step": 1874
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2921030540491482e-06,
+      "loss": 2.6193,
+      "step": 1876
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2907010249803052e-06,
+      "loss": 2.3594,
+      "step": 1878
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2892983714591825e-06,
+      "loss": 2.17,
+      "step": 1880
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2878950964988086e-06,
+      "loss": 2.4788,
+      "step": 1882
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2864912031135455e-06,
+      "loss": 2.4027,
+      "step": 1884
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2850866943190843e-06,
+      "loss": 2.4353,
+      "step": 1886
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2836815731324375e-06,
+      "loss": 2.6185,
+      "step": 1888
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2822758425719336e-06,
+      "loss": 2.2183,
+      "step": 1890
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.2808695056572098e-06,
+      "loss": 2.3426,
+      "step": 1892
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.2794625654092057e-06,
+      "loss": 2.2975,
+      "step": 1894
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2780550248501577e-06,
+      "loss": 2.5333,
+      "step": 1896
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2766468870035905e-06,
+      "loss": 2.3948,
+      "step": 1898
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.2752381548943122e-06,
+      "loss": 2.161,
+      "step": 1900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.2738288315484088e-06,
+      "loss": 2.7284,
+      "step": 1902
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.272418919993234e-06,
+      "loss": 2.2664,
+      "step": 1904
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.2710084232574073e-06,
+      "loss": 2.4391,
+      "step": 1906
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.2695973443708035e-06,
+      "loss": 2.5834,
+      "step": 1908
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.2681856863645485e-06,
+      "loss": 2.4076,
+      "step": 1910
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.2667734522710128e-06,
+      "loss": 2.4514,
+      "step": 1912
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.2653606451238035e-06,
+      "loss": 2.2832,
+      "step": 1914
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.263947267957759e-06,
+      "loss": 2.4629,
+      "step": 1916
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2625333238089421e-06,
+      "loss": 2.5803,
+      "step": 1918
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2611188157146341e-06,
+      "loss": 2.2946,
+      "step": 1920
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2597037467133267e-06,
+      "loss": 2.4157,
+      "step": 1922
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2582881198447172e-06,
+      "loss": 2.4456,
+      "step": 1924
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2568719381497013e-06,
+      "loss": 1.9894,
+      "step": 1926
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2554552046703658e-06,
+      "loss": 2.2723,
+      "step": 1928
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2540379224499838e-06,
+      "loss": 2.5817,
+      "step": 1930
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2526200945330056e-06,
+      "loss": 2.2494,
+      "step": 1932
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.251201723965056e-06,
+      "loss": 2.4479,
+      "step": 1934
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2497828137929233e-06,
+      "loss": 2.1995,
+      "step": 1936
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2483633670645562e-06,
+      "loss": 2.2064,
+      "step": 1938
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2469433868290552e-06,
+      "loss": 2.1307,
+      "step": 1940
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2455228761366674e-06,
+      "loss": 2.3635,
+      "step": 1942
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.244101838038779e-06,
+      "loss": 2.2071,
+      "step": 1944
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2426802755879096e-06,
+      "loss": 2.4216,
+      "step": 1946
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.2412581918377045e-06,
+      "loss": 2.4252,
+      "step": 1948
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.2398355898429295e-06,
+      "loss": 2.7158,
+      "step": 1950
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2384124726594626e-06,
+      "loss": 2.1449,
+      "step": 1952
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.23698884334429e-06,
+      "loss": 2.2603,
+      "step": 1954
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.235564704955496e-06,
+      "loss": 2.5274,
+      "step": 1956
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2341400605522604e-06,
+      "loss": 2.4358,
+      "step": 1958
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2327149131948492e-06,
+      "loss": 2.3367,
+      "step": 1960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2312892659446083e-06,
+      "loss": 2.2516,
+      "step": 1962
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2298631218639584e-06,
+      "loss": 2.5467,
+      "step": 1964
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.228436484016386e-06,
+      "loss": 2.6495,
+      "step": 1966
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.22700935546644e-06,
+      "loss": 2.3837,
+      "step": 1968
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.225581739279722e-06,
+      "loss": 2.4825,
+      "step": 1970
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.224153638522882e-06,
+      "loss": 2.3524,
+      "step": 1972
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2227250562636098e-06,
+      "loss": 2.6791,
+      "step": 1974
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2212959955706309e-06,
+      "loss": 2.3547,
+      "step": 1976
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2198664595136977e-06,
+      "loss": 2.4261,
+      "step": 1978
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2184364511635835e-06,
+      "loss": 2.2508,
+      "step": 1980
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2170059735920768e-06,
+      "loss": 2.3813,
+      "step": 1982
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2155750298719736e-06,
+      "loss": 2.4343,
+      "step": 1984
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2141436230770716e-06,
+      "loss": 2.2746,
+      "step": 1986
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2127117562821627e-06,
+      "loss": 2.343,
+      "step": 1988
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2112794325630273e-06,
+      "loss": 2.436,
+      "step": 1990
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2098466549964268e-06,
+      "loss": 2.3389,
+      "step": 1992
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2084134266600986e-06,
+      "loss": 2.7032,
+      "step": 1994
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2069797506327474e-06,
+      "loss": 2.7591,
+      "step": 1996
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2055456299940397e-06,
+      "loss": 2.3661,
+      "step": 1998
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2041110678245974e-06,
+      "loss": 2.2633,
+      "step": 2000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2026760672059908e-06,
+      "loss": 2.5338,
+      "step": 2002
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2012406312207316e-06,
+      "loss": 2.5867,
+      "step": 2004
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.199804762952268e-06,
+      "loss": 2.4737,
+      "step": 2006
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1983684654849741e-06,
+      "loss": 2.3638,
+      "step": 2008
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.1969317419041494e-06,
+      "loss": 2.2326,
+      "step": 2010
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.1954945952960058e-06,
+      "loss": 2.3512,
+      "step": 2012
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1940570287476661e-06,
+      "loss": 2.2958,
+      "step": 2014
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1926190453471537e-06,
+      "loss": 2.1707,
+      "step": 2016
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1911806481833875e-06,
+      "loss": 2.2141,
+      "step": 2018
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1897418403461762e-06,
+      "loss": 2.3797,
+      "step": 2020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1883026249262099e-06,
+      "loss": 2.4088,
+      "step": 2022
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1868630050150543e-06,
+      "loss": 2.3545,
+      "step": 2024
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1854229837051443e-06,
+      "loss": 2.4329,
+      "step": 2026
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1839825640897767e-06,
+      "loss": 2.3957,
+      "step": 2028
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1825417492631041e-06,
+      "loss": 2.2269,
+      "step": 2030
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1811005423201281e-06,
+      "loss": 2.3754,
+      "step": 2032
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1796589463566922e-06,
+      "loss": 2.5927,
+      "step": 2034
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1782169644694758e-06,
+      "loss": 2.5426,
+      "step": 2036
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1767745997559877e-06,
+      "loss": 2.5174,
+      "step": 2038
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1753318553145586e-06,
+      "loss": 2.3862,
+      "step": 2040
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1738887342443344e-06,
+      "loss": 2.2668,
+      "step": 2042
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1724452396452713e-06,
+      "loss": 2.1973,
+      "step": 2044
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.171001374618127e-06,
+      "loss": 2.6375,
+      "step": 2046
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.1695571422644549e-06,
+      "loss": 2.5391,
+      "step": 2048
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1681125456865975e-06,
+      "loss": 2.7426,
+      "step": 2050
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1666675879876805e-06,
+      "loss": 2.4048,
+      "step": 2052
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1652222722716039e-06,
+      "loss": 2.3801,
+      "step": 2054
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1637766016430379e-06,
+      "loss": 2.3919,
+      "step": 2056
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1623305792074146e-06,
+      "loss": 2.5378,
+      "step": 2058
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1608842080709217e-06,
+      "loss": 2.3905,
+      "step": 2060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1594374913404965e-06,
+      "loss": 2.2424,
+      "step": 2062
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1579904321238182e-06,
+      "loss": 2.4612,
+      "step": 2064
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1565430335293016e-06,
+      "loss": 2.238,
+      "step": 2066
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1550952986660909e-06,
+      "loss": 2.8252,
+      "step": 2068
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1536472306440526e-06,
+      "loss": 2.5775,
+      "step": 2070
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.1521988325737684e-06,
+      "loss": 2.4405,
+      "step": 2072
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.150750107566529e-06,
+      "loss": 2.5779,
+      "step": 2074
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1493010587343284e-06,
+      "loss": 2.303,
+      "step": 2076
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1478516891898548e-06,
+      "loss": 2.382,
+      "step": 2078
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1464020020464864e-06,
+      "loss": 2.4109,
+      "step": 2080
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1449520004182833e-06,
+      "loss": 2.4196,
+      "step": 2082
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1435016874199806e-06,
+      "loss": 2.3244,
+      "step": 2084
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1420510661669834e-06,
+      "loss": 2.5645,
+      "step": 2086
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1406001397753576e-06,
+      "loss": 2.1936,
+      "step": 2088
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.1391489113618255e-06,
+      "loss": 2.4046,
+      "step": 2090
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.1376973840437581e-06,
+      "loss": 2.3622,
+      "step": 2092
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.1362455609391678e-06,
+      "loss": 2.2028,
+      "step": 2094
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.1347934451667037e-06,
+      "loss": 2.1606,
+      "step": 2096
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.1333410398456418e-06,
+      "loss": 2.3497,
+      "step": 2098
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.1318883480958816e-06,
+      "loss": 2.3943,
+      "step": 2100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.130435373037937e-06,
+      "loss": 2.3429,
+      "step": 2102
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.128982117792931e-06,
+      "loss": 2.1936,
+      "step": 2104
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1275285854825882e-06,
+      "loss": 2.2376,
+      "step": 2106
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1260747792292283e-06,
+      "loss": 2.5502,
+      "step": 2108
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1246207021557595e-06,
+      "loss": 2.3205,
+      "step": 2110
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.1231663573856716e-06,
+      "loss": 2.3361,
+      "step": 2112
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.1217117480430302e-06,
+      "loss": 2.3491,
+      "step": 2114
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.1202568772524684e-06,
+      "loss": 2.3551,
+      "step": 2116
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.118801748139181e-06,
+      "loss": 2.4922,
+      "step": 2118
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.117346363828918e-06,
+      "loss": 2.3158,
+      "step": 2120
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1158907274479772e-06,
+      "loss": 2.4769,
+      "step": 2122
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1144348421231987e-06,
+      "loss": 2.4177,
+      "step": 2124
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1129787109819562e-06,
+      "loss": 2.2472,
+      "step": 2126
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1115223371521524e-06,
+      "loss": 2.3932,
+      "step": 2128
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.1100657237622109e-06,
+      "loss": 2.4636,
+      "step": 2130
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.1086088739410697e-06,
+      "loss": 2.1997,
+      "step": 2132
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.1071517908181752e-06,
+      "loss": 2.6186,
+      "step": 2134
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.105694477523474e-06,
+      "loss": 2.3176,
+      "step": 2136
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.1042369371874088e-06,
+      "loss": 2.4185,
+      "step": 2138
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.1027791729409084e-06,
+      "loss": 2.6309,
+      "step": 2140
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.1013211879153832e-06,
+      "loss": 2.4281,
+      "step": 2142
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0998629852427179e-06,
+      "loss": 2.0809,
+      "step": 2144
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0984045680552647e-06,
+      "loss": 2.3214,
+      "step": 2146
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0969459394858364e-06,
+      "loss": 2.3554,
+      "step": 2148
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0954871026677001e-06,
+      "loss": 2.3334,
+      "step": 2150
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0940280607345704e-06,
+      "loss": 2.6128,
+      "step": 2152
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0925688168206018e-06,
+      "loss": 2.3866,
+      "step": 2154
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0911093740603836e-06,
+      "loss": 2.2807,
+      "step": 2156
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0896497355889316e-06,
+      "loss": 2.2577,
+      "step": 2158
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0881899045416818e-06,
+      "loss": 2.3046,
+      "step": 2160
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0867298840544849e-06,
+      "loss": 2.5557,
+      "step": 2162
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.085269677263597e-06,
+      "loss": 2.4749,
+      "step": 2164
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0838092873056768e-06,
+      "loss": 2.4613,
+      "step": 2166
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0823487173177735e-06,
+      "loss": 2.4206,
+      "step": 2168
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0808879704373249e-06,
+      "loss": 2.5469,
+      "step": 2170
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0794270498021486e-06,
+      "loss": 2.7428,
+      "step": 2172
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.077965958550435e-06,
+      "loss": 2.1805,
+      "step": 2174
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.076504699820741e-06,
+      "loss": 2.4371,
+      "step": 2176
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0750432767519837e-06,
+      "loss": 2.5389,
+      "step": 2178
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.073581692483433e-06,
+      "loss": 2.3166,
+      "step": 2180
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.072119950154705e-06,
+      "loss": 2.4784,
+      "step": 2182
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0706580529057553e-06,
+      "loss": 2.2944,
+      "step": 2184
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.069196003876872e-06,
+      "loss": 2.4812,
+      "step": 2186
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.06773380620867e-06,
+      "loss": 2.2998,
+      "step": 2188
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0662714630420833e-06,
+      "loss": 2.1607,
+      "step": 2190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0648089775183575e-06,
+      "loss": 2.5073,
+      "step": 2192
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0633463527790457e-06,
+      "loss": 2.7015,
+      "step": 2194
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0618835919659981e-06,
+      "loss": 2.2588,
+      "step": 2196
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0604206982213585e-06,
+      "loss": 2.5762,
+      "step": 2198
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0589576746875556e-06,
+      "loss": 2.3897,
+      "step": 2200
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0574945245072979e-06,
+      "loss": 2.4374,
+      "step": 2202
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0560312508235648e-06,
+      "loss": 2.1639,
+      "step": 2204
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0545678567796016e-06,
+      "loss": 2.4232,
+      "step": 2206
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.053104345518912e-06,
+      "loss": 2.1925,
+      "step": 2208
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0516407201852513e-06,
+      "loss": 2.355,
+      "step": 2210
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0501769839226202e-06,
+      "loss": 2.6249,
+      "step": 2212
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.0487131398752573e-06,
+      "loss": 2.3273,
+      "step": 2214
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.0472491911876332e-06,
+      "loss": 2.2022,
+      "step": 2216
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.045785141004443e-06,
+      "loss": 2.5215,
+      "step": 2218
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.0443209924705999e-06,
+      "loss": 2.3913,
+      "step": 2220
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.0428567487312279e-06,
+      "loss": 2.0402,
+      "step": 2222
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.041392412931656e-06,
+      "loss": 2.2759,
+      "step": 2224
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.0399279882174114e-06,
+      "loss": 2.3752,
+      "step": 2226
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.0384634777342114e-06,
+      "loss": 2.4845,
+      "step": 2228
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.0369988846279576e-06,
+      "loss": 2.2978,
+      "step": 2230
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.03553421204473e-06,
+      "loss": 2.4405,
+      "step": 2232
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.0340694631307786e-06,
+      "loss": 2.2374,
+      "step": 2234
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.0326046410325166e-06,
+      "loss": 2.3413,
+      "step": 2236
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.0311397488965166e-06,
+      "loss": 2.4933,
+      "step": 2238
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.0296747898694994e-06,
+      "loss": 2.2753,
+      "step": 2240
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.028209767098331e-06,
+      "loss": 2.5901,
+      "step": 2242
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.0267446837300133e-06,
+      "loss": 2.5863,
+      "step": 2244
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.0252795429116792e-06,
+      "loss": 2.432,
+      "step": 2246
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.0238143477905842e-06,
+      "loss": 2.4135,
+      "step": 2248
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.0223491015141012e-06,
+      "loss": 2.2242,
+      "step": 2250
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.0208838072297129e-06,
+      "loss": 2.3453,
+      "step": 2252
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.019418468085004e-06,
+      "loss": 2.3431,
+      "step": 2254
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.0179530872276573e-06,
+      "loss": 2.1643,
+      "step": 2256
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.016487667805444e-06,
+      "loss": 2.2139,
+      "step": 2258
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.0150222129662182e-06,
+      "loss": 2.2468,
+      "step": 2260
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0135567258579111e-06,
+      "loss": 2.4914,
+      "step": 2262
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0120912096285222e-06,
+      "loss": 2.3727,
+      "step": 2264
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0106256674261136e-06,
+      "loss": 2.1403,
+      "step": 2266
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0091601023988031e-06,
+      "loss": 2.3905,
+      "step": 2268
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.007694517694759e-06,
+      "loss": 2.1131,
+      "step": 2270
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.00622891646219e-06,
+      "loss": 2.3395,
+      "step": 2272
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0047633018493408e-06,
+      "loss": 2.5324,
+      "step": 2274
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0032976770044854e-06,
+      "loss": 2.2844,
+      "step": 2276
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0018320450759195e-06,
+      "loss": 2.4777,
+      "step": 2278
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.000366409211954e-06,
+      "loss": 2.4501,
+      "step": 2280
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.989007725609083e-07,
+      "loss": 2.5747,
+      "step": 2282
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.974351382711028e-07,
+      "loss": 2.5167,
+      "step": 2284
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.95969509490854e-07,
+      "loss": 2.4251,
+      "step": 2286
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.945038893684663e-07,
+      "loss": 2.4821,
+      "step": 2288
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.930382810522246e-07,
+      "loss": 2.2526,
+      "step": 2290
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.915726876903891e-07,
+      "loss": 2.3031,
+      "step": 2292
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.901071124311883e-07,
+      "loss": 2.1421,
+      "step": 2294
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.88641558422811e-07,
+      "loss": 2.6568,
+      "step": 2296
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.871760288134006e-07,
+      "loss": 2.3728,
+      "step": 2298
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.857105267510484e-07,
+      "loss": 2.4358,
+      "step": 2300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.842450553837865e-07,
+      "loss": 2.4469,
+      "step": 2302
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.827796178595805e-07,
+      "loss": 2.2206,
+      "step": 2304
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.813142173263233e-07,
+      "loss": 2.2388,
+      "step": 2306
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.798488569318295e-07,
+      "loss": 2.5086,
+      "step": 2308
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.78383539823827e-07,
+      "loss": 2.3137,
+      "step": 2310
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.76918269149949e-07,
+      "loss": 2.0895,
+      "step": 2312
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.754530480577314e-07,
+      "loss": 2.4254,
+      "step": 2314
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.739878796946027e-07,
+      "loss": 2.4146,
+      "step": 2316
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.725227672078772e-07,
+      "loss": 2.5433,
+      "step": 2318
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.710577137447505e-07,
+      "loss": 2.4301,
+      "step": 2320
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.695927224522907e-07,
+      "loss": 2.4225,
+      "step": 2322
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.68127796477433e-07,
+      "loss": 2.3673,
+      "step": 2324
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.66662938966971e-07,
+      "loss": 2.2955,
+      "step": 2326
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.651981530675524e-07,
+      "loss": 2.2772,
+      "step": 2328
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.637334419256713e-07,
+      "loss": 2.4733,
+      "step": 2330
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.6226880868766e-07,
+      "loss": 2.8809,
+      "step": 2332
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.60804256499684e-07,
+      "loss": 2.3127,
+      "step": 2334
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.593397885077353e-07,
+      "loss": 2.6099,
+      "step": 2336
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.57875407857624e-07,
+      "loss": 2.3925,
+      "step": 2338
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.564111176949734e-07,
+      "loss": 2.4485,
+      "step": 2340
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.549469211652123e-07,
+      "loss": 2.2315,
+      "step": 2342
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.534828214135682e-07,
+      "loss": 2.2182,
+      "step": 2344
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.520188215850602e-07,
+      "loss": 2.4335,
+      "step": 2346
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.505549248244936e-07,
+      "loss": 2.5005,
+      "step": 2348
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.490911342764525e-07,
+      "loss": 2.2757,
+      "step": 2350
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.476274530852913e-07,
+      "loss": 2.3001,
+      "step": 2352
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.46163884395131e-07,
+      "loss": 2.173,
+      "step": 2354
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.447004313498507e-07,
+      "loss": 2.363,
+      "step": 2356
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.432370970930809e-07,
+      "loss": 2.2835,
+      "step": 2358
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.41773884768196e-07,
+      "loss": 2.307,
+      "step": 2360
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.403107975183102e-07,
+      "loss": 2.2666,
+      "step": 2362
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.388478384862682e-07,
+      "loss": 2.3614,
+      "step": 2364
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.373850108146386e-07,
+      "loss": 2.0857,
+      "step": 2366
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.359223176457087e-07,
+      "loss": 2.5566,
+      "step": 2368
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.344597621214771e-07,
+      "loss": 2.3728,
+      "step": 2370
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.329973473836458e-07,
+      "loss": 2.3273,
+      "step": 2372
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.315350765736148e-07,
+      "loss": 2.1808,
+      "step": 2374
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.300729528324756e-07,
+      "loss": 2.3019,
+      "step": 2376
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.286109793010026e-07,
+      "loss": 2.2624,
+      "step": 2378
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.271491591196482e-07,
+      "loss": 2.4553,
+      "step": 2380
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.256874954285354e-07,
+      "loss": 2.2152,
+      "step": 2382
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.242259913674509e-07,
+      "loss": 2.715,
+      "step": 2384
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.227646500758387e-07,
+      "loss": 2.1977,
+      "step": 2386
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.213034746927926e-07,
+      "loss": 2.5103,
+      "step": 2388
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.198424683570505e-07,
+      "loss": 2.3052,
+      "step": 2390
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.183816342069878e-07,
+      "loss": 2.1901,
+      "step": 2392
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.169209753806082e-07,
+      "loss": 2.2311,
+      "step": 2394
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.154604950155404e-07,
+      "loss": 2.2232,
+      "step": 2396
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.140001962490293e-07,
+      "loss": 2.3053,
+      "step": 2398
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.125400822179301e-07,
+      "loss": 2.4779,
+      "step": 2400
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.110801560586998e-07,
+      "loss": 2.441,
+      "step": 2402
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.096204209073933e-07,
+      "loss": 2.5225,
+      "step": 2404
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.081608798996549e-07,
+      "loss": 2.3408,
+      "step": 2406
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.067015361707109e-07,
+      "loss": 2.3998,
+      "step": 2408
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.05242392855365e-07,
+      "loss": 2.308,
+      "step": 2410
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.037834530879897e-07,
+      "loss": 2.3218,
+      "step": 2412
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.023247200025213e-07,
+      "loss": 2.5219,
+      "step": 2414
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.008661967324503e-07,
+      "loss": 2.346,
+      "step": 2416
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.994078864108179e-07,
+      "loss": 2.0946,
+      "step": 2418
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.979497921702081e-07,
+      "loss": 2.3819,
+      "step": 2420
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.964919171427389e-07,
+      "loss": 2.1452,
+      "step": 2422
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.950342644600595e-07,
+      "loss": 2.0626,
+      "step": 2424
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.935768372533408e-07,
+      "loss": 2.3418,
+      "step": 2426
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.921196386532685e-07,
+      "loss": 2.4828,
+      "step": 2428
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.906626717900383e-07,
+      "loss": 2.2321,
+      "step": 2430
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.892059397933476e-07,
+      "loss": 2.2295,
+      "step": 2432
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.877494457923899e-07,
+      "loss": 2.2563,
+      "step": 2434
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.862931929158462e-07,
+      "loss": 2.4069,
+      "step": 2436
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.848371842918804e-07,
+      "loss": 2.4871,
+      "step": 2438
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.833814230481327e-07,
+      "loss": 2.1938,
+      "step": 2440
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.819259123117096e-07,
+      "loss": 2.5121,
+      "step": 2442
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.804706552091811e-07,
+      "loss": 2.3073,
+      "step": 2444
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.790156548665723e-07,
+      "loss": 2.3042,
+      "step": 2446
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.775609144093562e-07,
+      "loss": 2.0554,
+      "step": 2448
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.761064369624477e-07,
+      "loss": 2.2464,
+      "step": 2450
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.746522256501973e-07,
+      "loss": 2.5404,
+      "step": 2452
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.73198283596383e-07,
+      "loss": 2.7003,
+      "step": 2454
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.717446139242044e-07,
+      "loss": 2.2159,
+      "step": 2456
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.702912197562771e-07,
+      "loss": 2.2748,
+      "step": 2458
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.688381042146236e-07,
+      "loss": 2.3034,
+      "step": 2460
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.673852704206689e-07,
+      "loss": 2.2095,
+      "step": 2462
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.65932721495232e-07,
+      "loss": 2.2086,
+      "step": 2464
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.644804605585201e-07,
+      "loss": 2.2465,
+      "step": 2466
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.630284907301225e-07,
+      "loss": 2.3245,
+      "step": 2468
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.615768151290024e-07,
+      "loss": 2.294,
+      "step": 2470
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.601254368734908e-07,
+      "loss": 2.3998,
+      "step": 2472
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.586743590812808e-07,
+      "loss": 2.5294,
+      "step": 2474
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.572235848694202e-07,
+      "loss": 2.3761,
+      "step": 2476
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.557731173543029e-07,
+      "loss": 2.1873,
+      "step": 2478
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.54322959651666e-07,
+      "loss": 2.6505,
+      "step": 2480
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.528731148765806e-07,
+      "loss": 2.1418,
+      "step": 2482
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.521483108101453e-07,
+      "loss": 2.2971,
+      "step": 2484
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.506989412656718e-07,
+      "loss": 2.178,
+      "step": 2486
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.492498924334711e-07,
+      "loss": 2.1965,
+      "step": 2488
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.47801167426232e-07,
+      "loss": 2.2415,
+      "step": 2490
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.463527693559475e-07,
+      "loss": 2.2727,
+      "step": 2492
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.44904701333909e-07,
+      "loss": 2.4316,
+      "step": 2494
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.434569664706982e-07,
+      "loss": 2.2054,
+      "step": 2496
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.420095678761817e-07,
+      "loss": 2.3326,
+      "step": 2498
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.405625086595034e-07,
+      "loss": 2.4215,
+      "step": 2500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.391157919290781e-07,
+      "loss": 2.3286,
+      "step": 2502
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.376694207925854e-07,
+      "loss": 2.34,
+      "step": 2504
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 8.362233983569622e-07,
+      "loss": 2.8653,
+      "step": 2506
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 8.347777277283961e-07,
+      "loss": 2.6285,
+      "step": 2508
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 8.333324120123194e-07,
+      "loss": 2.5004,
+      "step": 2510
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 8.318874543134025e-07,
+      "loss": 2.4015,
+      "step": 2512
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.304428577355452e-07,
+      "loss": 2.2038,
+      "step": 2514
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.297206958367709e-07,
+      "loss": 2.3619,
+      "step": 2516
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.282766467586193e-07,
+      "loss": 2.398,
+      "step": 2518
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.268329665578708e-07,
+      "loss": 2.1842,
+      "step": 2520
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.253896583356824e-07,
+      "loss": 2.4201,
+      "step": 2522
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.239467251924118e-07,
+      "loss": 2.3461,
+      "step": 2524
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.225041702276107e-07,
+      "loss": 2.5341,
+      "step": 2526
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.210619965400183e-07,
+      "loss": 2.5787,
+      "step": 2528
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.196202072275554e-07,
+      "loss": 2.222,
+      "step": 2530
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.181788053873167e-07,
+      "loss": 2.4516,
+      "step": 2532
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.167377941155642e-07,
+      "loss": 2.5411,
+      "step": 2534
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.152971765077219e-07,
+      "loss": 2.5724,
+      "step": 2536
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.138569556583672e-07,
+      "loss": 2.4833,
+      "step": 2538
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.124171346612263e-07,
+      "loss": 2.5704,
+      "step": 2540
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.109777166091652e-07,
+      "loss": 2.5729,
+      "step": 2542
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.095387045941852e-07,
+      "loss": 2.1751,
+      "step": 2544
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.081001017074152e-07,
+      "loss": 2.3624,
+      "step": 2546
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.066619110391054e-07,
+      "loss": 2.4528,
+      "step": 2548
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 8.0522413567862e-07,
+      "loss": 2.4156,
+      "step": 2550
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 8.037867787144314e-07,
+      "loss": 2.4374,
+      "step": 2552
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 8.02349843234114e-07,
+      "loss": 2.2263,
+      "step": 2554
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 8.009133323243351e-07,
+      "loss": 2.5188,
+      "step": 2556
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.994772490708514e-07,
+      "loss": 2.1436,
+      "step": 2558
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.980415965585011e-07,
+      "loss": 2.4208,
+      "step": 2560
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.966063778711956e-07,
+      "loss": 2.4573,
+      "step": 2562
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.951715960919157e-07,
+      "loss": 2.4805,
+      "step": 2564
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.937372543027035e-07,
+      "loss": 2.3773,
+      "step": 2566
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.92303355584656e-07,
+      "loss": 2.327,
+      "step": 2568
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.908699030179177e-07,
+      "loss": 2.354,
+      "step": 2570
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.894368996816755e-07,
+      "loss": 2.3706,
+      "step": 2572
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.880043486541517e-07,
+      "loss": 2.43,
+      "step": 2574
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.865722530125952e-07,
+      "loss": 2.613,
+      "step": 2576
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.851406158332786e-07,
+      "loss": 2.1504,
+      "step": 2578
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.837094401914889e-07,
+      "loss": 2.5713,
+      "step": 2580
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.82278729161522e-07,
+      "loss": 2.2792,
+      "step": 2582
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.808484858166749e-07,
+      "loss": 2.4921,
+      "step": 2584
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.794187132292408e-07,
+      "loss": 2.4913,
+      "step": 2586
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.77989414470502e-07,
+      "loss": 2.2147,
+      "step": 2588
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.765605926107213e-07,
+      "loss": 2.7114,
+      "step": 2590
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.75132250719139e-07,
+      "loss": 2.3006,
+      "step": 2592
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.737043918639634e-07,
+      "loss": 2.4085,
+      "step": 2594
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.722770191123658e-07,
+      "loss": 2.2064,
+      "step": 2596
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.708501355304719e-07,
+      "loss": 2.5217,
+      "step": 2598
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.694237441833584e-07,
+      "loss": 2.4016,
+      "step": 2600
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.679978481350437e-07,
+      "loss": 2.2748,
+      "step": 2602
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.665724504484819e-07,
+      "loss": 2.7744,
+      "step": 2604
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.651475541855577e-07,
+      "loss": 2.5243,
+      "step": 2606
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.637231624070776e-07,
+      "loss": 2.5171,
+      "step": 2608
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.622992781727652e-07,
+      "loss": 2.6107,
+      "step": 2610
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.608759045412534e-07,
+      "loss": 2.3215,
+      "step": 2612
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.594530445700781e-07,
+      "loss": 2.2903,
+      "step": 2614
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.580307013156726e-07,
+      "loss": 2.4102,
+      "step": 2616
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.566088778333597e-07,
+      "loss": 2.3011,
+      "step": 2618
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.551875771773451e-07,
+      "loss": 2.4777,
+      "step": 2620
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.53766802400713e-07,
+      "loss": 2.2883,
+      "step": 2622
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.523465565554165e-07,
+      "loss": 2.603,
+      "step": 2624
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.509268426922729e-07,
+      "loss": 2.3403,
+      "step": 2626
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.495076638609573e-07,
+      "loss": 2.2766,
+      "step": 2628
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.48089023109995e-07,
+      "loss": 2.5495,
+      "step": 2630
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.466709234867551e-07,
+      "loss": 2.2699,
+      "step": 2632
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.452533680374451e-07,
+      "loss": 2.232,
+      "step": 2634
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.438363598071037e-07,
+      "loss": 2.4567,
+      "step": 2636
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.424199018395927e-07,
+      "loss": 2.3238,
+      "step": 2638
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.410039971775932e-07,
+      "loss": 2.2977,
+      "step": 2640
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.395886488625975e-07,
+      "loss": 2.513,
+      "step": 2642
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.38173859934903e-07,
+      "loss": 2.4299,
+      "step": 2644
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.367596334336046e-07,
+      "loss": 2.2186,
+      "step": 2646
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.353459723965896e-07,
+      "loss": 2.6373,
+      "step": 2648
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.339328798605315e-07,
+      "loss": 2.1771,
+      "step": 2650
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.325203588608808e-07,
+      "loss": 2.3566,
+      "step": 2652
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.311084124318617e-07,
+      "loss": 2.3604,
+      "step": 2654
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.296970436064634e-07,
+      "loss": 2.6705,
+      "step": 2656
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.282862554164355e-07,
+      "loss": 2.306,
+      "step": 2658
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.268760508922786e-07,
+      "loss": 2.3193,
+      "step": 2660
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.25466433063241e-07,
+      "loss": 2.4695,
+      "step": 2662
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.240574049573105e-07,
+      "loss": 2.3212,
+      "step": 2664
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.226489696012068e-07,
+      "loss": 2.3396,
+      "step": 2666
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.212411300203778e-07,
+      "loss": 2.7672,
+      "step": 2668
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.198338892389915e-07,
+      "loss": 2.3035,
+      "step": 2670
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.184272502799291e-07,
+      "loss": 2.2063,
+      "step": 2672
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.170212161647789e-07,
+      "loss": 2.3545,
+      "step": 2674
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.156157899138305e-07,
+      "loss": 2.1973,
+      "step": 2676
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.142109745460676e-07,
+      "loss": 2.3296,
+      "step": 2678
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.128067730791611e-07,
+      "loss": 2.4783,
+      "step": 2680
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.114031885294642e-07,
+      "loss": 2.6186,
+      "step": 2682
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.10000223912004e-07,
+      "loss": 2.1012,
+      "step": 2684
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.085978822404766e-07,
+      "loss": 2.3392,
+      "step": 2686
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.071961665272392e-07,
+      "loss": 2.3101,
+      "step": 2688
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.05795079783305e-07,
+      "loss": 2.2984,
+      "step": 2690
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.043946250183361e-07,
+      "loss": 2.3071,
+      "step": 2692
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.02994805240637e-07,
+      "loss": 2.5653,
+      "step": 2694
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.015956234571474e-07,
+      "loss": 2.4521,
+      "step": 2696
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.001970826734378e-07,
+      "loss": 2.5733,
+      "step": 2698
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.987991858937015e-07,
+      "loss": 2.0661,
+      "step": 2700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.97401936120747e-07,
+      "loss": 2.5841,
+      "step": 2702
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.96005336355995e-07,
+      "loss": 2.2641,
+      "step": 2704
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.94609389599469e-07,
+      "loss": 2.2637,
+      "step": 2706
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.932140988497892e-07,
+      "loss": 2.2499,
+      "step": 2708
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.918194671041674e-07,
+      "loss": 2.509,
+      "step": 2710
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.904254973584002e-07,
+      "loss": 2.2785,
+      "step": 2712
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.890321926068608e-07,
+      "loss": 2.4755,
+      "step": 2714
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.876395558424949e-07,
+      "loss": 2.4773,
+      "step": 2716
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.862475900568132e-07,
+      "loss": 2.333,
+      "step": 2718
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.848562982398855e-07,
+      "loss": 2.4779,
+      "step": 2720
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.834656833803326e-07,
+      "loss": 2.4282,
+      "step": 2722
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.82075748465322e-07,
+      "loss": 2.104,
+      "step": 2724
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.806864964805612e-07,
+      "loss": 2.483,
+      "step": 2726
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.792979304102888e-07,
+      "loss": 2.2719,
+      "step": 2728
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.77910053237272e-07,
+      "loss": 2.5435,
+      "step": 2730
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.765228679427973e-07,
+      "loss": 2.5445,
+      "step": 2732
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.751363775066652e-07,
+      "loss": 2.3369,
+      "step": 2734
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.737505849071828e-07,
+      "loss": 2.3555,
+      "step": 2736
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.723654931211593e-07,
+      "loss": 2.1379,
+      "step": 2738
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.709811051238982e-07,
+      "loss": 2.1932,
+      "step": 2740
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.695974238891906e-07,
+      "loss": 2.5776,
+      "step": 2742
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.682144523893098e-07,
+      "loss": 2.1765,
+      "step": 2744
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.66832193595005e-07,
+      "loss": 2.1806,
+      "step": 2746
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.654506504754937e-07,
+      "loss": 2.3733,
+      "step": 2748
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.640698259984563e-07,
+      "loss": 2.3373,
+      "step": 2750
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.626897231300294e-07,
+      "loss": 2.2696,
+      "step": 2752
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.613103448348004e-07,
+      "loss": 2.2713,
+      "step": 2754
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.599316940757986e-07,
+      "loss": 2.2902,
+      "step": 2756
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.585537738144916e-07,
+      "loss": 2.3606,
+      "step": 2758
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.571765870107782e-07,
+      "loss": 2.3262,
+      "step": 2760
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.558001366229807e-07,
+      "loss": 2.2879,
+      "step": 2762
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.544244256078395e-07,
+      "loss": 2.2362,
+      "step": 2764
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.530494569205078e-07,
+      "loss": 2.5146,
+      "step": 2766
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.516752335145435e-07,
+      "loss": 2.6295,
+      "step": 2768
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.503017583419033e-07,
+      "loss": 2.291,
+      "step": 2770
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.48929034352937e-07,
+      "loss": 2.3465,
+      "step": 2772
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.475570644963811e-07,
+      "loss": 2.2647,
+      "step": 2774
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.461858517193517e-07,
+      "loss": 2.5567,
+      "step": 2776
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.448153989673384e-07,
+      "loss": 2.3326,
+      "step": 2778
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.434457091841988e-07,
+      "loss": 2.4799,
+      "step": 2780
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.420767853121518e-07,
+      "loss": 2.2515,
+      "step": 2782
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.407086302917694e-07,
+      "loss": 2.0771,
+      "step": 2784
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.393412470619741e-07,
+      "loss": 2.4317,
+      "step": 2786
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.379746385600294e-07,
+      "loss": 2.4658,
+      "step": 2788
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.366088077215349e-07,
+      "loss": 2.333,
+      "step": 2790
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.352437574804194e-07,
+      "loss": 2.3936,
+      "step": 2792
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.338794907689351e-07,
+      "loss": 2.3265,
+      "step": 2794
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.325160105176519e-07,
+      "loss": 2.1475,
+      "step": 2796
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.311533196554486e-07,
+      "loss": 2.2806,
+      "step": 2798
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.297914211095097e-07,
+      "loss": 2.4101,
+      "step": 2800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.284303178053179e-07,
+      "loss": 2.2414,
+      "step": 2802
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.270700126666457e-07,
+      "loss": 2.3114,
+      "step": 2804
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.257105086155537e-07,
+      "loss": 2.0424,
+      "step": 2806
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.243518085723795e-07,
+      "loss": 2.4947,
+      "step": 2808
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.229939154557358e-07,
+      "loss": 2.1634,
+      "step": 2810
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.216368321824992e-07,
+      "loss": 2.2303,
+      "step": 2812
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.20280561667809e-07,
+      "loss": 2.4121,
+      "step": 2814
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.189251068250582e-07,
+      "loss": 2.3059,
+      "step": 2816
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.175704705658861e-07,
+      "loss": 2.1436,
+      "step": 2818
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.162166558001756e-07,
+      "loss": 2.4704,
+      "step": 2820
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.148636654360437e-07,
+      "loss": 2.3432,
+      "step": 2822
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.135115023798378e-07,
+      "loss": 2.4694,
+      "step": 2824
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.12160169536126e-07,
+      "loss": 2.1774,
+      "step": 2826
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.108096698076952e-07,
+      "loss": 2.3985,
+      "step": 2828
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.09460006095542e-07,
+      "loss": 2.1717,
+      "step": 2830
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.08111181298866e-07,
+      "loss": 2.3996,
+      "step": 2832
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.067631983150665e-07,
+      "loss": 2.3045,
+      "step": 2834
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.054160600397336e-07,
+      "loss": 2.4791,
+      "step": 2836
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.04069769366643e-07,
+      "loss": 2.2783,
+      "step": 2838
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.027243291877495e-07,
+      "loss": 2.3579,
+      "step": 2840
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.013797423931813e-07,
+      "loss": 2.3617,
+      "step": 2842
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 6.000360118712336e-07,
+      "loss": 2.3048,
+      "step": 2844
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.986931405083611e-07,
+      "loss": 2.4872,
+      "step": 2846
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.973511311891747e-07,
+      "loss": 2.2881,
+      "step": 2848
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.960099867964319e-07,
+      "loss": 2.4967,
+      "step": 2850
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.94669710211034e-07,
+      "loss": 2.6076,
+      "step": 2852
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.933303043120162e-07,
+      "loss": 2.1849,
+      "step": 2854
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.919917719765446e-07,
+      "loss": 2.2264,
+      "step": 2856
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.906541160799092e-07,
+      "loss": 2.6351,
+      "step": 2858
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.893173394955158e-07,
+      "loss": 2.2375,
+      "step": 2860
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.879814450948826e-07,
+      "loss": 2.337,
+      "step": 2862
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.866464357476324e-07,
+      "loss": 2.5032,
+      "step": 2864
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.853123143214876e-07,
+      "loss": 2.6742,
+      "step": 2866
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.839790836822614e-07,
+      "loss": 2.2436,
+      "step": 2868
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.826467466938553e-07,
+      "loss": 2.2489,
+      "step": 2870
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.813153062182504e-07,
+      "loss": 2.4944,
+      "step": 2872
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.799847651155022e-07,
+      "loss": 2.242,
+      "step": 2874
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.786551262437341e-07,
+      "loss": 2.2145,
+      "step": 2876
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.773263924591309e-07,
+      "loss": 2.3368,
+      "step": 2878
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.759985666159348e-07,
+      "loss": 2.3857,
+      "step": 2880
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.746716515664363e-07,
+      "loss": 2.5574,
+      "step": 2882
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.733456501609697e-07,
+      "loss": 2.6827,
+      "step": 2884
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.720205652479067e-07,
+      "loss": 2.3012,
+      "step": 2886
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.706963996736504e-07,
+      "loss": 2.4365,
+      "step": 2888
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.693731562826288e-07,
+      "loss": 2.0728,
+      "step": 2890
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.6805083791729e-07,
+      "loss": 2.3692,
+      "step": 2892
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.667294474180937e-07,
+      "loss": 2.4155,
+      "step": 2894
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.654089876235073e-07,
+      "loss": 2.1968,
+      "step": 2896
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.640894613699982e-07,
+      "loss": 2.0421,
+      "step": 2898
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.6277087149203e-07,
+      "loss": 2.4892,
+      "step": 2900
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.614532208220528e-07,
+      "loss": 2.2188,
+      "step": 2902
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.601365121905007e-07,
+      "loss": 2.3704,
+      "step": 2904
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.588207484257839e-07,
+      "loss": 2.3284,
+      "step": 2906
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.575059323542825e-07,
+      "loss": 2.3581,
+      "step": 2908
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.561920668003413e-07,
+      "loss": 2.2069,
+      "step": 2910
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.548791545862632e-07,
+      "loss": 2.377,
+      "step": 2912
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.535671985323034e-07,
+      "loss": 2.552,
+      "step": 2914
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.522562014566628e-07,
+      "loss": 2.4803,
+      "step": 2916
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.509461661754822e-07,
+      "loss": 2.2848,
+      "step": 2918
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.496370955028379e-07,
+      "loss": 2.3721,
+      "step": 2920
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.483289922507312e-07,
+      "loss": 2.2689,
+      "step": 2922
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.470218592290883e-07,
+      "loss": 2.4126,
+      "step": 2924
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.457156992457493e-07,
+      "loss": 2.4723,
+      "step": 2926
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.444105151064657e-07,
+      "loss": 2.5204,
+      "step": 2928
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.431063096148908e-07,
+      "loss": 2.2726,
+      "step": 2930
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.418030855725776e-07,
+      "loss": 2.3093,
+      "step": 2932
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.405008457789697e-07,
+      "loss": 2.5789,
+      "step": 2934
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.391995930313969e-07,
+      "loss": 2.496,
+      "step": 2936
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.378993301250682e-07,
+      "loss": 2.2452,
+      "step": 2938
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.366000598530674e-07,
+      "loss": 2.3148,
+      "step": 2940
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.353017850063452e-07,
+      "loss": 2.7162,
+      "step": 2942
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.340045083737142e-07,
+      "loss": 2.4087,
+      "step": 2944
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.327082327418419e-07,
+      "loss": 2.2211,
+      "step": 2946
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.314129608952484e-07,
+      "loss": 2.0549,
+      "step": 2948
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.301186956162935e-07,
+      "loss": 2.5625,
+      "step": 2950
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.288254396851782e-07,
+      "loss": 2.2932,
+      "step": 2952
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.275331958799335e-07,
+      "loss": 2.3089,
+      "step": 2954
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.262419669764171e-07,
+      "loss": 2.4793,
+      "step": 2956
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.24951755748306e-07,
+      "loss": 2.2383,
+      "step": 2958
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.236625649670913e-07,
+      "loss": 2.4058,
+      "step": 2960
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.223743974020727e-07,
+      "loss": 2.4452,
+      "step": 2962
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.210872558203514e-07,
+      "loss": 2.232,
+      "step": 2964
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.198011429868245e-07,
+      "loss": 2.6852,
+      "step": 2966
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.185160616641793e-07,
+      "loss": 2.0647,
+      "step": 2968
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.172320146128884e-07,
+      "loss": 2.5407,
+      "step": 2970
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.159490045912006e-07,
+      "loss": 2.313,
+      "step": 2972
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.146670343551394e-07,
+      "loss": 2.291,
+      "step": 2974
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.133861066584929e-07,
+      "loss": 2.2558,
+      "step": 2976
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.121062242528109e-07,
+      "loss": 2.6145,
+      "step": 2978
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.108273898873967e-07,
+      "loss": 2.4894,
+      "step": 2980
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.095496063093044e-07,
+      "loss": 2.5229,
+      "step": 2982
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.082728762633278e-07,
+      "loss": 2.1288,
+      "step": 2984
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.069972024920003e-07,
+      "loss": 2.467,
+      "step": 2986
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.057225877355851e-07,
+      "loss": 2.5339,
+      "step": 2988
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.044490347320715e-07,
+      "loss": 2.123,
+      "step": 2990
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.031765462171659e-07,
+      "loss": 2.5362,
+      "step": 2992
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.019051249242907e-07,
+      "loss": 2.3589,
+      "step": 2994
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.006347735845744e-07,
+      "loss": 2.612,
+      "step": 2996
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.993654949268472e-07,
+      "loss": 2.2357,
+      "step": 2998
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.980972916776349e-07,
+      "loss": 2.0088,
+      "step": 3000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.968301665611545e-07,
+      "loss": 2.4529,
+      "step": 3002
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.955641222993057e-07,
+      "loss": 2.352,
+      "step": 3004
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.942991616116667e-07,
+      "loss": 2.2461,
+      "step": 3006
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.930352872154881e-07,
+      "loss": 2.4761,
+      "step": 3008
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.917725018256885e-07,
+      "loss": 2.3513,
+      "step": 3010
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.905108081548441e-07,
+      "loss": 2.0589,
+      "step": 3012
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.892502089131891e-07,
+      "loss": 2.3859,
+      "step": 3014
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.879907068086047e-07,
+      "loss": 2.4326,
+      "step": 3016
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.867323045466173e-07,
+      "loss": 2.2585,
+      "step": 3018
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.854750048303881e-07,
+      "loss": 2.2026,
+      "step": 3020
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.842188103607126e-07,
+      "loss": 2.4944,
+      "step": 3022
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.829637238360103e-07,
+      "loss": 2.3826,
+      "step": 3024
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.817097479523214e-07,
+      "loss": 2.4484,
+      "step": 3026
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.804568854033003e-07,
+      "loss": 2.1429,
+      "step": 3028
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.792051388802103e-07,
+      "loss": 2.5137,
+      "step": 3030
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.779545110719167e-07,
+      "loss": 2.6513,
+      "step": 3032
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.767050046648818e-07,
+      "loss": 2.4917,
+      "step": 3034
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.7545662234315933e-07,
+      "loss": 2.3492,
+      "step": 3036
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.742093667883881e-07,
+      "loss": 2.1966,
+      "step": 3038
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.729632406797868e-07,
+      "loss": 2.5651,
+      "step": 3040
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.717182466941472e-07,
+      "loss": 2.2645,
+      "step": 3042
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.704743875058307e-07,
+      "loss": 2.2814,
+      "step": 3044
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.6923166578675976e-07,
+      "loss": 2.247,
+      "step": 3046
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.679900842064137e-07,
+      "loss": 2.3914,
+      "step": 3048
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.667496454318225e-07,
+      "loss": 2.5838,
+      "step": 3050
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.6551035212756284e-07,
+      "loss": 2.348,
+      "step": 3052
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.642722069557481e-07,
+      "loss": 2.4168,
+      "step": 3054
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.63035212576028e-07,
+      "loss": 2.3571,
+      "step": 3056
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.6179937164557824e-07,
+      "loss": 2.1961,
+      "step": 3058
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.60564686819099e-07,
+      "loss": 2.2318,
+      "step": 3060
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.593311607488042e-07,
+      "loss": 2.3089,
+      "step": 3062
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.5809879608442127e-07,
+      "loss": 2.32,
+      "step": 3064
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.5686759547318145e-07,
+      "loss": 2.3229,
+      "step": 3066
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.556375615598157e-07,
+      "loss": 2.1972,
+      "step": 3068
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.5440869698654847e-07,
+      "loss": 2.0506,
+      "step": 3070
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.5318100439309383e-07,
+      "loss": 2.4581,
+      "step": 3072
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.519544864166467e-07,
+      "loss": 2.428,
+      "step": 3074
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.5072914569187936e-07,
+      "loss": 2.3864,
+      "step": 3076
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.4950498485093504e-07,
+      "loss": 2.2819,
+      "step": 3078
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.4828200652342376e-07,
+      "loss": 2.3175,
+      "step": 3080
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.4706021333641286e-07,
+      "loss": 2.4947,
+      "step": 3082
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.458396079144267e-07,
+      "loss": 2.6006,
+      "step": 3084
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.4462019287943633e-07,
+      "loss": 2.1521,
+      "step": 3086
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.434019708508563e-07,
+      "loss": 2.1405,
+      "step": 3088
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.421849444455382e-07,
+      "loss": 2.3048,
+      "step": 3090
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.4096911627776633e-07,
+      "loss": 2.2149,
+      "step": 3092
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.397544889592499e-07,
+      "loss": 2.2301,
+      "step": 3094
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.38541065099119e-07,
+      "loss": 2.2929,
+      "step": 3096
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.373288473039185e-07,
+      "loss": 2.5705,
+      "step": 3098
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.361178381776034e-07,
+      "loss": 2.1664,
+      "step": 3100
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.3490804032153027e-07,
+      "loss": 2.2688,
+      "step": 3102
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.3369945633445626e-07,
+      "loss": 2.391,
+      "step": 3104
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.324920888125294e-07,
+      "loss": 2.2777,
+      "step": 3106
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.3128594034928514e-07,
+      "loss": 2.0721,
+      "step": 3108
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.300810135356397e-07,
+      "loss": 2.3379,
+      "step": 3110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.2887731095988664e-07,
+      "loss": 2.2214,
+      "step": 3112
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.2767483520768833e-07,
+      "loss": 2.6695,
+      "step": 3114
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.26473588862072e-07,
+      "loss": 2.3539,
+      "step": 3116
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.252735745034243e-07,
+      "loss": 2.3112,
+      "step": 3118
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.2407479470948495e-07,
+      "loss": 2.1182,
+      "step": 3120
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.22877252055343e-07,
+      "loss": 2.1283,
+      "step": 3122
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.2168094911342867e-07,
+      "loss": 2.2434,
+      "step": 3124
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.2048588845350974e-07,
+      "loss": 2.3956,
+      "step": 3126
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.192920726426853e-07,
+      "loss": 2.2923,
+      "step": 3128
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.1809950424538043e-07,
+      "loss": 2.4565,
+      "step": 3130
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.169081858233404e-07,
+      "loss": 2.3932,
+      "step": 3132
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.157181199356262e-07,
+      "loss": 2.1525,
+      "step": 3134
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.145293091386076e-07,
+      "loss": 2.1397,
+      "step": 3136
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.1334175598595845e-07,
+      "loss": 2.2545,
+      "step": 3138
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.121554630286506e-07,
+      "loss": 2.7604,
+      "step": 3140
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.109704328149505e-07,
+      "loss": 2.1315,
+      "step": 3142
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.097866678904096e-07,
+      "loss": 2.3746,
+      "step": 3144
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.086041707978638e-07,
+      "loss": 2.262,
+      "step": 3146
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.0742294407742386e-07,
+      "loss": 2.5298,
+      "step": 3148
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.0624299026647335e-07,
+      "loss": 2.035,
+      "step": 3150
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.050643118996593e-07,
+      "loss": 2.2675,
+      "step": 3152
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.038869115088911e-07,
+      "loss": 2.2519,
+      "step": 3154
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.0271079162333175e-07,
+      "loss": 2.474,
+      "step": 3156
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.015359547693938e-07,
+      "loss": 2.5323,
+      "step": 3158
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.003624034707336e-07,
+      "loss": 2.2379,
+      "step": 3160
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.991901402482468e-07,
+      "loss": 2.2635,
+      "step": 3162
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.9801916762006137e-07,
+      "loss": 2.6183,
+      "step": 3164
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.9684948810153305e-07,
+      "loss": 2.5431,
+      "step": 3166
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.9568110420523957e-07,
+      "loss": 2.3616,
+      "step": 3168
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.945140184409772e-07,
+      "loss": 2.2793,
+      "step": 3170
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.9334823331575086e-07,
+      "loss": 2.224,
+      "step": 3172
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.9218375133377445e-07,
+      "loss": 2.3404,
+      "step": 3174
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.910205749964608e-07,
+      "loss": 2.4168,
+      "step": 3176
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.898587068024186e-07,
+      "loss": 2.3829,
+      "step": 3178
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8869814924744637e-07,
+      "loss": 2.4255,
+      "step": 3180
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.875389048245281e-07,
+      "loss": 2.2857,
+      "step": 3182
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.86380976023826e-07,
+      "loss": 2.5262,
+      "step": 3184
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8522436533267654e-07,
+      "loss": 2.2654,
+      "step": 3186
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8406907523558496e-07,
+      "loss": 2.2859,
+      "step": 3188
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.829151082142196e-07,
+      "loss": 2.5575,
+      "step": 3190
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.817624667474062e-07,
+      "loss": 2.1387,
+      "step": 3192
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.806111533111245e-07,
+      "loss": 2.1788,
+      "step": 3194
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.794611703785002e-07,
+      "loss": 2.3718,
+      "step": 3196
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7831252041980143e-07,
+      "loss": 2.4759,
+      "step": 3198
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.771652059024328e-07,
+      "loss": 2.2483,
+      "step": 3200
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.760192292909303e-07,
+      "loss": 2.1328,
+      "step": 3202
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7487459304695655e-07,
+      "loss": 2.521,
+      "step": 3204
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.737312996292941e-07,
+      "loss": 2.1955,
+      "step": 3206
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.725893514938414e-07,
+      "loss": 2.2915,
+      "step": 3208
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.7144875109360675e-07,
+      "loss": 2.4738,
+      "step": 3210
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.703095008787047e-07,
+      "loss": 2.2094,
+      "step": 3212
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.6917160329634714e-07,
+      "loss": 2.3182,
+      "step": 3214
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.6803506079084256e-07,
+      "loss": 2.1551,
+      "step": 3216
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.6689987580358773e-07,
+      "loss": 2.1328,
+      "step": 3218
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.657660507730631e-07,
+      "loss": 2.5026,
+      "step": 3220
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.6463358813482805e-07,
+      "loss": 2.1657,
+      "step": 3222
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.6350249032151594e-07,
+      "loss": 2.2007,
+      "step": 3224
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.6237275976282753e-07,
+      "loss": 2.4264,
+      "step": 3226
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.6124439888552725e-07,
+      "loss": 2.4743,
+      "step": 3228
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.6011741011343645e-07,
+      "loss": 2.1036,
+      "step": 3230
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.589917958674307e-07,
+      "loss": 2.572,
+      "step": 3232
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.578675585654305e-07,
+      "loss": 2.0887,
+      "step": 3234
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.5674470062240125e-07,
+      "loss": 2.4678,
+      "step": 3236
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.556232244503432e-07,
+      "loss": 2.4499,
+      "step": 3238
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.5450313245829034e-07,
+      "loss": 2.3848,
+      "step": 3240
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.5338442705230097e-07,
+      "loss": 2.1436,
+      "step": 3242
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.522671106354572e-07,
+      "loss": 2.5811,
+      "step": 3244
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.5115118560785607e-07,
+      "loss": 2.5358,
+      "step": 3246
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.5003665436660636e-07,
+      "loss": 2.4689,
+      "step": 3248
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.489235193058222e-07,
+      "loss": 2.2282,
+      "step": 3250
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.478117828166197e-07,
+      "loss": 2.2035,
+      "step": 3252
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.467014472871099e-07,
+      "loss": 2.1824,
+      "step": 3254
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.455925151023946e-07,
+      "loss": 2.3074,
+      "step": 3256
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.444849886445612e-07,
+      "loss": 2.1646,
+      "step": 3258
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.433788702926772e-07,
+      "loss": 2.2208,
+      "step": 3260
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.422741624227855e-07,
+      "loss": 2.397,
+      "step": 3262
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.4117086740789967e-07,
+      "loss": 2.2592,
+      "step": 3264
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.4006898761799776e-07,
+      "loss": 2.4963,
+      "step": 3266
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.3896852542001777e-07,
+      "loss": 2.5038,
+      "step": 3268
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.378694831778527e-07,
+      "loss": 2.4851,
+      "step": 3270
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.367718632523452e-07,
+      "loss": 2.43,
+      "step": 3272
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.356756680012833e-07,
+      "loss": 2.5678,
+      "step": 3274
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.345808997793942e-07,
+      "loss": 2.1539,
+      "step": 3276
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.3348756093833965e-07,
+      "loss": 2.1828,
+      "step": 3278
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.3239565382671087e-07,
+      "loss": 2.4646,
+      "step": 3280
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.313051807900239e-07,
+      "loss": 2.1353,
+      "step": 3282
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.30216144170714e-07,
+      "loss": 2.4583,
+      "step": 3284
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.2912854630813137e-07,
+      "loss": 2.2403,
+      "step": 3286
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.2804238953853524e-07,
+      "loss": 2.5755,
+      "step": 3288
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.2695767619508905e-07,
+      "loss": 2.4918,
+      "step": 3290
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.2587440860785567e-07,
+      "loss": 2.4181,
+      "step": 3292
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.2479258910379347e-07,
+      "loss": 2.1952,
+      "step": 3294
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.237122200067479e-07,
+      "loss": 2.3317,
+      "step": 3296
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.2263330363745136e-07,
+      "loss": 2.0153,
+      "step": 3298
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.215558423135136e-07,
+      "loss": 2.2844,
+      "step": 3300
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.2047983834942085e-07,
+      "loss": 2.4493,
+      "step": 3302
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.1940529405652627e-07,
+      "loss": 2.4057,
+      "step": 3304
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.183322117430497e-07,
+      "loss": 2.1862,
+      "step": 3306
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.1726059371406953e-07,
+      "loss": 2.6167,
+      "step": 3308
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.1619044227151894e-07,
+      "loss": 2.4327,
+      "step": 3310
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.151217597141802e-07,
+      "loss": 2.264,
+      "step": 3312
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.140545483376815e-07,
+      "loss": 2.1945,
+      "step": 3314
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.1298881043448976e-07,
+      "loss": 2.5421,
+      "step": 3316
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.11924548293907e-07,
+      "loss": 2.1653,
+      "step": 3318
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.108617642020651e-07,
+      "loss": 2.3909,
+      "step": 3320
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.0980046044192187e-07,
+      "loss": 2.5704,
+      "step": 3322
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.087406392932532e-07,
+      "loss": 2.3593,
+      "step": 3324
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.0768230303265254e-07,
+      "loss": 2.1842,
+      "step": 3326
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.066254539335218e-07,
+      "loss": 2.2524,
+      "step": 3328
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.0557009426607007e-07,
+      "loss": 2.4896,
+      "step": 3330
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.045162262973048e-07,
+      "loss": 2.3859,
+      "step": 3332
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.0346385229103125e-07,
+      "loss": 2.2594,
+      "step": 3334
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.0241297450784443e-07,
+      "loss": 2.3181,
+      "step": 3336
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.0136359520512545e-07,
+      "loss": 2.2603,
+      "step": 3338
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.003157166370367e-07,
+      "loss": 2.2832,
+      "step": 3340
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.992693410545165e-07,
+      "loss": 2.1501,
+      "step": 3342
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9822447070527547e-07,
+      "loss": 2.5098,
+      "step": 3344
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.971811078337899e-07,
+      "loss": 2.3524,
+      "step": 3346
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.961392546812983e-07,
+      "loss": 2.5895,
+      "step": 3348
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.950989134857963e-07,
+      "loss": 2.3269,
+      "step": 3350
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.940600864820314e-07,
+      "loss": 2.4572,
+      "step": 3352
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.9302277590149835e-07,
+      "loss": 2.3999,
+      "step": 3354
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.9198698397243535e-07,
+      "loss": 2.4635,
+      "step": 3356
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.9095271291981747e-07,
+      "loss": 2.3415,
+      "step": 3358
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.899199649653532e-07,
+      "loss": 2.3258,
+      "step": 3360
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.8888874232747896e-07,
+      "loss": 2.227,
+      "step": 3362
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.8785904722135567e-07,
+      "loss": 2.3294,
+      "step": 3364
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.868308818588611e-07,
+      "loss": 2.5475,
+      "step": 3366
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.8580424844858884e-07,
+      "loss": 2.5204,
+      "step": 3368
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.8477914919584076e-07,
+      "loss": 2.2726,
+      "step": 3370
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.837555863026236e-07,
+      "loss": 2.2483,
+      "step": 3372
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.8273356196764307e-07,
+      "loss": 2.3288,
+      "step": 3374
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.8171307838630144e-07,
+      "loss": 2.4139,
+      "step": 3376
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.8069413775068986e-07,
+      "loss": 2.4272,
+      "step": 3378
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.796767422495856e-07,
+      "loss": 2.1728,
+      "step": 3380
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.786608940684467e-07,
+      "loss": 2.1175,
+      "step": 3382
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.776465953894085e-07,
+      "loss": 2.3987,
+      "step": 3384
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.766338483912757e-07,
+      "loss": 2.4435,
+      "step": 3386
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.756226552495219e-07,
+      "loss": 2.2848,
+      "step": 3388
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.7461301813628155e-07,
+      "loss": 2.3723,
+      "step": 3390
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.73604939220348e-07,
+      "loss": 2.3791,
+      "step": 3392
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.7259842066716543e-07,
+      "loss": 2.4471,
+      "step": 3394
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.7159346463882826e-07,
+      "loss": 2.2133,
+      "step": 3396
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.7059007329407326e-07,
+      "loss": 2.3595,
+      "step": 3398
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.6958824878827635e-07,
+      "loss": 2.3127,
+      "step": 3400
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.685879932734476e-07,
+      "loss": 2.3173,
+      "step": 3402
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.675893088982275e-07,
+      "loss": 2.1802,
+      "step": 3404
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.665921978078809e-07,
+      "loss": 2.3848,
+      "step": 3406
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.6559666214429297e-07,
+      "loss": 2.3093,
+      "step": 3408
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.646027040459651e-07,
+      "loss": 2.1086,
+      "step": 3410
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.636103256480102e-07,
+      "loss": 2.4374,
+      "step": 3412
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.626195290821467e-07,
+      "loss": 2.2971,
+      "step": 3414
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.616303164766965e-07,
+      "loss": 2.42,
+      "step": 3416
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6064268995657823e-07,
+      "loss": 2.3013,
+      "step": 3418
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5965665164330354e-07,
+      "loss": 2.3163,
+      "step": 3420
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5867220365497254e-07,
+      "loss": 2.376,
+      "step": 3422
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5768934810626885e-07,
+      "loss": 2.3108,
+      "step": 3424
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5670808710845624e-07,
+      "loss": 2.4786,
+      "step": 3426
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.557284227693727e-07,
+      "loss": 2.1118,
+      "step": 3428
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.5475035719342606e-07,
+      "loss": 2.444,
+      "step": 3430
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.5377389248159033e-07,
+      "loss": 2.1952,
+      "step": 3432
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.527990307314013e-07,
+      "loss": 2.283,
+      "step": 3434
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.518257740369497e-07,
+      "loss": 2.3963,
+      "step": 3436
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.5085412448888045e-07,
+      "loss": 2.1796,
+      "step": 3438
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.49884084174385e-07,
+      "loss": 2.3689,
+      "step": 3440
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.489156551771981e-07,
+      "loss": 2.4479,
+      "step": 3442
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.479488395775935e-07,
+      "loss": 2.2417,
+      "step": 3444
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.469836394523794e-07,
+      "loss": 2.4555,
+      "step": 3446
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.460200568748936e-07,
+      "loss": 2.7688,
+      "step": 3448
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.45058093914999e-07,
+      "loss": 2.7413,
+      "step": 3450
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.440977526390795e-07,
+      "loss": 2.3979,
+      "step": 3452
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.431390351100364e-07,
+      "loss": 2.3415,
+      "step": 3454
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.421819433872813e-07,
+      "loss": 2.6412,
+      "step": 3456
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.4122647952673504e-07,
+      "loss": 2.4298,
+      "step": 3458
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.4027264558082085e-07,
+      "loss": 2.186,
+      "step": 3460
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.393204435984608e-07,
+      "loss": 2.3957,
+      "step": 3462
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3836987562507138e-07,
+      "loss": 2.5346,
+      "step": 3464
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3742094370255972e-07,
+      "loss": 2.325,
+      "step": 3466
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.364736498693175e-07,
+      "loss": 2.4938,
+      "step": 3468
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3552799616021845e-07,
+      "loss": 2.4997,
+      "step": 3470
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3458398460661233e-07,
+      "loss": 2.3212,
+      "step": 3472
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3364161723632302e-07,
+      "loss": 2.3675,
+      "step": 3474
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3270089607364018e-07,
+      "loss": 2.1652,
+      "step": 3476
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3176182313931948e-07,
+      "loss": 2.4536,
+      "step": 3478
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.308244004505745e-07,
+      "loss": 2.2274,
+      "step": 3480
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.2988863002107538e-07,
+      "loss": 2.4275,
+      "step": 3482
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.2895451386094124e-07,
+      "loss": 2.1462,
+      "step": 3484
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.2802205397673935e-07,
+      "loss": 2.4634,
+      "step": 3486
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.270912523714782e-07,
+      "loss": 2.435,
+      "step": 3488
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.2616211104460459e-07,
+      "loss": 2.4373,
+      "step": 3490
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.2523463199199822e-07,
+      "loss": 2.3186,
+      "step": 3492
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.2430881720596927e-07,
+      "loss": 2.44,
+      "step": 3494
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.2338466867525196e-07,
+      "loss": 2.4308,
+      "step": 3496
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.2246218838500143e-07,
+      "loss": 2.2673,
+      "step": 3498
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.2154137831678932e-07,
+      "loss": 2.2352,
+      "step": 3500
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.2062224044859966e-07,
+      "loss": 2.202,
+      "step": 3502
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.1970477675482423e-07,
+      "loss": 2.2827,
+      "step": 3504
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.1878898920625832e-07,
+      "loss": 2.1762,
+      "step": 3506
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.1787487977009765e-07,
+      "loss": 2.474,
+      "step": 3508
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.169624504099321e-07,
+      "loss": 2.2976,
+      "step": 3510
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.1605170308574318e-07,
+      "loss": 2.1453,
+      "step": 3512
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.1514263975389879e-07,
+      "loss": 2.1924,
+      "step": 3514
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.1423526236715052e-07,
+      "loss": 2.2609,
+      "step": 3516
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.1378220651243138e-07,
+      "loss": 2.4962,
+      "step": 3518
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.128773616968108e-07,
+      "loss": 2.186,
+      "step": 3520
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.1197420769230912e-07,
+      "loss": 2.2364,
+      "step": 3522
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.110727464389843e-07,
+      "loss": 2.5511,
+      "step": 3524
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.101729798732561e-07,
+      "loss": 2.1022,
+      "step": 3526
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0927490992790587e-07,
+      "loss": 2.501,
+      "step": 3528
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0837853853206934e-07,
+      "loss": 2.3158,
+      "step": 3530
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.0748386761123404e-07,
+      "loss": 2.3638,
+      "step": 3532
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.0659089908723415e-07,
+      "loss": 2.2822,
+      "step": 3534
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.0614505382362125e-07,
+      "loss": 2.4747,
+      "step": 3536
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.0525464249031454e-07,
+      "loss": 2.1788,
+      "step": 3538
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.043659383424222e-07,
+      "loss": 2.0549,
+      "step": 3540
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.0347894328896177e-07,
+      "loss": 2.1212,
+      "step": 3542
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.0259365923527894e-07,
+      "loss": 2.4062,
+      "step": 3544
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.0171008808304547e-07,
+      "loss": 2.3898,
+      "step": 3546
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.0082823173025142e-07,
+      "loss": 2.2393,
+      "step": 3548
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.9994809207120556e-07,
+      "loss": 2.1458,
+      "step": 3550
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.9906967099652771e-07,
+      "loss": 2.3189,
+      "step": 3552
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.981929703931463e-07,
+      "loss": 2.1485,
+      "step": 3554
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.973179921442938e-07,
+      "loss": 2.2943,
+      "step": 3556
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.964447381295038e-07,
+      "loss": 2.2134,
+      "step": 3558
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.9557321022460505e-07,
+      "loss": 2.2823,
+      "step": 3560
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.9470341030171898e-07,
+      "loss": 2.5778,
+      "step": 3562
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.9383534022925484e-07,
+      "loss": 2.3283,
+      "step": 3564
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.929690018719069e-07,
+      "loss": 2.0778,
+      "step": 3566
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.9210439709064785e-07,
+      "loss": 2.3643,
+      "step": 3568
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.9124152774272827e-07,
+      "loss": 2.1915,
+      "step": 3570
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.9038039568166974e-07,
+      "loss": 2.372,
+      "step": 3572
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.8952100275726268e-07,
+      "loss": 2.1317,
+      "step": 3574
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.8866335081556072e-07,
+      "loss": 2.1938,
+      "step": 3576
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.8780744169887918e-07,
+      "loss": 2.3531,
+      "step": 3578
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.8695327724578836e-07,
+      "loss": 2.0558,
+      "step": 3580
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.861008592911113e-07,
+      "loss": 2.1199,
+      "step": 3582
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.8525018966591953e-07,
+      "loss": 2.1674,
+      "step": 3584
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.844012701975286e-07,
+      "loss": 2.5179,
+      "step": 3586
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.8355410270949545e-07,
+      "loss": 2.2588,
+      "step": 3588
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.8270868902161219e-07,
+      "loss": 2.4867,
+      "step": 3590
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.8186503094990512e-07,
+      "loss": 2.5123,
+      "step": 3592
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.810231303066283e-07,
+      "loss": 2.067,
+      "step": 3594
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.8018298890026106e-07,
+      "loss": 2.1489,
+      "step": 3596
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.7934460853550336e-07,
+      "loss": 2.4436,
+      "step": 3598
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.7850799101327318e-07,
+      "loss": 2.6137,
+      "step": 3600
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.7767313813070084e-07,
+      "loss": 2.4861,
+      "step": 3602
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.7684005168112647e-07,
+      "loss": 2.2893,
+      "step": 3604
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.7600873345409517e-07,
+      "loss": 2.3534,
+      "step": 3606
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.7517918523535514e-07,
+      "loss": 2.479,
+      "step": 3608
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.743514088068504e-07,
+      "loss": 2.2013,
+      "step": 3610
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.7352540594672105e-07,
+      "loss": 2.4611,
+      "step": 3612
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.7270117842929577e-07,
+      "loss": 2.2597,
+      "step": 3614
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.7187872802509117e-07,
+      "loss": 2.5705,
+      "step": 3616
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.7105805650080462e-07,
+      "loss": 2.1431,
+      "step": 3618
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.7023916561931405e-07,
+      "loss": 2.3859,
+      "step": 3620
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.6942205713967138e-07,
+      "loss": 2.4055,
+      "step": 3622
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.6860673281710013e-07,
+      "loss": 2.2329,
+      "step": 3624
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.677931944029909e-07,
+      "loss": 2.5085,
+      "step": 3626
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.6698144364489875e-07,
+      "loss": 2.3044,
+      "step": 3628
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.661714822865381e-07,
+      "loss": 2.5598,
+      "step": 3630
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.6536331206777965e-07,
+      "loss": 2.4912,
+      "step": 3632
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.6455693472464638e-07,
+      "loss": 2.1866,
+      "step": 3634
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.6375235198931113e-07,
+      "loss": 2.245,
+      "step": 3636
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.6294956559008976e-07,
+      "loss": 2.3741,
+      "step": 3638
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.6214857725144137e-07,
+      "loss": 2.4793,
+      "step": 3640
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.6134938869396163e-07,
+      "loss": 2.3812,
+      "step": 3642
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.6055200163438032e-07,
+      "loss": 2.203,
+      "step": 3644
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.5975641778555738e-07,
+      "loss": 2.4009,
+      "step": 3646
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.5896263885647964e-07,
+      "loss": 2.2717,
+      "step": 3648
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.581706665522564e-07,
+      "loss": 2.3982,
+      "step": 3650
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.5738050257411628e-07,
+      "loss": 2.2986,
+      "step": 3652
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.5659214861940352e-07,
+      "loss": 2.3759,
+      "step": 3654
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.5580560638157425e-07,
+      "loss": 2.7631,
+      "step": 3656
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.5502087755019245e-07,
+      "loss": 2.4175,
+      "step": 3658
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.542379638109278e-07,
+      "loss": 2.4338,
+      "step": 3660
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.5345686684554994e-07,
+      "loss": 2.4166,
+      "step": 3662
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.5267758833192625e-07,
+      "loss": 2.1653,
+      "step": 3664
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.5190012994401814e-07,
+      "loss": 2.2565,
+      "step": 3666
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.511244933518766e-07,
+      "loss": 2.2848,
+      "step": 3668
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.5035068022164033e-07,
+      "loss": 2.4534,
+      "step": 3670
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4957869221552998e-07,
+      "loss": 2.4818,
+      "step": 3672
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4880853099184631e-07,
+      "loss": 2.5356,
+      "step": 3674
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.480401982049655e-07,
+      "loss": 2.2729,
+      "step": 3676
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4727369550533687e-07,
+      "loss": 2.3927,
+      "step": 3678
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4650902453947734e-07,
+      "loss": 2.2466,
+      "step": 3680
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4574618694997032e-07,
+      "loss": 2.5247,
+      "step": 3682
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.4498518437546025e-07,
+      "loss": 2.3951,
+      "step": 3684
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4422601845065009e-07,
+      "loss": 2.2343,
+      "step": 3686
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.4346869080629698e-07,
+      "loss": 2.3271,
+      "step": 3688
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.427132030692102e-07,
+      "loss": 2.4791,
+      "step": 3690
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.4195955686224604e-07,
+      "loss": 2.3011,
+      "step": 3692
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4120775380430517e-07,
+      "loss": 2.5667,
+      "step": 3694
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.4045779551032877e-07,
+      "loss": 2.2304,
+      "step": 3696
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.3970968359129632e-07,
+      "loss": 2.2826,
+      "step": 3698
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.3896341965421943e-07,
+      "loss": 2.5231,
+      "step": 3700
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.382190053021416e-07,
+      "loss": 2.0304,
+      "step": 3702
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.3747644213413224e-07,
+      "loss": 2.4519,
+      "step": 3704
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.367357317452854e-07,
+      "loss": 2.3202,
+      "step": 3706
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.3599687572671358e-07,
+      "loss": 2.3184,
+      "step": 3708
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.352598756655474e-07,
+      "loss": 2.0781,
+      "step": 3710
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.3452473314493007e-07,
+      "loss": 2.3513,
+      "step": 3712
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.3379144974401436e-07,
+      "loss": 2.2974,
+      "step": 3714
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.3306002703795983e-07,
+      "loss": 2.2226,
+      "step": 3716
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.3233046659792947e-07,
+      "loss": 2.2723,
+      "step": 3718
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.3160276999108533e-07,
+      "loss": 2.3053,
+      "step": 3720
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.3087693878058591e-07,
+      "loss": 2.4518,
+      "step": 3722
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.3015297452558283e-07,
+      "loss": 2.2635,
+      "step": 3724
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.2943087878121727e-07,
+      "loss": 2.2041,
+      "step": 3726
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.287106530986165e-07,
+      "loss": 2.196,
+      "step": 3728
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2799229902489117e-07,
+      "loss": 2.3408,
+      "step": 3730
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2727581810313114e-07,
+      "loss": 2.5223,
+      "step": 3732
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2656121187240266e-07,
+      "loss": 2.4009,
+      "step": 3734
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.25848481867745e-07,
+      "loss": 2.4088,
+      "step": 3736
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2513762962016695e-07,
+      "loss": 2.3094,
+      "step": 3738
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2442865665664414e-07,
+      "loss": 2.354,
+      "step": 3740
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2372156450011518e-07,
+      "loss": 2.5011,
+      "step": 3742
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2301635466947812e-07,
+      "loss": 2.5396,
+      "step": 3744
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2231302867958805e-07,
+      "loss": 2.2474,
+      "step": 3746
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2161158804125316e-07,
+      "loss": 2.141,
+      "step": 3748
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2091203426123175e-07,
+      "loss": 1.9833,
+      "step": 3750
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.2021436884222946e-07,
+      "loss": 2.4734,
+      "step": 3752
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.1951859328289493e-07,
+      "loss": 2.401,
+      "step": 3754
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.1882470907781727e-07,
+      "loss": 2.3442,
+      "step": 3756
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.1813271771752298e-07,
+      "loss": 2.3096,
+      "step": 3758
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.17442620688473e-07,
+      "loss": 2.3408,
+      "step": 3760
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.1675441947305787e-07,
+      "loss": 2.3648,
+      "step": 3762
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.1606811554959706e-07,
+      "loss": 2.0118,
+      "step": 3764
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.1538371039233341e-07,
+      "loss": 2.5944,
+      "step": 3766
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.1470120547143236e-07,
+      "loss": 2.4937,
+      "step": 3768
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.1402060225297561e-07,
+      "loss": 2.3807,
+      "step": 3770
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.1334190219896155e-07,
+      "loss": 2.5853,
+      "step": 3772
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.1266510676729956e-07,
+      "loss": 2.4372,
+      "step": 3774
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.1199021741180781e-07,
+      "loss": 2.3959,
+      "step": 3776
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.1131723558220996e-07,
+      "loss": 2.3788,
+      "step": 3778
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.1064616272413262e-07,
+      "loss": 2.4389,
+      "step": 3780
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.0997700027910128e-07,
+      "loss": 2.1827,
+      "step": 3782
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.0930974968453777e-07,
+      "loss": 2.3847,
+      "step": 3784
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.0864441237375698e-07,
+      "loss": 2.2319,
+      "step": 3786
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.0798098977596448e-07,
+      "loss": 2.1224,
+      "step": 3788
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.0731948331625195e-07,
+      "loss": 2.3572,
+      "step": 3790
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.0665989441559586e-07,
+      "loss": 2.2529,
+      "step": 3792
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.0600222449085294e-07,
+      "loss": 2.4526,
+      "step": 3794
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.0534647495475857e-07,
+      "loss": 2.1667,
+      "step": 3796
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.0469264721592197e-07,
+      "loss": 2.6476,
+      "step": 3798
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.0404074267882502e-07,
+      "loss": 2.3072,
+      "step": 3800
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.0339076274381787e-07,
+      "loss": 2.5948,
+      "step": 3802
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.0274270880711666e-07,
+      "loss": 2.2217,
+      "step": 3804
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.0209658226080032e-07,
+      "loss": 2.5243,
+      "step": 3806
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.0145238449280724e-07,
+      "loss": 2.3269,
+      "step": 3808
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.0081011688693341e-07,
+      "loss": 2.2734,
+      "step": 3810
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.0016978082282779e-07,
+      "loss": 2.1973,
+      "step": 3812
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.953137767599073e-08,
+      "loss": 2.1697,
+      "step": 3814
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.889490881777018e-08,
+      "loss": 2.2464,
+      "step": 3816
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.826037561535938e-08,
+      "loss": 2.4254,
+      "step": 3818
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.762777943179312e-08,
+      "loss": 2.3955,
+      "step": 3820
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.699712162594609e-08,
+      "loss": 2.1837,
+      "step": 3822
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 9.63684035525284e-08,
+      "loss": 2.0977,
+      "step": 3824
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 9.574162656208384e-08,
+      "loss": 2.4142,
+      "step": 3826
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.511679200098632e-08,
+      "loss": 2.2772,
+      "step": 3828
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.449390121143774e-08,
+      "loss": 2.24,
+      "step": 3830
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 9.387295553146379e-08,
+      "loss": 2.444,
+      "step": 3832
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 9.325395629491262e-08,
+      "loss": 2.5115,
+      "step": 3834
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 9.263690483145103e-08,
+      "loss": 2.2708,
+      "step": 3836
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 9.202180246656155e-08,
+      "loss": 2.465,
+      "step": 3838
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 9.140865052154012e-08,
+      "loss": 2.9963,
+      "step": 3840
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 9.079745031349317e-08,
+      "loss": 2.159,
+      "step": 3842
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 9.018820315533426e-08,
+      "loss": 2.5147,
+      "step": 3844
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.958091035578174e-08,
+      "loss": 2.3305,
+      "step": 3846
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.89755732193559e-08,
+      "loss": 2.3214,
+      "step": 3848
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.837219304637633e-08,
+      "loss": 2.3948,
+      "step": 3850
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.777077113295794e-08,
+      "loss": 2.428,
+      "step": 3852
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.717130877101031e-08,
+      "loss": 2.279,
+      "step": 3854
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.657380724823294e-08,
+      "loss": 2.2065,
+      "step": 3856
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 8.597826784811402e-08,
+      "loss": 2.5204,
+      "step": 3858
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 8.538469184992559e-08,
+      "loss": 2.1397,
+      "step": 3860
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.479308052872359e-08,
+      "loss": 2.4221,
+      "step": 3862
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.42034351553429e-08,
+      "loss": 2.1741,
+      "step": 3864
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.361575699639534e-08,
+      "loss": 2.199,
+      "step": 3866
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 8.30300473142671e-08,
+      "loss": 2.5473,
+      "step": 3868
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 8.244630736711621e-08,
+      "loss": 2.536,
+      "step": 3870
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 8.186453840886897e-08,
+      "loss": 2.2939,
+      "step": 3872
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 8.128474168921828e-08,
+      "loss": 2.2578,
+      "step": 3874
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 8.070691845361999e-08,
+      "loss": 2.4713,
+      "step": 3876
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 8.013106994329155e-08,
+      "loss": 2.2558,
+      "step": 3878
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.95571973952075e-08,
+      "loss": 2.4457,
+      "step": 3880
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.898530204209864e-08,
+      "loss": 2.419,
+      "step": 3882
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.841538511244816e-08,
+      "loss": 2.0561,
+      "step": 3884
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.784744783048958e-08,
+      "loss": 2.2682,
+      "step": 3886
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.7281491416204e-08,
+      "loss": 2.4426,
+      "step": 3888
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.671751708531692e-08,
+      "loss": 2.0675,
+      "step": 3890
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.61555260492972e-08,
+      "loss": 2.0942,
+      "step": 3892
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.559551951535237e-08,
+      "loss": 2.2835,
+      "step": 3894
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.503749868642762e-08,
+      "loss": 2.2816,
+      "step": 3896
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.448146476120232e-08,
+      "loss": 2.3509,
+      "step": 3898
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.392741893408838e-08,
+      "loss": 2.3625,
+      "step": 3900
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.337536239522623e-08,
+      "loss": 2.3323,
+      "step": 3902
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.282529633048385e-08,
+      "loss": 2.3537,
+      "step": 3904
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.227722192145325e-08,
+      "loss": 2.3887,
+      "step": 3906
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.173114034544825e-08,
+      "loss": 2.1123,
+      "step": 3908
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.118705277550142e-08,
+      "loss": 2.197,
+      "step": 3910
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.064496038036293e-08,
+      "loss": 2.387,
+      "step": 3912
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.010486432449636e-08,
+      "loss": 2.4442,
+      "step": 3914
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.95667657680773e-08,
+      "loss": 2.4656,
+      "step": 3916
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.903066586699014e-08,
+      "loss": 2.1219,
+      "step": 3918
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.849656577282703e-08,
+      "loss": 2.4213,
+      "step": 3920
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.79644666328828e-08,
+      "loss": 2.1535,
+      "step": 3922
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.743436959015547e-08,
+      "loss": 2.6134,
+      "step": 3924
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.69062757833414e-08,
+      "loss": 2.198,
+      "step": 3926
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.63801863468344e-08,
+      "loss": 2.2652,
+      "step": 3928
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.585610241072226e-08,
+      "loss": 2.3245,
+      "step": 3930
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.533402510078533e-08,
+      "loss": 2.1564,
+      "step": 3932
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.4813955538493e-08,
+      "loss": 2.2783,
+      "step": 3934
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.429589484100217e-08,
+      "loss": 2.4365,
+      "step": 3936
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 6.377984412115423e-08,
+      "loss": 2.4163,
+      "step": 3938
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 6.326580448747343e-08,
+      "loss": 2.1423,
+      "step": 3940
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 6.27537770441633e-08,
+      "loss": 2.2192,
+      "step": 3942
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 6.224376289110589e-08,
+      "loss": 2.2585,
+      "step": 3944
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 6.173576312385765e-08,
+      "loss": 2.2334,
+      "step": 3946
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 6.122977883364877e-08,
+      "loss": 2.2312,
+      "step": 3948
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 6.072581110737907e-08,
+      "loss": 2.5184,
+      "step": 3950
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 6.022386102761756e-08,
+      "loss": 2.5329,
+      "step": 3952
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.972392967259865e-08,
+      "loss": 2.418,
+      "step": 3954
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.92260181162203e-08,
+      "loss": 2.2043,
+      "step": 3956
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.873012742804173e-08,
+      "loss": 2.469,
+      "step": 3958
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.823625867328175e-08,
+      "loss": 2.3175,
+      "step": 3960
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.774441291281518e-08,
+      "loss": 2.5495,
+      "step": 3962
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.725459120317156e-08,
+      "loss": 2.2427,
+      "step": 3964
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.676679459653233e-08,
+      "loss": 2.5635,
+      "step": 3966
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.628102414072933e-08,
+      "loss": 2.2622,
+      "step": 3968
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.579728087924162e-08,
+      "loss": 2.3046,
+      "step": 3970
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.531556585119357e-08,
+      "loss": 2.1263,
+      "step": 3972
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.4835880091353314e-08,
+      "loss": 2.1675,
+      "step": 3974
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.4358224630129404e-08,
+      "loss": 2.579,
+      "step": 3976
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.388260049356919e-08,
+      "loss": 2.4221,
+      "step": 3978
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.3409008703356626e-08,
+      "loss": 2.3077,
+      "step": 3980
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.293745027681029e-08,
+      "loss": 2.4017,
+      "step": 3982
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.246792622688023e-08,
+      "loss": 2.1001,
+      "step": 3984
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2000437562147225e-08,
+      "loss": 2.3728,
+      "step": 3986
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.15349852868191e-08,
+      "loss": 2.3596,
+      "step": 3988
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.1071570400730405e-08,
+      "loss": 2.3265,
+      "step": 3990
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.061019389933774e-08,
+      "loss": 2.3075,
+      "step": 3992
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.015085677372044e-08,
+      "loss": 2.2687,
+      "step": 3994
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.969356001057612e-08,
+      "loss": 2.3342,
+      "step": 3996
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.9238304592220117e-08,
+      "loss": 2.2943,
+      "step": 3998
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.8785091496582385e-08,
+      "loss": 2.1758,
+      "step": 4000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.833392169720607e-08,
+      "loss": 2.2717,
+      "step": 4002
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.788479616324481e-08,
+      "loss": 2.3099,
+      "step": 4004
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.743771585946144e-08,
+      "loss": 2.584,
+      "step": 4006
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.699268174622473e-08,
+      "loss": 2.16,
+      "step": 4008
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.6549694779509094e-08,
+      "loss": 2.6906,
+      "step": 4010
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.610875591089025e-08,
+      "loss": 2.1513,
+      "step": 4012
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.566986608754553e-08,
+      "loss": 2.2294,
+      "step": 4014
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5233026252250005e-08,
+      "loss": 2.1508,
+      "step": 4016
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.479823734337551e-08,
+      "loss": 2.2828,
+      "step": 4018
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.4365500294888056e-08,
+      "loss": 2.3614,
+      "step": 4020
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.3934816036346525e-08,
+      "loss": 2.3381,
+      "step": 4022
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.350618549289986e-08,
+      "loss": 1.9958,
+      "step": 4024
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.307960958528534e-08,
+      "loss": 2.2663,
+      "step": 4026
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.265508922982686e-08,
+      "loss": 2.0485,
+      "step": 4028
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.223262533843319e-08,
+      "loss": 2.4741,
+      "step": 4030
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.1812218818594757e-08,
+      "loss": 2.2407,
+      "step": 4032
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.1393870573383394e-08,
+      "loss": 2.2279,
+      "step": 4034
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.097758150144903e-08,
+      "loss": 2.5935,
+      "step": 4036
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.05633524970187e-08,
+      "loss": 2.4199,
+      "step": 4038
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.015118444989374e-08,
+      "loss": 2.3565,
+      "step": 4040
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.974107824544892e-08,
+      "loss": 2.7166,
+      "step": 4042
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.9333034764629793e-08,
+      "loss": 2.3456,
+      "step": 4044
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.892705488395065e-08,
+      "loss": 2.4363,
+      "step": 4046
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.852313947549335e-08,
+      "loss": 2.2772,
+      "step": 4048
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.812128940690496e-08,
+      "loss": 2.4964,
+      "step": 4050
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.7721505541396305e-08,
+      "loss": 2.3199,
+      "step": 4052
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.7323788737739005e-08,
+      "loss": 2.2969,
+      "step": 4054
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.6928139850265436e-08,
+      "loss": 2.1398,
+      "step": 4056
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.6534559728865324e-08,
+      "loss": 2.2636,
+      "step": 4058
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.6143049218984586e-08,
+      "loss": 2.3242,
+      "step": 4060
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.575360916162329e-08,
+      "loss": 2.4013,
+      "step": 4062
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.536624039333447e-08,
+      "loss": 2.0722,
+      "step": 4064
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4980943746221295e-08,
+      "loss": 2.2253,
+      "step": 4066
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.459772004793615e-08,
+      "loss": 2.5274,
+      "step": 4068
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.421657012167834e-08,
+      "loss": 2.185,
+      "step": 4070
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.383749478619291e-08,
+      "loss": 2.4211,
+      "step": 4072
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.346049485576774e-08,
+      "loss": 2.2866,
+      "step": 4074
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.308557114023347e-08,
+      "loss": 2.4034,
+      "step": 4076
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.271272444495998e-08,
+      "loss": 2.2678,
+      "step": 4078
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.2341955570856506e-08,
+      "loss": 2.1473,
+      "step": 4080
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.197326531436773e-08,
+      "loss": 1.9607,
+      "step": 4082
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.1606654467474236e-08,
+      "loss": 2.3081,
+      "step": 4084
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.124212381768942e-08,
+      "loss": 2.3196,
+      "step": 4086
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.087967414805848e-08,
+      "loss": 2.2704,
+      "step": 4088
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.051930623715604e-08,
+      "loss": 2.2809,
+      "step": 4090
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.016102085908534e-08,
+      "loss": 2.5148,
+      "step": 4092
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.9804818783476184e-08,
+      "loss": 2.3634,
+      "step": 4094
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.945070077548284e-08,
+      "loss": 2.2721,
+      "step": 4096
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.9098667595782945e-08,
+      "loss": 2.5452,
+      "step": 4098
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.8748720000576265e-08,
+      "loss": 2.4443,
+      "step": 4100
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.8400858741581602e-08,
+      "loss": 2.5627,
+      "step": 4102
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.805508456603689e-08,
+      "loss": 2.5705,
+      "step": 4104
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.7711398216696658e-08,
+      "loss": 2.1941,
+      "step": 4106
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.7369800431830236e-08,
+      "loss": 2.3704,
+      "step": 4108
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.7030291945220885e-08,
+      "loss": 2.5346,
+      "step": 4110
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.669287348616378e-08,
+      "loss": 2.0938,
+      "step": 4112
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.6357545779464584e-08,
+      "loss": 2.6161,
+      "step": 4114
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.602430954543755e-08,
+      "loss": 2.261,
+      "step": 4116
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.5693165499904635e-08,
+      "loss": 2.2991,
+      "step": 4118
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.5364114354193277e-08,
+      "loss": 1.9709,
+      "step": 4120
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.5037156815135408e-08,
+      "loss": 2.7381,
+      "step": 4122
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.4712293585065546e-08,
+      "loss": 2.0963,
+      "step": 4124
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.4389525361819487e-08,
+      "loss": 2.1209,
+      "step": 4126
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.4068852838732945e-08,
+      "loss": 2.2181,
+      "step": 4128
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.3750276704639472e-08,
+      "loss": 2.2637,
+      "step": 4130
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.3433797643869658e-08,
+      "loss": 2.1917,
+      "step": 4132
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.3119416336249588e-08,
+      "loss": 2.1542,
+      "step": 4134
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.2807133457098504e-08,
+      "loss": 2.492,
+      "step": 4136
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.2496949677228927e-08,
+      "loss": 2.3315,
+      "step": 4138
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.2188865662943536e-08,
+      "loss": 2.2493,
+      "step": 4140
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.188288207603517e-08,
+      "loss": 2.5447,
+      "step": 4142
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.1578999573784063e-08,
+      "loss": 2.3446,
+      "step": 4144
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.127721880895783e-08,
+      "loss": 2.1998,
+      "step": 4146
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0977540429808926e-08,
+      "loss": 2.0618,
+      "step": 4148
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.067996508007386e-08,
+      "loss": 2.1654,
+      "step": 4150
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0384493398971303e-08,
+      "loss": 2.1868,
+      "step": 4152
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.0091126021201775e-08,
+      "loss": 2.3577,
+      "step": 4154
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9799863576944853e-08,
+      "loss": 2.1735,
+      "step": 4156
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9510706691858835e-08,
+      "loss": 2.4039,
+      "step": 4158
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.922365598707909e-08,
+      "loss": 2.4402,
+      "step": 4160
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.893871207921671e-08,
+      "loss": 2.2745,
+      "step": 4162
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8655875580356974e-08,
+      "loss": 2.5174,
+      "step": 4164
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8375147098058653e-08,
+      "loss": 2.7251,
+      "step": 4166
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8096527235351934e-08,
+      "loss": 2.0941,
+      "step": 4168
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7820016590737842e-08,
+      "loss": 2.4482,
+      "step": 4170
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.754561575818625e-08,
+      "loss": 2.3424,
+      "step": 4172
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7273325327135215e-08,
+      "loss": 2.3049,
+      "step": 4174
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.700314588248952e-08,
+      "loss": 2.5217,
+      "step": 4176
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.673507800461893e-08,
+      "loss": 2.3833,
+      "step": 4178
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6469122269357817e-08,
+      "loss": 2.323,
+      "step": 4180
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6205279248003413e-08,
+      "loss": 2.3853,
+      "step": 4182
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.5943549507314468e-08,
+      "loss": 2.3385,
+      "step": 4184
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.568393360951026e-08,
+      "loss": 2.1541,
+      "step": 4186
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.542643211226946e-08,
+      "loss": 2.5234,
+      "step": 4188
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5171045568728723e-08,
+      "loss": 1.9526,
+      "step": 4190
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.4917774527481442e-08,
+      "loss": 2.4961,
+      "step": 4192
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.4666619532577196e-08,
+      "loss": 2.3186,
+      "step": 4194
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.441758112351954e-08,
+      "loss": 2.2356,
+      "step": 4196
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.4170659835265774e-08,
+      "loss": 2.155,
+      "step": 4198
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3925856198225283e-08,
+      "loss": 2.2409,
+      "step": 4200
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3683170738258532e-08,
+      "loss": 2.4088,
+      "step": 4202
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.344260397667607e-08,
+      "loss": 2.0696,
+      "step": 4204
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.32041564302372e-08,
+      "loss": 2.1969,
+      "step": 4206
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.2967828611148868e-08,
+      "loss": 2.1321,
+      "step": 4208
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.2733621027064989e-08,
+      "loss": 2.1224,
+      "step": 4210
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.2501534181084461e-08,
+      "loss": 2.6715,
+      "step": 4212
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.2271568571751157e-08,
+      "loss": 2.4569,
+      "step": 4214
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.204372469305226e-08,
+      "loss": 2.4008,
+      "step": 4216
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1818003034416935e-08,
+      "loss": 2.3672,
+      "step": 4218
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1594404080716103e-08,
+      "loss": 2.4229,
+      "step": 4220
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1372928312260333e-08,
+      "loss": 2.4505,
+      "step": 4222
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1153576204800285e-08,
+      "loss": 2.4657,
+      "step": 4224
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.0936348229523717e-08,
+      "loss": 2.6675,
+      "step": 4226
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.0721244853056366e-08,
+      "loss": 2.1615,
+      "step": 4228
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.0508266537459843e-08,
+      "loss": 2.3642,
+      "step": 4230
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.0297413740231076e-08,
+      "loss": 2.2433,
+      "step": 4232
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.008868691430087e-08,
+      "loss": 2.5839,
+      "step": 4234
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.882086508033571e-09,
+      "loss": 2.6095,
+      "step": 4236
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.677612965225734e-09,
+      "loss": 2.4396,
+      "step": 4238
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.475266725105014e-09,
+      "loss": 2.5627,
+      "step": 4240
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.275048222329607e-09,
+      "loss": 2.188,
+      "step": 4242
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 9.076957886987146e-09,
+      "loss": 2.2413,
+      "step": 4244
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 8.88099614459381e-09,
+      "loss": 2.3452,
+      "step": 4246
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 8.68716341609299e-09,
+      "loss": 2.3178,
+      "step": 4248
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 8.49546011785518e-09,
+      "loss": 2.4892,
+      "step": 4250
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 8.305886661676752e-09,
+      "loss": 2.4523,
+      "step": 4252
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 8.118443454778302e-09,
+      "loss": 2.4388,
+      "step": 4254
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 7.933130899805296e-09,
+      "loss": 2.1198,
+      "step": 4256
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 7.74994939482576e-09,
+      "loss": 2.5573,
+      "step": 4258
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 7.56889933333038e-09,
+      "loss": 2.2659,
+      "step": 4260
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 7.389981104231169e-09,
+      "loss": 2.1934,
+      "step": 4262
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 7.213195091860691e-09,
+      "loss": 2.2468,
+      "step": 4264
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 7.038541675971399e-09,
+      "loss": 2.3865,
+      "step": 4266
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 6.866021231734853e-09,
+      "loss": 2.3968,
+      "step": 4268
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 6.695634129740613e-09,
+      "loss": 2.3712,
+      "step": 4270
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 6.52738073599568e-09,
+      "loss": 2.2353,
+      "step": 4272
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 6.361261411923724e-09,
+      "loss": 2.5092,
+      "step": 4274
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 6.19727651436408e-09,
+      "loss": 2.2469,
+      "step": 4276
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 6.035426395571419e-09,
+      "loss": 2.1285,
+      "step": 4278
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.875711403214412e-09,
+      "loss": 2.2678,
+      "step": 4280
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.718131880375398e-09,
+      "loss": 2.3838,
+      "step": 4282
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.5626881655495006e-09,
+      "loss": 2.2587,
+      "step": 4284
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.4093805926440635e-09,
+      "loss": 2.3756,
+      "step": 4286
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.258209490977772e-09,
+      "loss": 2.2056,
+      "step": 4288
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.109175185279757e-09,
+      "loss": 2.1925,
+      "step": 4290
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.96227799568949e-09,
+      "loss": 2.3737,
+      "step": 4292
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.8175182377553355e-09,
+      "loss": 2.2732,
+      "step": 4294
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.674896222434555e-09,
+      "loss": 2.326,
+      "step": 4296
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5344122560920795e-09,
+      "loss": 2.5249,
+      "step": 4298
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.396066640500406e-09,
+      "loss": 2.2555,
+      "step": 4300
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.259859672838484e-09,
+      "loss": 2.5123,
+      "step": 4302
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.125791645691268e-09,
+      "loss": 2.1902,
+      "step": 4304
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.993862847049167e-09,
+      "loss": 2.2976,
+      "step": 4306
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.864073560307046e-09,
+      "loss": 2.2722,
+      "step": 4308
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.736424064264443e-09,
+      "loss": 2.1837,
+      "step": 4310
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.610914633123685e-09,
+      "loss": 2.5105,
+      "step": 4312
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.487545536490888e-09,
+      "loss": 2.0634,
+      "step": 4314
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.366317039373845e-09,
+      "loss": 2.4091,
+      "step": 4316
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.247229402182472e-09,
+      "loss": 2.4849,
+      "step": 4318
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.130282880727808e-09,
+      "loss": 2.7144,
+      "step": 4320
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.0154777262217934e-09,
+      "loss": 2.5959,
+      "step": 4322
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.902814185276159e-09,
+      "loss": 2.1911,
+      "step": 4324
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.7922924999025375e-09,
+      "loss": 2.4163,
+      "step": 4326
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.6839129075116873e-09,
+      "loss": 2.195,
+      "step": 4328
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.577675640912602e-09,
+      "loss": 2.1792,
+      "step": 4330
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.4735809283128463e-09,
+      "loss": 2.3718,
+      "step": 4332
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.3716289933172208e-09,
+      "loss": 2.4066,
+      "step": 4334
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.2718200549277644e-09,
+      "loss": 2.2239,
+      "step": 4336
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.1741543275431983e-09,
+      "loss": 2.3306,
+      "step": 4338
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.0786320209584817e-09,
+      "loss": 2.4882,
+      "step": 4340
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.9852533403640347e-09,
+      "loss": 2.5039,
+      "step": 4342
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.894018486345961e-09,
+      "loss": 2.3346,
+      "step": 4344
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.8049276548848247e-09,
+      "loss": 2.5675,
+      "step": 4346
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7179810373560977e-09,
+      "loss": 2.2032,
+      "step": 4348
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.6331788205288245e-09,
+      "loss": 1.9619,
+      "step": 4350
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.5505211865660673e-09,
+      "loss": 2.184,
+      "step": 4352
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.470008313023796e-09,
+      "loss": 2.4945,
+      "step": 4354
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3916403728509995e-09,
+      "loss": 2.3055,
+      "step": 4356
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.3154175343893514e-09,
+      "loss": 2.3081,
+      "step": 4358
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.241339961372212e-09,
+      "loss": 2.671,
+      "step": 4360
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1694078129250717e-09,
+      "loss": 2.052,
+      "step": 4362
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.0996212435647745e-09,
+      "loss": 2.3833,
+      "step": 4364
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.0319804031990732e-09,
+      "loss": 2.4489,
+      "step": 4366
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 9.664854371268516e-10,
+      "loss": 2.2591,
+      "step": 4368
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 9.03136486037126e-10,
+      "loss": 2.4332,
+      "step": 4370
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 8.419336860092663e-10,
+      "loss": 2.1944,
+      "step": 4372
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 7.828771685124413e-10,
+      "loss": 2.1891,
+      "step": 4374
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 7.259670604053969e-10,
+      "loss": 2.4247,
+      "step": 4376
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.712034839363445e-10,
+      "loss": 2.172,
+      "step": 4378
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.185865567422955e-10,
+      "loss": 2.3444,
+      "step": 4380
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.681163918492827e-10,
+      "loss": 2.3733,
+      "step": 4382
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.197930976718056e-10,
+      "loss": 2.1766,
+      "step": 4384
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.736167780127199e-10,
+      "loss": 2.3506,
+      "step": 4386
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.2958753206290275e-10,
+      "loss": 2.2205,
+      "step": 4388
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.877054544011438e-10,
+      "loss": 2.1202,
+      "step": 4390
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.479706349941436e-10,
+      "loss": 2.5056,
+      "step": 4392
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.1038315919584834e-10,
+      "loss": 2.2159,
+      "step": 4394
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.7494310774744955e-10,
+      "loss": 2.1418,
+      "step": 4396
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.4165055677749515e-10,
+      "loss": 2.5672,
+      "step": 4398
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.105055778014453e-10,
+      "loss": 2.844,
+      "step": 4400
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.8150823772156155e-10,
+      "loss": 2.2968,
+      "step": 4402
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.5465859882668463e-10,
+      "loss": 2.2367,
+      "step": 4404
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.299567187923456e-10,
+      "loss": 2.4825,
+      "step": 4406
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0740265068043264e-10,
+      "loss": 2.3133,
+      "step": 4408
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 8.699644293908016e-11,
+      "loss": 2.3099,
+      "step": 4410
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.873813940277973e-11,
+      "loss": 2.3648,
+      "step": 4412
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.262777929193607e-11,
+      "loss": 2.439,
+      "step": 4414
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.8665397213089034e-11,
+      "loss": 2.4944,
+      "step": 4416
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.685102315880261e-11,
+      "loss": 2.1788,
+      "step": 4418
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.718468250733185e-11,
+      "loss": 2.394,
+      "step": 4420
+    },
+    {
+      "epoch": 10.0,
+      "step": 4420,
+      "total_flos": 1.0214099064245453e+17,
+      "train_loss": 2.5490156790250027,
+      "train_runtime": 26958.9225,
+      "train_samples_per_second": 10.496,
+      "train_steps_per_second": 0.164
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 4420,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 4000,
+  "total_flos": 1.0214099064245453e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}