diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15162 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.998761456527124,
+  "eval_steps": 500,
+  "global_step": 5045,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6315789473684213e-07,
+      "loss": 3.2514,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 3.2746,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894736842105263e-07,
+      "loss": 3.3116,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 3.2935,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3157894736842106e-06,
+      "loss": 3.1796,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 3.2044,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8421052631578948e-06,
+      "loss": 3.23,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 3.1374,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.368421052631579e-06,
+      "loss": 3.2225,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 3.0712,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8947368421052634e-06,
+      "loss": 3.1417,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 3.1394,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.421052631578948e-06,
+      "loss": 3.0458,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 3.0644,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.947368421052632e-06,
+      "loss": 3.07,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 3.0137,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.473684210526316e-06,
+      "loss": 3.127,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 3.0631,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 2.9491,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 3.0062,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.526315789473685e-06,
+      "loss": 3.0418,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 3.0384,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0526315789473685e-06,
+      "loss": 2.9976,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 2.8664,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.578947368421054e-06,
+      "loss": 2.9123,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 2.851,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.1052631578947375e-06,
+      "loss": 2.9273,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 2.9114,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.631578947368423e-06,
+      "loss": 2.9303,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 2.8576,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.157894736842106e-06,
+      "loss": 2.943,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 2.8336,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.68421052631579e-06,
+      "loss": 2.8879,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 2.8228,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.210526315789474e-06,
+      "loss": 2.9397,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 2.8934,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.736842105263159e-06,
+      "loss": 2.8464,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 2.8852,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0263157894736844e-05,
+      "loss": 2.7809,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 2.8093,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0789473684210528e-05,
+      "loss": 2.8095,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 2.78,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1315789473684212e-05,
+      "loss": 2.7971,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 2.7739,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 2.8226,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 2.6981,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.236842105263158e-05,
+      "loss": 2.7664,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 2.7412,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2894736842105264e-05,
+      "loss": 2.794,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 2.7895,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3421052631578948e-05,
+      "loss": 2.8016,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 2.7087,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3947368421052631e-05,
+      "loss": 2.7312,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 2.7034,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4473684210526317e-05,
+      "loss": 2.7121,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 2.6458,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.709,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 2.7198,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5526315789473686e-05,
+      "loss": 2.6623,
+      "step": 118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 2.7108,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.605263157894737e-05,
+      "loss": 2.6155,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 2.6367,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6578947368421053e-05,
+      "loss": 2.6831,
+      "step": 126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 2.6989,
+      "step": 128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 2.6432,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 2.639,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.763157894736842e-05,
+      "loss": 2.6208,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 2.6826,
+      "step": 136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8157894736842107e-05,
+      "loss": 2.634,
+      "step": 138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 2.5627,
+      "step": 140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.868421052631579e-05,
+      "loss": 2.6513,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 2.5878,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9210526315789474e-05,
+      "loss": 2.6553,
+      "step": 146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 2.5739,
+      "step": 148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 2.5694,
+      "step": 150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 2.6193,
+      "step": 152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999991755216616e-05,
+      "loss": 2.6145,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996702088006e-05,
+      "loss": 2.6207,
+      "step": 156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999925797031122e-05,
+      "loss": 2.5708,
+      "step": 158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999986808373777e-05,
+      "loss": 2.5894,
+      "step": 160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999793881095173e-05,
+      "loss": 2.5574,
+      "step": 162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999703189225693e-05,
+      "loss": 2.5237,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999959600827887e-05,
+      "loss": 2.5201,
+      "step": 166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999472338431445e-05,
+      "loss": 2.5322,
+      "step": 168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999332179887346e-05,
+      "loss": 2.5435,
+      "step": 170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999175532877685e-05,
+      "loss": 2.4825,
+      "step": 172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999002397660767e-05,
+      "loss": 2.5491,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998812774522084e-05,
+      "loss": 2.4988,
+      "step": 176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998606663774318e-05,
+      "loss": 2.5125,
+      "step": 178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998384065757334e-05,
+      "loss": 2.5019,
+      "step": 180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999814498083819e-05,
+      "loss": 2.5975,
+      "step": 182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997889409411126e-05,
+      "loss": 2.48,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997617351897564e-05,
+      "loss": 2.4492,
+      "step": 186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999732880874612e-05,
+      "loss": 2.5289,
+      "step": 188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997023780432584e-05,
+      "loss": 2.5206,
+      "step": 190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996702267459943e-05,
+      "loss": 2.4659,
+      "step": 192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996364270358346e-05,
+      "loss": 2.4068,
+      "step": 194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999600978968515e-05,
+      "loss": 2.5316,
+      "step": 196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995638826024863e-05,
+      "loss": 2.5007,
+      "step": 198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99952513799892e-05,
+      "loss": 2.5139,
+      "step": 200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994847452217035e-05,
+      "loss": 2.486,
+      "step": 202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994427043374433e-05,
+      "loss": 2.4763,
+      "step": 204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993990154154628e-05,
+      "loss": 2.4561,
+      "step": 206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999353678527803e-05,
+      "loss": 2.4208,
+      "step": 208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993066937492224e-05,
+      "loss": 2.405,
+      "step": 210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992580611571975e-05,
+      "loss": 2.5216,
+      "step": 212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992077808319203e-05,
+      "loss": 2.4548,
+      "step": 214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991558528563018e-05,
+      "loss": 2.4286,
+      "step": 216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991022773159686e-05,
+      "loss": 2.4838,
+      "step": 218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990470542992645e-05,
+      "loss": 2.4973,
+      "step": 220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989901838972496e-05,
+      "loss": 2.4462,
+      "step": 222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998931666203701e-05,
+      "loss": 2.47,
+      "step": 224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998871501315112e-05,
+      "loss": 2.4337,
+      "step": 226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988096893306918e-05,
+      "loss": 2.4194,
+      "step": 228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998746230352365e-05,
+      "loss": 2.3912,
+      "step": 230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986811244847738e-05,
+      "loss": 2.4434,
+      "step": 232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998614371835274e-05,
+      "loss": 2.4184,
+      "step": 234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998545972513939e-05,
+      "loss": 2.3803,
+      "step": 236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984759266335548e-05,
+      "loss": 2.4425,
+      "step": 238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984042343096254e-05,
+      "loss": 2.4629,
+      "step": 240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983308956603677e-05,
+      "loss": 2.3591,
+      "step": 242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982559108067135e-05,
+      "loss": 2.4035,
+      "step": 244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981792798723105e-05,
+      "loss": 2.4544,
+      "step": 246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9981010029835192e-05,
+      "loss": 2.4415,
+      "step": 248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998021080269415e-05,
+      "loss": 2.4851,
+      "step": 250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997939511861787e-05,
+      "loss": 2.413,
+      "step": 252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978562978951378e-05,
+      "loss": 2.4147,
+      "step": 254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997771438506684e-05,
+      "loss": 2.4271,
+      "step": 256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976849338363546e-05,
+      "loss": 2.398,
+      "step": 258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997596784026792e-05,
+      "loss": 2.4301,
+      "step": 260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975069892233522e-05,
+      "loss": 2.3703,
+      "step": 262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974155495741024e-05,
+      "loss": 2.3411,
+      "step": 264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973224652298226e-05,
+      "loss": 2.3897,
+      "step": 266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972277363440044e-05,
+      "loss": 2.4257,
+      "step": 268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971313630728527e-05,
+      "loss": 2.4108,
+      "step": 270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997033345575282e-05,
+      "loss": 2.3462,
+      "step": 272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969336840129192e-05,
+      "loss": 2.3499,
+      "step": 274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968323785501018e-05,
+      "loss": 2.3375,
+      "step": 276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996729429353878e-05,
+      "loss": 2.3252,
+      "step": 278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966248365940073e-05,
+      "loss": 2.4312,
+      "step": 280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965186004429577e-05,
+      "loss": 2.3603,
+      "step": 282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964107210759087e-05,
+      "loss": 2.3189,
+      "step": 284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963011986707483e-05,
+      "loss": 2.3251,
+      "step": 286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996190033408074e-05,
+      "loss": 2.3982,
+      "step": 288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960772254711932e-05,
+      "loss": 2.3182,
+      "step": 290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959627750461208e-05,
+      "loss": 2.2967,
+      "step": 292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958466823215806e-05,
+      "loss": 2.3536,
+      "step": 294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957289474890044e-05,
+      "loss": 2.306,
+      "step": 296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9956095707425326e-05,
+      "loss": 2.3741,
+      "step": 298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995488552279011e-05,
+      "loss": 2.2472,
+      "step": 300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953658922979948e-05,
+      "loss": 2.3548,
+      "step": 302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995241591001745e-05,
+      "loss": 2.3481,
+      "step": 304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995115648595228e-05,
+      "loss": 2.311,
+      "step": 306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949880652861187e-05,
+      "loss": 2.4276,
+      "step": 308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994858841284796e-05,
+      "loss": 2.3092,
+      "step": 310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994727976804344e-05,
+      "loss": 2.3157,
+      "step": 312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945954720605534e-05,
+      "loss": 2.297,
+      "step": 314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944613272719185e-05,
+      "loss": 2.3921,
+      "step": 316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9943255426596383e-05,
+      "loss": 2.3172,
+      "step": 318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941881184476154e-05,
+      "loss": 2.2716,
+      "step": 320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994049054862457e-05,
+      "loss": 2.2714,
+      "step": 322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939083521334722e-05,
+      "loss": 2.3345,
+      "step": 324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937660104926743e-05,
+      "loss": 2.3128,
+      "step": 326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993622030174778e-05,
+      "loss": 2.2616,
+      "step": 328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993476411417201e-05,
+      "loss": 2.3352,
+      "step": 330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933291544600626e-05,
+      "loss": 2.3523,
+      "step": 332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931802595461827e-05,
+      "loss": 2.2428,
+      "step": 334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993029726921082e-05,
+      "loss": 2.3379,
+      "step": 336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928775568329834e-05,
+      "loss": 2.3561,
+      "step": 338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927237495328082e-05,
+      "loss": 2.3072,
+      "step": 340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992568305274178e-05,
+      "loss": 2.323,
+      "step": 342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924112243134134e-05,
+      "loss": 2.3019,
+      "step": 344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992252506909535e-05,
+      "loss": 2.2791,
+      "step": 346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920921533242597e-05,
+      "loss": 2.2754,
+      "step": 348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919301638220044e-05,
+      "loss": 2.2824,
+      "step": 350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917665386698826e-05,
+      "loss": 2.3103,
+      "step": 352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916012781377055e-05,
+      "loss": 2.284,
+      "step": 354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99143438249798e-05,
+      "loss": 2.2943,
+      "step": 356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912658520259097e-05,
+      "loss": 2.2663,
+      "step": 358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991095686999394e-05,
+      "loss": 2.3155,
+      "step": 360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909238876990283e-05,
+      "loss": 2.3001,
+      "step": 362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990750454408102e-05,
+      "loss": 2.2771,
+      "step": 364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905753874125987e-05,
+      "loss": 2.2087,
+      "step": 366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903986870011965e-05,
+      "loss": 2.2836,
+      "step": 368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990220353465267e-05,
+      "loss": 2.3225,
+      "step": 370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900403870988735e-05,
+      "loss": 2.2992,
+      "step": 372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989858788198774e-05,
+      "loss": 2.2237,
+      "step": 374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896755570644165e-05,
+      "loss": 2.2087,
+      "step": 376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894906939979415e-05,
+      "loss": 2.2498,
+      "step": 378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893041993041803e-05,
+      "loss": 2.2796,
+      "step": 380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891160732906538e-05,
+      "loss": 2.2606,
+      "step": 382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889263162675742e-05,
+      "loss": 2.2303,
+      "step": 384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887349285478427e-05,
+      "loss": 2.3025,
+      "step": 386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885419104470496e-05,
+      "loss": 2.2087,
+      "step": 388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883472622834724e-05,
+      "loss": 2.2149,
+      "step": 390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881509843780785e-05,
+      "loss": 2.2532,
+      "step": 392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879530770545212e-05,
+      "loss": 2.3051,
+      "step": 394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877535406391412e-05,
+      "loss": 2.2932,
+      "step": 396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875523754609655e-05,
+      "loss": 2.2128,
+      "step": 398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873495818517066e-05,
+      "loss": 2.3022,
+      "step": 400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871451601457622e-05,
+      "loss": 2.2225,
+      "step": 402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869391106802152e-05,
+      "loss": 2.2464,
+      "step": 404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986731433794832e-05,
+      "loss": 2.2404,
+      "step": 406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986522129832063e-05,
+      "loss": 2.2789,
+      "step": 408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863111991370414e-05,
+      "loss": 2.2627,
+      "step": 410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860986420575826e-05,
+      "loss": 2.2023,
+      "step": 412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985884458944184e-05,
+      "loss": 2.2156,
+      "step": 414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856686501500243e-05,
+      "loss": 2.2241,
+      "step": 416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854512160309626e-05,
+      "loss": 2.2255,
+      "step": 418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985232156945539e-05,
+      "loss": 2.2222,
+      "step": 420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985011473254972e-05,
+      "loss": 2.2924,
+      "step": 422
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847891653231594e-05,
+      "loss": 2.2019,
+      "step": 424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845652335166774e-05,
+      "loss": 2.254,
+      "step": 426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98433967820478e-05,
+      "loss": 2.1814,
+      "step": 428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984112499759398e-05,
+      "loss": 2.2191,
+      "step": 430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838836985551388e-05,
+      "loss": 2.2808,
+      "step": 432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983653274969286e-05,
+      "loss": 2.2299,
+      "step": 434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834212293817976e-05,
+      "loss": 2.2136,
+      "step": 436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983187562175307e-05,
+      "loss": 2.1976,
+      "step": 438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982952273735121e-05,
+      "loss": 2.1628,
+      "step": 440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827153644492208e-05,
+      "loss": 2.2554,
+      "step": 442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824768347082588e-05,
+      "loss": 2.2471,
+      "step": 444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822366849055603e-05,
+      "loss": 2.1757,
+      "step": 446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819949154371227e-05,
+      "loss": 2.2213,
+      "step": 448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817515267016117e-05,
+      "loss": 2.232,
+      "step": 450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981506519100367e-05,
+      "loss": 2.2565,
+      "step": 452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812598930373937e-05,
+      "loss": 2.1681,
+      "step": 454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810116489193683e-05,
+      "loss": 2.2391,
+      "step": 456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807617871556343e-05,
+      "loss": 2.1956,
+      "step": 458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805103081582032e-05,
+      "loss": 2.2029,
+      "step": 460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802572123417528e-05,
+      "loss": 2.2794,
+      "step": 462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980002500123627e-05,
+      "loss": 2.1982,
+      "step": 464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797461719238357e-05,
+      "loss": 2.2152,
+      "step": 466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794882281650527e-05,
+      "loss": 2.1839,
+      "step": 468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979228669272616e-05,
+      "loss": 2.2442,
+      "step": 470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978967495674527e-05,
+      "loss": 2.2155,
+      "step": 472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787047078014496e-05,
+      "loss": 2.2649,
+      "step": 474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784403060867098e-05,
+      "loss": 2.1753,
+      "step": 476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781742909662945e-05,
+      "loss": 2.2015,
+      "step": 478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779066628788514e-05,
+      "loss": 2.1838,
+      "step": 480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776374222656866e-05,
+      "loss": 2.0713,
+      "step": 482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773665695707676e-05,
+      "loss": 2.205,
+      "step": 484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770941052407178e-05,
+      "loss": 2.152,
+      "step": 486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768200297248193e-05,
+      "loss": 2.1674,
+      "step": 488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976544343475011e-05,
+      "loss": 2.1727,
+      "step": 490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976267046945887e-05,
+      "loss": 2.1173,
+      "step": 492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975988140594698e-05,
+      "loss": 2.1853,
+      "step": 494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975707624881348e-05,
+      "loss": 2.1542,
+      "step": 496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754255002683954e-05,
+      "loss": 2.1707,
+      "step": 498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9751417672210516e-05,
+      "loss": 2.1689,
+      "step": 500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748564262071798e-05,
+      "loss": 2.1635,
+      "step": 502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974569477697295e-05,
+      "loss": 2.1734,
+      "step": 504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742809221645633e-05,
+      "loss": 2.1961,
+      "step": 506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739907600848e-05,
+      "loss": 2.1399,
+      "step": 508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736989919364696e-05,
+      "loss": 2.1333,
+      "step": 510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734056182006856e-05,
+      "loss": 2.2111,
+      "step": 512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973110639361208e-05,
+      "loss": 2.1823,
+      "step": 514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728140559044443e-05,
+      "loss": 2.179,
+      "step": 516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725158683194482e-05,
+      "loss": 2.2008,
+      "step": 518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9722160770979176e-05,
+      "loss": 2.1732,
+      "step": 520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719146827341953e-05,
+      "loss": 2.2042,
+      "step": 522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971611685725268e-05,
+      "loss": 2.197,
+      "step": 524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971307086570764e-05,
+      "loss": 2.1346,
+      "step": 526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710008857729548e-05,
+      "loss": 2.1145,
+      "step": 528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706930838367517e-05,
+      "loss": 2.1761,
+      "step": 530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970383681269707e-05,
+      "loss": 2.1898,
+      "step": 532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700726785820116e-05,
+      "loss": 2.1953,
+      "step": 534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697600762864963e-05,
+      "loss": 2.148,
+      "step": 536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694458748986286e-05,
+      "loss": 2.1921,
+      "step": 538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9691300749365124e-05,
+      "loss": 2.2032,
+      "step": 540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688126769208885e-05,
+      "loss": 2.1631,
+      "step": 542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968493681375133e-05,
+      "loss": 2.1507,
+      "step": 544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681730888252546e-05,
+      "loss": 2.1865,
+      "step": 546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678508997998972e-05,
+      "loss": 2.1402,
+      "step": 548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675271148303365e-05,
+      "loss": 2.1377,
+      "step": 550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96720173445048e-05,
+      "loss": 2.2131,
+      "step": 552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966874759196866e-05,
+      "loss": 2.1884,
+      "step": 554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966546189608662e-05,
+      "loss": 2.1447,
+      "step": 556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966216026227665e-05,
+      "loss": 2.1327,
+      "step": 558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658842695983006e-05,
+      "loss": 2.129,
+      "step": 560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655509202676207e-05,
+      "loss": 2.1419,
+      "step": 562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652159787853045e-05,
+      "loss": 2.1318,
+      "step": 564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648794457036553e-05,
+      "loss": 2.1677,
+      "step": 566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645413215776018e-05,
+      "loss": 2.15,
+      "step": 568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642016069646962e-05,
+      "loss": 2.1764,
+      "step": 570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963860302425113e-05,
+      "loss": 2.1015,
+      "step": 572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635174085216485e-05,
+      "loss": 2.188,
+      "step": 574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631729258197204e-05,
+      "loss": 2.133,
+      "step": 576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628268548873652e-05,
+      "loss": 2.1666,
+      "step": 578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624791962952395e-05,
+      "loss": 2.1124,
+      "step": 580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621299506166163e-05,
+      "loss": 2.1674,
+      "step": 582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617791184273877e-05,
+      "loss": 2.1116,
+      "step": 584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614267003060594e-05,
+      "loss": 2.1396,
+      "step": 586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961072696833755e-05,
+      "loss": 2.0693,
+      "step": 588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96071710859421e-05,
+      "loss": 2.1861,
+      "step": 590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603599361737747e-05,
+      "loss": 2.1392,
+      "step": 592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600011801614104e-05,
+      "loss": 2.0486,
+      "step": 594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95964084114869e-05,
+      "loss": 2.1426,
+      "step": 596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959278919729798e-05,
+      "loss": 2.0353,
+      "step": 598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589154165015262e-05,
+      "loss": 2.1299,
+      "step": 600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958550332063276e-05,
+      "loss": 2.1517,
+      "step": 602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581836670170555e-05,
+      "loss": 2.0953,
+      "step": 604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95781542196748e-05,
+      "loss": 2.1526,
+      "step": 606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957445597521769e-05,
+      "loss": 2.1637,
+      "step": 608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570741942897474e-05,
+      "loss": 2.1758,
+      "step": 610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956701212883843e-05,
+      "loss": 2.1157,
+      "step": 612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563266539190864e-05,
+      "loss": 2.0837,
+      "step": 614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559505180131082e-05,
+      "loss": 2.1175,
+      "step": 616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 2.0993,
+      "step": 618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551935178610152e-05,
+      "loss": 2.1066,
+      "step": 620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548126548631606e-05,
+      "loss": 2.0927,
+      "step": 622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954430217420604e-05,
+      "loss": 2.1272,
+      "step": 624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540462061639678e-05,
+      "loss": 2.1036,
+      "step": 626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95366062172647e-05,
+      "loss": 2.0732,
+      "step": 628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532734647439228e-05,
+      "loss": 2.1357,
+      "step": 630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528847358547312e-05,
+      "loss": 2.0959,
+      "step": 632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952494435699892e-05,
+      "loss": 2.1112,
+      "step": 634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521025649229934e-05,
+      "loss": 2.09,
+      "step": 636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517091241702134e-05,
+      "loss": 2.0943,
+      "step": 638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513141140903192e-05,
+      "loss": 2.1057,
+      "step": 640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509175353346646e-05,
+      "loss": 2.1132,
+      "step": 642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505193885571908e-05,
+      "loss": 2.0786,
+      "step": 644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501196744144253e-05,
+      "loss": 2.1316,
+      "step": 646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949718393565479e-05,
+      "loss": 2.0988,
+      "step": 648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493155466720465e-05,
+      "loss": 2.1128,
+      "step": 650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489111343984047e-05,
+      "loss": 2.098,
+      "step": 652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485051574114124e-05,
+      "loss": 2.0799,
+      "step": 654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948097616380508e-05,
+      "loss": 2.0453,
+      "step": 656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476885119777088e-05,
+      "loss": 2.0585,
+      "step": 658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472778448776098e-05,
+      "loss": 2.0882,
+      "step": 660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468656157573846e-05,
+      "loss": 2.0342,
+      "step": 662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464518252967798e-05,
+      "loss": 2.0629,
+      "step": 664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460364741781185e-05,
+      "loss": 2.0726,
+      "step": 666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456195630862965e-05,
+      "loss": 2.1322,
+      "step": 668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452010927087827e-05,
+      "loss": 2.097,
+      "step": 670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447810637356163e-05,
+      "loss": 2.0981,
+      "step": 672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443594768594062e-05,
+      "loss": 2.075,
+      "step": 674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439363327753318e-05,
+      "loss": 2.128,
+      "step": 676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943511632181139e-05,
+      "loss": 2.0814,
+      "step": 678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943085375777141e-05,
+      "loss": 2.1763,
+      "step": 680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426575642662155e-05,
+      "loss": 2.0245,
+      "step": 682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422281983538056e-05,
+      "loss": 2.056,
+      "step": 684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941797278747917e-05,
+      "loss": 2.083,
+      "step": 686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941364806159118e-05,
+      "loss": 2.0602,
+      "step": 688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409307813005357e-05,
+      "loss": 2.1267,
+      "step": 690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.94049520488786e-05,
+      "loss": 2.0688,
+      "step": 692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400580776393365e-05,
+      "loss": 2.1494,
+      "step": 694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939619400275769e-05,
+      "loss": 2.1451,
+      "step": 696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391791735205182e-05,
+      "loss": 2.0589,
+      "step": 698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387373980994985e-05,
+      "loss": 2.0711,
+      "step": 700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382940747411786e-05,
+      "loss": 2.1432,
+      "step": 702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9378492041765794e-05,
+      "loss": 2.1217,
+      "step": 704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937402787139273e-05,
+      "loss": 2.142,
+      "step": 706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936954824365382e-05,
+      "loss": 2.0659,
+      "step": 708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365053165935774e-05,
+      "loss": 2.0815,
+      "step": 710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360542645650783e-05,
+      "loss": 2.0562,
+      "step": 712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356016690236503e-05,
+      "loss": 2.1152,
+      "step": 714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351475307156028e-05,
+      "loss": 2.0711,
+      "step": 716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346918503897907e-05,
+      "loss": 2.0945,
+      "step": 718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342346287976113e-05,
+      "loss": 2.0916,
+      "step": 720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337758666930034e-05,
+      "loss": 2.0951,
+      "step": 722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933315564832446e-05,
+      "loss": 2.0957,
+      "step": 724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328537239749553e-05,
+      "loss": 2.1126,
+      "step": 726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9323903448820886e-05,
+      "loss": 2.0479,
+      "step": 728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319254283179375e-05,
+      "loss": 2.0838,
+      "step": 730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931458975049129e-05,
+      "loss": 2.0931,
+      "step": 732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309909858448245e-05,
+      "loss": 2.0772,
+      "step": 734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930521461476718e-05,
+      "loss": 2.082,
+      "step": 736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300504027190344e-05,
+      "loss": 2.0263,
+      "step": 738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295778103485297e-05,
+      "loss": 2.0403,
+      "step": 740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9291036851444885e-05,
+      "loss": 2.0782,
+      "step": 742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286280278887218e-05,
+      "loss": 2.0964,
+      "step": 744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281508393655683e-05,
+      "loss": 2.0317,
+      "step": 746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276721203618916e-05,
+      "loss": 2.087,
+      "step": 748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927191871667078e-05,
+      "loss": 2.0234,
+      "step": 750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9267100940730367e-05,
+      "loss": 2.0623,
+      "step": 752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262267883741986e-05,
+      "loss": 2.1058,
+      "step": 754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257419553675136e-05,
+      "loss": 2.047,
+      "step": 756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252555958524502e-05,
+      "loss": 1.993,
+      "step": 758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924767710630995e-05,
+      "loss": 2.0882,
+      "step": 760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924278300507648e-05,
+      "loss": 2.0796,
+      "step": 762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923787366289427e-05,
+      "loss": 2.0676,
+      "step": 764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92329490878586e-05,
+      "loss": 2.0807,
+      "step": 766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9228009288089886e-05,
+      "loss": 2.0079,
+      "step": 768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922305427173364e-05,
+      "loss": 2.012,
+      "step": 770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921808404696048e-05,
+      "loss": 2.0989,
+      "step": 772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921309862196608e-05,
+      "loss": 2.0741,
+      "step": 774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208098004971193e-05,
+      "loss": 2.0979,
+      "step": 776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203082204221622e-05,
+      "loss": 1.9967,
+      "step": 778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198051227988202e-05,
+      "loss": 2.0629,
+      "step": 780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193005084566798e-05,
+      "loss": 2.038,
+      "step": 782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918794378227828e-05,
+      "loss": 2.0628,
+      "step": 784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918286732946852e-05,
+      "loss": 2.078,
+      "step": 786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917777573450836e-05,
+      "loss": 2.0428,
+      "step": 788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917266900579363e-05,
+      "loss": 2.0763,
+      "step": 790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167547151745096e-05,
+      "loss": 2.063,
+      "step": 792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916241018080848e-05,
+      "loss": 2.0455,
+      "step": 794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157258101454417e-05,
+      "loss": 2.1249,
+      "step": 796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152090922178466e-05,
+      "loss": 2.0257,
+      "step": 798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146908651501087e-05,
+      "loss": 2.0711,
+      "step": 800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141711297967614e-05,
+      "loss": 2.099,
+      "step": 802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913649887014826e-05,
+      "loss": 2.0683,
+      "step": 804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913127137663809e-05,
+      "loss": 1.9897,
+      "step": 806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126028826057016e-05,
+      "loss": 2.0291,
+      "step": 808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912077122704978e-05,
+      "loss": 2.0337,
+      "step": 810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115498588285932e-05,
+      "loss": 2.0845,
+      "step": 812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110210918459823e-05,
+      "loss": 2.0833,
+      "step": 814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104908226290594e-05,
+      "loss": 2.0388,
+      "step": 816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099590520522155e-05,
+      "loss": 2.0332,
+      "step": 818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909425780992317e-05,
+      "loss": 2.0105,
+      "step": 820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908891010328705e-05,
+      "loss": 2.0848,
+      "step": 822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908354740943193e-05,
+      "loss": 2.0553,
+      "step": 824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078169737200663e-05,
+      "loss": 2.043,
+      "step": 826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907277709546079e-05,
+      "loss": 2.0327,
+      "step": 828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906736949310456e-05,
+      "loss": 2.04,
+      "step": 830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906194693904885e-05,
+      "loss": 1.9995,
+      "step": 832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056509442235242e-05,
+      "loss": 2.0288,
+      "step": 834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051057011629915e-05,
+      "loss": 2.0136,
+      "step": 836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045589656223698e-05,
+      "loss": 2.133,
+      "step": 838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040107385032023e-05,
+      "loss": 2.0006,
+      "step": 840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034610207094917e-05,
+      "loss": 2.0015,
+      "step": 842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029098131476988e-05,
+      "loss": 2.0462,
+      "step": 844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9023571167267407e-05,
+      "loss": 2.0338,
+      "step": 846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018029323579904e-05,
+      "loss": 1.9753,
+      "step": 848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901247260955274e-05,
+      "loss": 2.0099,
+      "step": 850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900690103434869e-05,
+      "loss": 1.9688,
+      "step": 852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900131460715504e-05,
+      "loss": 2.0587,
+      "step": 854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8995713337183573e-05,
+      "loss": 2.0002,
+      "step": 856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8990097233670533e-05,
+      "loss": 2.0339,
+      "step": 858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898446630587664e-05,
+      "loss": 1.9846,
+      "step": 860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8978820563087042e-05,
+      "loss": 2.01,
+      "step": 862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973160014611326e-05,
+      "loss": 2.0621,
+      "step": 864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967484669783494e-05,
+      "loss": 2.0265,
+      "step": 866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896179453796194e-05,
+      "loss": 2.0072,
+      "step": 868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956089628529448e-05,
+      "loss": 2.0323,
+      "step": 870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895036995089317e-05,
+      "loss": 2.0687,
+      "step": 872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944635514484594e-05,
+      "loss": 2.071,
+      "step": 874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938886328759573e-05,
+      "loss": 2.0765,
+      "step": 876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933122403198254e-05,
+      "loss": 2.0222,
+      "step": 878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927343747305103e-05,
+      "loss": 1.9957,
+      "step": 880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921550370608876e-05,
+      "loss": 1.9802,
+      "step": 882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915742282662596e-05,
+      "loss": 2.042,
+      "step": 884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8909919493043552e-05,
+      "loss": 1.9863,
+      "step": 886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904082011353275e-05,
+      "loss": 2.0574,
+      "step": 888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889822984721751e-05,
+      "loss": 2.0093,
+      "step": 890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889236301028623e-05,
+      "loss": 2.0526,
+      "step": 892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8886481510233593e-05,
+      "loss": 2.0059,
+      "step": 894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880585356757933e-05,
+      "loss": 2.0308,
+      "step": 896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887467455958176e-05,
+      "loss": 2.0072,
+      "step": 898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868749128451715e-05,
+      "loss": 2.0051,
+      "step": 900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8862809073138577e-05,
+      "loss": 2.0101,
+      "step": 902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885685440343725e-05,
+      "loss": 1.9896,
+      "step": 904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850885129166716e-05,
+      "loss": 2.014,
+      "step": 906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844901260170052e-05,
+      "loss": 2.0076,
+      "step": 908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838902806314402e-05,
+      "loss": 2.021,
+      "step": 910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8832889777490953e-05,
+      "loss": 2.0469,
+      "step": 912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882686218361493e-05,
+      "loss": 2.0043,
+      "step": 914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882082003462558e-05,
+      "loss": 2.0016,
+      "step": 916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814763340486134e-05,
+      "loss": 2.0217,
+      "step": 918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808692111183825e-05,
+      "loss": 2.1145,
+      "step": 920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802606356729845e-05,
+      "loss": 1.987,
+      "step": 922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796506087159343e-05,
+      "loss": 2.0033,
+      "step": 924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790391312531394e-05,
+      "loss": 1.9621,
+      "step": 926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784262042929004e-05,
+      "loss": 2.0327,
+      "step": 928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877811828845906e-05,
+      "loss": 2.1008,
+      "step": 930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877196005925236e-05,
+      "loss": 1.9997,
+      "step": 932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876578736546355e-05,
+      "loss": 2.062,
+      "step": 934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875960021727114e-05,
+      "loss": 2.0136,
+      "step": 936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8753398624877463e-05,
+      "loss": 2.0292,
+      "step": 938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874718259850868e-05,
+      "loss": 2.0066,
+      "step": 940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874095214841475e-05,
+      "loss": 2.0487,
+      "step": 942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734707284869413e-05,
+      "loss": 2.0171,
+      "step": 944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728448018170178e-05,
+      "loss": 2.0562,
+      "step": 946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722174358638304e-05,
+      "loss": 1.9528,
+      "step": 948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871588631661879e-05,
+      "loss": 2.0178,
+      "step": 950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870958390248034e-05,
+      "loss": 2.0379,
+      "step": 952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870326712661536e-05,
+      "loss": 2.0562,
+      "step": 954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869693599943994e-05,
+      "loss": 1.9802,
+      "step": 956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869059053139384e-05,
+      "loss": 1.9725,
+      "step": 958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868423073294046e-05,
+      "loss": 2.0041,
+      "step": 960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867785661456683e-05,
+      "loss": 2.0382,
+      "step": 962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8671468186783594e-05,
+      "loss": 1.976,
+      "step": 964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665065460124993e-05,
+      "loss": 2.0281,
+      "step": 966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865864844514885e-05,
+      "loss": 2.0112,
+      "step": 968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865221715243654e-05,
+      "loss": 2.0198,
+      "step": 970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864577159259299e-05,
+      "loss": 1.9579,
+      "step": 972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863931177624664e-05,
+      "loss": 1.9744,
+      "step": 974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632837714049462e-05,
+      "loss": 1.9999,
+      "step": 976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862634941667689e-05,
+      "loss": 2.0096,
+      "step": 978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8619846894827855e-05,
+      "loss": 1.9432,
+      "step": 980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861333015922473e-05,
+      "loss": 2.004,
+      "step": 982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860679922061333e-05,
+      "loss": 1.9629,
+      "step": 984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8600254089762886e-05,
+      "loss": 1.972,
+      "step": 986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593694777466048e-05,
+      "loss": 1.9807,
+      "step": 988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8587121294538822e-05,
+      "loss": 2.0297,
+      "step": 990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85805336518206e-05,
+      "loss": 1.9988,
+      "step": 992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857393186017413e-05,
+      "loss": 1.9464,
+      "step": 994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856731593048547e-05,
+      "loss": 1.9911,
+      "step": 996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560685873664008e-05,
+      "loss": 2.018,
+      "step": 998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554041700642412e-05,
+      "loss": 1.9837,
+      "step": 1000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547383422376643e-05,
+      "loss": 1.9525,
+      "step": 1002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854071104984591e-05,
+      "loss": 1.9358,
+      "step": 1004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534024594052668e-05,
+      "loss": 1.968,
+      "step": 1006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527324066022593e-05,
+      "loss": 1.9762,
+      "step": 1008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8520609476804565e-05,
+      "loss": 1.9542,
+      "step": 1010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513880837470648e-05,
+      "loss": 2.0005,
+      "step": 1012
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850713815911608e-05,
+      "loss": 1.9702,
+      "step": 1014
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500381452859245e-05,
+      "loss": 1.9957,
+      "step": 1016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849361072984166e-05,
+      "loss": 1.9995,
+      "step": 1018
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848682600122795e-05,
+      "loss": 1.9441,
+      "step": 1020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848002727820584e-05,
+      "loss": 1.9178,
+      "step": 1022
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473214571986133e-05,
+      "loss": 1.9069,
+      "step": 1024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846638789380268e-05,
+      "loss": 1.9558,
+      "step": 1026
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459547254912387e-05,
+      "loss": 1.9686,
+      "step": 1028
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845269266659516e-05,
+      "loss": 1.9835,
+      "step": 1030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445824140153928e-05,
+      "loss": 2.0155,
+      "step": 1032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843894168691459e-05,
+      "loss": 1.8726,
+      "step": 1034
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843204531822601e-05,
+      "loss": 1.9797,
+      "step": 1036
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425135045460004e-05,
+      "loss": 1.9303,
+      "step": 1038
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418210880011314e-05,
+      "loss": 1.9986,
+      "step": 1040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411272833297587e-05,
+      "loss": 1.9149,
+      "step": 1042
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8404320916759362e-05,
+      "loss": 1.9124,
+      "step": 1044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8397355141860046e-05,
+      "loss": 1.9599,
+      "step": 1046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390375520085905e-05,
+      "loss": 1.9861,
+      "step": 1048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383382062946025e-05,
+      "loss": 1.9648,
+      "step": 1050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8376374781972324e-05,
+      "loss": 1.9811,
+      "step": 1052
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369353688719496e-05,
+      "loss": 1.9469,
+      "step": 1054
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8362318794765025e-05,
+      "loss": 1.9891,
+      "step": 1056
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835527011170914e-05,
+      "loss": 1.9981,
+      "step": 1058
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8351740602899463e-05,
+      "loss": 2.0261,
+      "step": 1060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834467125799164e-05,
+      "loss": 1.9084,
+      "step": 1062
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337588153082434e-05,
+      "loss": 1.9241,
+      "step": 1064
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8330491299851573e-05,
+      "loss": 1.9124,
+      "step": 1066
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832338071000146e-05,
+      "loss": 2.0013,
+      "step": 1068
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316256395257156e-05,
+      "loss": 1.9589,
+      "step": 1070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309118367366344e-05,
+      "loss": 1.9428,
+      "step": 1072
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8301966638099317e-05,
+      "loss": 1.9133,
+      "step": 1074
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8294801219248975e-05,
+      "loss": 1.9266,
+      "step": 1076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828762212263078e-05,
+      "loss": 1.9474,
+      "step": 1078
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8280429360082754e-05,
+      "loss": 1.9653,
+      "step": 1080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8273222943465446e-05,
+      "loss": 1.9784,
+      "step": 1082
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266002884661928e-05,
+      "loss": 1.9589,
+      "step": 1084
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8258769195577763e-05,
+      "loss": 1.994,
+      "step": 1086
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825152188814099e-05,
+      "loss": 1.8801,
+      "step": 1088
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824426097430211e-05,
+      "loss": 1.9744,
+      "step": 1090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8236986466034048e-05,
+      "loss": 1.9841,
+      "step": 1092
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8229698375332158e-05,
+      "loss": 1.876,
+      "step": 1094
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8222396714214182e-05,
+      "loss": 1.9546,
+      "step": 1096
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8215081494720248e-05,
+      "loss": 1.9897,
+      "step": 1098
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8207752728912835e-05,
+      "loss": 1.9276,
+      "step": 1100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8200410428876756e-05,
+      "loss": 1.9676,
+      "step": 1102
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193054606719144e-05,
+      "loss": 1.9912,
+      "step": 1104
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8185685274569443e-05,
+      "loss": 1.9544,
+      "step": 1106
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817830244457935e-05,
+      "loss": 1.9816,
+      "step": 1108
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170906128922842e-05,
+      "loss": 1.9886,
+      "step": 1110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816349633979611e-05,
+      "loss": 1.969,
+      "step": 1112
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815607308941759e-05,
+      "loss": 1.9711,
+      "step": 1114
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8148636390027884e-05,
+      "loss": 1.9506,
+      "step": 1116
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8141186253889805e-05,
+      "loss": 1.9346,
+      "step": 1118
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8133722693288294e-05,
+      "loss": 1.9867,
+      "step": 1120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812624572053044e-05,
+      "loss": 1.9695,
+      "step": 1122
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118755347945448e-05,
+      "loss": 1.9308,
+      "step": 1124
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8111251587884616e-05,
+      "loss": 1.9865,
+      "step": 1126
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103734452721327e-05,
+      "loss": 1.9177,
+      "step": 1128
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8096203954851e-05,
+      "loss": 1.9324,
+      "step": 1130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808866010669111e-05,
+      "loss": 1.9745,
+      "step": 1132
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8081102920681128e-05,
+      "loss": 2.0365,
+      "step": 1134
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8073532409282532e-05,
+      "loss": 1.9127,
+      "step": 1136
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8065948584978762e-05,
+      "loss": 1.8919,
+      "step": 1138
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8058351460275222e-05,
+      "loss": 1.9756,
+      "step": 1140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8050741047699233e-05,
+      "loss": 1.9265,
+      "step": 1142
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8043117359800043e-05,
+      "loss": 1.917,
+      "step": 1144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8035480409148785e-05,
+      "loss": 1.9275,
+      "step": 1146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802783020833845e-05,
+      "loss": 1.9117,
+      "step": 1148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020166769983894e-05,
+      "loss": 1.9065,
+      "step": 1150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012490106721798e-05,
+      "loss": 1.9555,
+      "step": 1152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8004800231210637e-05,
+      "loss": 1.9785,
+      "step": 1154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7997097156130692e-05,
+      "loss": 1.9686,
+      "step": 1156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7989380894183998e-05,
+      "loss": 1.9114,
+      "step": 1158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981651458094334e-05,
+      "loss": 1.9694,
+      "step": 1160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797390886060721e-05,
+      "loss": 1.9375,
+      "step": 1162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7966153114489824e-05,
+      "loss": 1.9916,
+      "step": 1164
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7958384232531076e-05,
+      "loss": 1.9377,
+      "step": 1166
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7950602227541513e-05,
+      "loss": 2.0273,
+      "step": 1168
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794280711235332e-05,
+      "loss": 1.9149,
+      "step": 1170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7934998899820306e-05,
+      "loss": 1.9642,
+      "step": 1172
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792717760281788e-05,
+      "loss": 1.9948,
+      "step": 1174
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7919343234243015e-05,
+      "loss": 1.9323,
+      "step": 1176
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911495807014248e-05,
+      "loss": 1.9253,
+      "step": 1178
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790363533407165e-05,
+      "loss": 1.9426,
+      "step": 1180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895761828376795e-05,
+      "loss": 1.9204,
+      "step": 1182
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7887875302912755e-05,
+      "loss": 1.9679,
+      "step": 1184
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787997577068407e-05,
+      "loss": 1.9159,
+      "step": 1186
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7872063244716726e-05,
+      "loss": 1.9367,
+      "step": 1188
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864137738058135e-05,
+      "loss": 1.8958,
+      "step": 1190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856199263777113e-05,
+      "loss": 1.8911,
+      "step": 1192
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784824783496386e-05,
+      "loss": 1.8908,
+      "step": 1194
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7840283464729942e-05,
+      "loss": 1.9384,
+      "step": 1196
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832306166208252e-05,
+      "loss": 1.8674,
+      "step": 1198
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782431595255302e-05,
+      "loss": 1.935,
+      "step": 1200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7816312836939754e-05,
+      "loss": 1.9806,
+      "step": 1202
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780829683256525e-05,
+      "loss": 1.9114,
+      "step": 1204
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780026795264755e-05,
+      "loss": 1.9293,
+      "step": 1206
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779222621042593e-05,
+      "loss": 1.9296,
+      "step": 1208
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778417161916087e-05,
+      "loss": 1.879,
+      "step": 1210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7776104192134047e-05,
+      "loss": 1.946,
+      "step": 1212
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77680239426483e-05,
+      "loss": 1.9035,
+      "step": 1214
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77599308840276e-05,
+      "loss": 1.9086,
+      "step": 1216
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751825029617066e-05,
+      "loss": 1.986,
+      "step": 1218
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774370639278289e-05,
+      "loss": 1.9371,
+      "step": 1220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735574986912353e-05,
+      "loss": 1.9208,
+      "step": 1222
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7727430825413794e-05,
+      "loss": 1.9345,
+      "step": 1224
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7719273921716582e-05,
+      "loss": 1.8995,
+      "step": 1226
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7711104289271097e-05,
+      "loss": 1.8612,
+      "step": 1228
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770292194154871e-05,
+      "loss": 1.9183,
+      "step": 1230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7694726892041757e-05,
+      "loss": 1.9048,
+      "step": 1232
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7686519154263513e-05,
+      "loss": 1.9318,
+      "step": 1234
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7678298741748196e-05,
+      "loss": 1.9337,
+      "step": 1236
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76700656680509e-05,
+      "loss": 1.8913,
+      "step": 1238
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7661819946747613e-05,
+      "loss": 1.9245,
+      "step": 1240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653561591435168e-05,
+      "loss": 1.9101,
+      "step": 1242
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645290615731232e-05,
+      "loss": 1.8744,
+      "step": 1244
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7637007033274293e-05,
+      "loss": 1.936,
+      "step": 1246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7628710857723615e-05,
+      "loss": 1.963,
+      "step": 1248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7620402102759232e-05,
+      "loss": 1.8766,
+      "step": 1250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7612080782081924e-05,
+      "loss": 1.9541,
+      "step": 1252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7603746909413185e-05,
+      "loss": 1.9236,
+      "step": 1254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595400498495214e-05,
+      "loss": 1.9755,
+      "step": 1256
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587041563090873e-05,
+      "loss": 1.9019,
+      "step": 1258
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578670116983696e-05,
+      "loss": 1.928,
+      "step": 1260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7570286173977825e-05,
+      "loss": 1.9529,
+      "step": 1262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7561889747898026e-05,
+      "loss": 1.8803,
+      "step": 1264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7553480852589635e-05,
+      "loss": 1.9164,
+      "step": 1266
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754505950191856e-05,
+      "loss": 1.9268,
+      "step": 1268
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7536625709771244e-05,
+      "loss": 1.963,
+      "step": 1270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7528179490054644e-05,
+      "loss": 1.9253,
+      "step": 1272
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519720856696206e-05,
+      "loss": 1.9441,
+      "step": 1274
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751124982364386e-05,
+      "loss": 1.9124,
+      "step": 1276
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7507009661596783e-05,
+      "loss": 1.9467,
+      "step": 1278
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7498520055200636e-05,
+      "loss": 1.9075,
+      "step": 1280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7490018084069778e-05,
+      "loss": 1.9486,
+      "step": 1282
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7481503762223587e-05,
+      "loss": 1.8724,
+      "step": 1284
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7472977103701818e-05,
+      "loss": 1.9223,
+      "step": 1286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.746443812256456e-05,
+      "loss": 1.906,
+      "step": 1288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745588683289222e-05,
+      "loss": 1.9044,
+      "step": 1290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447323248785504e-05,
+      "loss": 1.9773,
+      "step": 1292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7438747384365396e-05,
+      "loss": 1.9311,
+      "step": 1294
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743015925377312e-05,
+      "loss": 1.8699,
+      "step": 1296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7421558871170138e-05,
+      "loss": 1.8862,
+      "step": 1298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412946250738096e-05,
+      "loss": 1.8753,
+      "step": 1300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404321406678844e-05,
+      "loss": 1.9164,
+      "step": 1302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7395684353214372e-05,
+      "loss": 1.9007,
+      "step": 1304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7387035104586804e-05,
+      "loss": 1.844,
+      "step": 1306
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7378373675058383e-05,
+      "loss": 1.8854,
+      "step": 1308
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7369700078911423e-05,
+      "loss": 1.9168,
+      "step": 1310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7361014330448312e-05,
+      "loss": 1.9127,
+      "step": 1312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7352316443991476e-05,
+      "loss": 1.936,
+      "step": 1314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7343606433883348e-05,
+      "loss": 1.9184,
+      "step": 1316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.733488431448636e-05,
+      "loss": 1.9462,
+      "step": 1318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732615010018291e-05,
+      "loss": 1.9393,
+      "step": 1320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7317403805375338e-05,
+      "loss": 1.8828,
+      "step": 1322
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7308645444485905e-05,
+      "loss": 1.9101,
+      "step": 1324
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7299875031956763e-05,
+      "loss": 1.9021,
+      "step": 1326
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7291092582249953e-05,
+      "loss": 1.9104,
+      "step": 1328
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7282298109847346e-05,
+      "loss": 1.9196,
+      "step": 1330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7273491629250648e-05,
+      "loss": 1.8977,
+      "step": 1332
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264673154981368e-05,
+      "loss": 1.9413,
+      "step": 1334
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725584270158078e-05,
+      "loss": 1.867,
+      "step": 1336
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724700028360993e-05,
+      "loss": 1.8905,
+      "step": 1338
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238145915649572e-05,
+      "loss": 1.9282,
+      "step": 1340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722927961230018e-05,
+      "loss": 1.8926,
+      "step": 1342
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220401388181903e-05,
+      "loss": 1.951,
+      "step": 1344
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721151125793455e-05,
+      "loss": 1.8906,
+      "step": 1346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7202609236217558e-05,
+      "loss": 1.8654,
+      "step": 1348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7193695337709973e-05,
+      "loss": 1.9713,
+      "step": 1350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184769577110436e-05,
+      "loss": 1.8856,
+      "step": 1352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7175831969137128e-05,
+      "loss": 1.9382,
+      "step": 1354
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166882528527784e-05,
+      "loss": 1.902,
+      "step": 1356
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7157921270039644e-05,
+      "loss": 1.899,
+      "step": 1358
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148948208449437e-05,
+      "loss": 1.9271,
+      "step": 1360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713996335855335e-05,
+      "loss": 1.9244,
+      "step": 1362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7130966735167006e-05,
+      "loss": 1.9116,
+      "step": 1364
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712195835312546e-05,
+      "loss": 1.9426,
+      "step": 1366
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.711293822728313e-05,
+      "loss": 1.8533,
+      "step": 1368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710390637251382e-05,
+      "loss": 1.9061,
+      "step": 1370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7094862803710665e-05,
+      "loss": 1.8954,
+      "step": 1372
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708580753578612e-05,
+      "loss": 1.8956,
+      "step": 1374
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707674058367193e-05,
+      "loss": 1.9322,
+      "step": 1376
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7067661962319104e-05,
+      "loss": 1.8735,
+      "step": 1378
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7058571686697898e-05,
+      "loss": 1.8245,
+      "step": 1380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704946977179778e-05,
+      "loss": 1.8964,
+      "step": 1382
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7040356232627414e-05,
+      "loss": 1.9361,
+      "step": 1384
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7031231084214633e-05,
+      "loss": 1.9151,
+      "step": 1386
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7022094341606407e-05,
+      "loss": 1.844,
+      "step": 1388
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7012946019868835e-05,
+      "loss": 1.9306,
+      "step": 1390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7003786134087094e-05,
+      "loss": 1.9036,
+      "step": 1392
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699461469936545e-05,
+      "loss": 1.9098,
+      "step": 1394
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6985431730827196e-05,
+      "loss": 1.8859,
+      "step": 1396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6976237243614648e-05,
+      "loss": 1.8612,
+      "step": 1398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696703125288912e-05,
+      "loss": 1.9084,
+      "step": 1400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695781377383089e-05,
+      "loss": 1.8829,
+      "step": 1402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6948584821639183e-05,
+      "loss": 1.8867,
+      "step": 1404
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6939344411532136e-05,
+      "loss": 1.8786,
+      "step": 1406
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6930092558746792e-05,
+      "loss": 1.8937,
+      "step": 1408
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6920829278539055e-05,
+      "loss": 1.9332,
+      "step": 1410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6911554586183666e-05,
+      "loss": 1.8744,
+      "step": 1412
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.69022684969742e-05,
+      "loss": 1.8573,
+      "step": 1414
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689297102622301e-05,
+      "loss": 1.8946,
+      "step": 1416
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6883662189261223e-05,
+      "loss": 1.8793,
+      "step": 1418
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687434200143871e-05,
+      "loss": 1.9337,
+      "step": 1420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865010478124053e-05,
+      "loss": 1.8389,
+      "step": 1422
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6855667634704534e-05,
+      "loss": 1.923,
+      "step": 1424
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6846313486586093e-05,
+      "loss": 1.8946,
+      "step": 1426
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836948049193318e-05,
+      "loss": 1.9016,
+      "step": 1428
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682757133796941e-05,
+      "loss": 1.8891,
+      "step": 1430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681818336837616e-05,
+      "loss": 1.8951,
+      "step": 1432
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6808784155893917e-05,
+      "loss": 1.8809,
+      "step": 1434
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6799373716021582e-05,
+      "loss": 1.8548,
+      "step": 1436
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678995206427656e-05,
+      "loss": 1.9315,
+      "step": 1438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.678051921619475e-05,
+      "loss": 1.9095,
+      "step": 1440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67710751873305e-05,
+      "loss": 1.8919,
+      "step": 1442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6761619993256615e-05,
+      "loss": 1.9301,
+      "step": 1444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67521536495643e-05,
+      "loss": 1.8959,
+      "step": 1446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6742676171863137e-05,
+      "loss": 1.9556,
+      "step": 1448
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.673318757578108e-05,
+      "loss": 1.8719,
+      "step": 1450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6723687876964417e-05,
+      "loss": 1.8372,
+      "step": 1452
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6714177091077736e-05,
+      "loss": 1.8446,
+      "step": 1454
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6704655233803912e-05,
+      "loss": 1.926,
+      "step": 1456
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669512232084408e-05,
+      "loss": 1.8995,
+      "step": 1458
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6685578367917588e-05,
+      "loss": 1.8692,
+      "step": 1460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667602339076201e-05,
+      "loss": 1.8713,
+      "step": 1462
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666645740513309e-05,
+      "loss": 1.9277,
+      "step": 1464
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665688042680472e-05,
+      "loss": 1.9008,
+      "step": 1466
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6647292471568924e-05,
+      "loss": 1.8923,
+      "step": 1468
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6637693555235825e-05,
+      "loss": 1.8421,
+      "step": 1470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662808369363362e-05,
+      "loss": 1.8996,
+      "step": 1472
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6618462902608557e-05,
+      "loss": 1.9306,
+      "step": 1474
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6608831198024896e-05,
+      "loss": 1.8789,
+      "step": 1476
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659918859576491e-05,
+      "loss": 1.9054,
+      "step": 1478
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6589535111728823e-05,
+      "loss": 1.9482,
+      "step": 1480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657987076183482e-05,
+      "loss": 1.8926,
+      "step": 1482
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570195562018994e-05,
+      "loss": 1.9385,
+      "step": 1484
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6560509528235323e-05,
+      "loss": 1.9298,
+      "step": 1486
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550812676455667e-05,
+      "loss": 1.8858,
+      "step": 1488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.654110502266971e-05,
+      "loss": 1.8779,
+      "step": 1490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653138658288495e-05,
+      "loss": 1.8904,
+      "step": 1492
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6521657373126677e-05,
+      "loss": 1.9057,
+      "step": 1494
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6511917409437936e-05,
+      "loss": 1.8549,
+      "step": 1496
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6502166707879505e-05,
+      "loss": 1.8362,
+      "step": 1498
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6492405284529868e-05,
+      "loss": 1.8259,
+      "step": 1500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6482633155485187e-05,
+      "loss": 1.8785,
+      "step": 1502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647285033685928e-05,
+      "loss": 1.8787,
+      "step": 1504
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6463056844783595e-05,
+      "loss": 1.9552,
+      "step": 1506
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645325269540717e-05,
+      "loss": 1.876,
+      "step": 1508
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.644343790489663e-05,
+      "loss": 1.9042,
+      "step": 1510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6433612489436136e-05,
+      "loss": 1.8759,
+      "step": 1512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642377646522737e-05,
+      "loss": 1.8307,
+      "step": 1514
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641392984848951e-05,
+      "loss": 1.8962,
+      "step": 1516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6404072655459205e-05,
+      "loss": 1.9286,
+      "step": 1518
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6394204902390535e-05,
+      "loss": 1.8264,
+      "step": 1520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6384326605555e-05,
+      "loss": 1.845,
+      "step": 1522
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637443778124148e-05,
+      "loss": 1.9033,
+      "step": 1524
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6364538445756224e-05,
+      "loss": 1.8926,
+      "step": 1526
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6354628615422803e-05,
+      "loss": 1.9188,
+      "step": 1528
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6344708306582096e-05,
+      "loss": 1.8789,
+      "step": 1530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6334777535592267e-05,
+      "loss": 1.8315,
+      "step": 1532
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632483631882872e-05,
+      "loss": 1.8662,
+      "step": 1534
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6314884672684098e-05,
+      "loss": 1.8962,
+      "step": 1536
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6304922613568238e-05,
+      "loss": 1.9008,
+      "step": 1538
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6294950157908133e-05,
+      "loss": 1.9109,
+      "step": 1540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628496732214793e-05,
+      "loss": 1.8242,
+      "step": 1542
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627497412274891e-05,
+      "loss": 1.8859,
+      "step": 1544
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6264970576189404e-05,
+      "loss": 1.849,
+      "step": 1546
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625495669896484e-05,
+      "loss": 1.9163,
+      "step": 1548
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6244932507587667e-05,
+      "loss": 1.8745,
+      "step": 1550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.9016,
+      "step": 1552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622485324851029e-05,
+      "loss": 1.9251,
+      "step": 1554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6214798213919916e-05,
+      "loss": 1.8905,
+      "step": 1556
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620473293139654e-05,
+      "loss": 1.8237,
+      "step": 1558
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.619465741753736e-05,
+      "loss": 1.8935,
+      "step": 1560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6184571688956475e-05,
+      "loss": 1.8845,
+      "step": 1562
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6174475762284815e-05,
+      "loss": 1.9002,
+      "step": 1564
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6164369654170117e-05,
+      "loss": 1.905,
+      "step": 1566
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615425338127692e-05,
+      "loss": 1.8864,
+      "step": 1568
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.614412696028652e-05,
+      "loss": 1.8777,
+      "step": 1570
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6133990407896943e-05,
+      "loss": 1.8574,
+      "step": 1572
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6123843740822934e-05,
+      "loss": 1.8752,
+      "step": 1574
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6113686975795894e-05,
+      "loss": 1.9041,
+      "step": 1576
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6103520129563902e-05,
+      "loss": 1.8679,
+      "step": 1578
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6093343218891638e-05,
+      "loss": 1.8465,
+      "step": 1580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6083156260560387e-05,
+      "loss": 1.8536,
+      "step": 1582
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607295927136801e-05,
+      "loss": 1.8119,
+      "step": 1584
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6062752268128893e-05,
+      "loss": 1.8917,
+      "step": 1586
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605253526767394e-05,
+      "loss": 1.9185,
+      "step": 1588
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6042308286850553e-05,
+      "loss": 1.881,
+      "step": 1590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603207134252257e-05,
+      "loss": 1.8416,
+      "step": 1592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6021824451570268e-05,
+      "loss": 1.9432,
+      "step": 1594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6011567630890336e-05,
+      "loss": 1.8077,
+      "step": 1596
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6001300897395817e-05,
+      "loss": 1.8498,
+      "step": 1598
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5991024268016118e-05,
+      "loss": 1.9153,
+      "step": 1600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598073775969695e-05,
+      "loss": 1.9,
+      "step": 1602
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970441389400325e-05,
+      "loss": 1.8553,
+      "step": 1604
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5960135174104504e-05,
+      "loss": 1.8686,
+      "step": 1606
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594981913080399e-05,
+      "loss": 1.8364,
+      "step": 1608
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.59394932765095e-05,
+      "loss": 1.9332,
+      "step": 1610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5929157628247914e-05,
+      "loss": 1.8831,
+      "step": 1612
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591881220306227e-05,
+      "loss": 1.8376,
+      "step": 1614
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908457018011723e-05,
+      "loss": 1.921,
+      "step": 1616
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5898092090171527e-05,
+      "loss": 1.8809,
+      "step": 1618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5887717436633003e-05,
+      "loss": 1.8824,
+      "step": 1620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5877333074503502e-05,
+      "loss": 1.8498,
+      "step": 1622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5866939020906378e-05,
+      "loss": 1.8118,
+      "step": 1624
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585653529298099e-05,
+      "loss": 1.8527,
+      "step": 1626
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846121907882627e-05,
+      "loss": 1.888,
+      "step": 1628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.583569888278251e-05,
+      "loss": 1.8718,
+      "step": 1630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5825266234867755e-05,
+      "loss": 1.8921,
+      "step": 1632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581482398134135e-05,
+      "loss": 1.865,
+      "step": 1634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580437213942212e-05,
+      "loss": 1.9139,
+      "step": 1636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5793910726344693e-05,
+      "loss": 1.8188,
+      "step": 1638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578343975935949e-05,
+      "loss": 1.9076,
+      "step": 1640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5772959255732678e-05,
+      "loss": 1.8577,
+      "step": 1642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5762469232746164e-05,
+      "loss": 1.9053,
+      "step": 1644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5751969707697528e-05,
+      "loss": 1.8732,
+      "step": 1646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574146069790004e-05,
+      "loss": 1.8364,
+      "step": 1648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57309422206826e-05,
+      "loss": 1.8468,
+      "step": 1650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.572041429338972e-05,
+      "loss": 1.9047,
+      "step": 1652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.57098769333815e-05,
+      "loss": 1.9308,
+      "step": 1654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5699330158033586e-05,
+      "loss": 1.8317,
+      "step": 1656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568877398473715e-05,
+      "loss": 1.8465,
+      "step": 1658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5678208430898874e-05,
+      "loss": 1.8262,
+      "step": 1660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5667633513940896e-05,
+      "loss": 1.8662,
+      "step": 1662
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565704925130079e-05,
+      "loss": 1.8938,
+      "step": 1664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5646455660431552e-05,
+      "loss": 1.8407,
+      "step": 1666
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5635852758801553e-05,
+      "loss": 1.9128,
+      "step": 1668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5625240563894516e-05,
+      "loss": 1.8344,
+      "step": 1670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5614619093209498e-05,
+      "loss": 1.8707,
+      "step": 1672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5603988364260836e-05,
+      "loss": 1.8308,
+      "step": 1674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5593348394578142e-05,
+      "loss": 1.846,
+      "step": 1676
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558269920170627e-05,
+      "loss": 1.8492,
+      "step": 1678
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5572040803205276e-05,
+      "loss": 1.8695,
+      "step": 1680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5561373216650393e-05,
+      "loss": 1.8366,
+      "step": 1682
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5550696459632013e-05,
+      "loss": 1.8485,
+      "step": 1684
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554001054975565e-05,
+      "loss": 1.8534,
+      "step": 1686
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5529315504641896e-05,
+      "loss": 1.9195,
+      "step": 1688
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5518611341926426e-05,
+      "loss": 1.8489,
+      "step": 1690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5507898079259938e-05,
+      "loss": 1.8777,
+      "step": 1692
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5497175734308135e-05,
+      "loss": 1.8732,
+      "step": 1694
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5486444324751705e-05,
+      "loss": 1.8554,
+      "step": 1696
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547570386828627e-05,
+      "loss": 1.8472,
+      "step": 1698
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5464954382622383e-05,
+      "loss": 1.8893,
+      "step": 1700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5454195885485476e-05,
+      "loss": 1.8841,
+      "step": 1702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544342839461585e-05,
+      "loss": 1.8779,
+      "step": 1704
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5432651927768633e-05,
+      "loss": 1.8164,
+      "step": 1706
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542186650271374e-05,
+      "loss": 1.8556,
+      "step": 1708
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541107213723588e-05,
+      "loss": 1.8268,
+      "step": 1710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5400268849134495e-05,
+      "loss": 1.8364,
+      "step": 1712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5389456656223728e-05,
+      "loss": 1.8285,
+      "step": 1714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5378635576332426e-05,
+      "loss": 1.8637,
+      "step": 1716
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536780562730408e-05,
+      "loss": 1.8795,
+      "step": 1718
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5356966826996806e-05,
+      "loss": 1.8281,
+      "step": 1720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5346119193283314e-05,
+      "loss": 1.7917,
+      "step": 1722
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5335262744050883e-05,
+      "loss": 1.8092,
+      "step": 1724
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532439749720133e-05,
+      "loss": 1.8648,
+      "step": 1726
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5313523470650972e-05,
+      "loss": 1.8402,
+      "step": 1728
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530264068233061e-05,
+      "loss": 1.85,
+      "step": 1730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5291749150185493e-05,
+      "loss": 1.8404,
+      "step": 1732
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528084889217528e-05,
+      "loss": 1.8103,
+      "step": 1734
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526993992627403e-05,
+      "loss": 1.8754,
+      "step": 1736
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525902227047015e-05,
+      "loss": 1.8897,
+      "step": 1738
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5248095942766385e-05,
+      "loss": 1.8159,
+      "step": 1740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5237160961179776e-05,
+      "loss": 1.8398,
+      "step": 1742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5226217343741631e-05,
+      "loss": 1.8377,
+      "step": 1744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5215265108497503e-05,
+      "loss": 1.8445,
+      "step": 1746
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5204304273507158e-05,
+      "loss": 1.8961,
+      "step": 1748
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193334856844528e-05,
+      "loss": 1.8835,
+      "step": 1750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5182356876597713e-05,
+      "loss": 1.8358,
+      "step": 1752
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5171370350868926e-05,
+      "loss": 1.8507,
+      "step": 1754
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5160375297774473e-05,
+      "loss": 1.8028,
+      "step": 1756
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149371735444713e-05,
+      "loss": 1.8311,
+      "step": 1758
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5138359682024053e-05,
+      "loss": 1.8867,
+      "step": 1760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5127339155670883e-05,
+      "loss": 1.8758,
+      "step": 1762
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5116310174557582e-05,
+      "loss": 1.8581,
+      "step": 1764
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510527275687046e-05,
+      "loss": 1.8084,
+      "step": 1766
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5094226920809739e-05,
+      "loss": 1.8172,
+      "step": 1768
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5083172684589516e-05,
+      "loss": 1.8039,
+      "step": 1770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5072110066437764e-05,
+      "loss": 1.8359,
+      "step": 1772
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5061039084596249e-05,
+      "loss": 1.8925,
+      "step": 1774
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5049959757320546e-05,
+      "loss": 1.7985,
+      "step": 1776
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5038872102879982e-05,
+      "loss": 1.8193,
+      "step": 1778
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5027776139557624e-05,
+      "loss": 1.8556,
+      "step": 1780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5016671885650227e-05,
+      "loss": 1.8331,
+      "step": 1782
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5005559359468235e-05,
+      "loss": 1.8088,
+      "step": 1784
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4994438579335713e-05,
+      "loss": 1.813,
+      "step": 1786
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4983309563590347e-05,
+      "loss": 1.866,
+      "step": 1788
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4972172330583406e-05,
+      "loss": 1.8127,
+      "step": 1790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4961026898679703e-05,
+      "loss": 1.8615,
+      "step": 1792
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494987328625757e-05,
+      "loss": 1.8595,
+      "step": 1794
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4938711511708834e-05,
+      "loss": 1.8932,
+      "step": 1796
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4927541593438775e-05,
+      "loss": 1.8665,
+      "step": 1798
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4916363549866109e-05,
+      "loss": 1.874,
+      "step": 1800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.490517739942294e-05,
+      "loss": 1.818,
+      "step": 1802
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4893983160554746e-05,
+      "loss": 1.8448,
+      "step": 1804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4882780851720344e-05,
+      "loss": 1.8519,
+      "step": 1806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4871570491391858e-05,
+      "loss": 1.8886,
+      "step": 1808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4860352098054682e-05,
+      "loss": 1.8753,
+      "step": 1810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4849125690207465e-05,
+      "loss": 1.8978,
+      "step": 1812
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483789128636206e-05,
+      "loss": 1.8951,
+      "step": 1814
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826648905043521e-05,
+      "loss": 1.8288,
+      "step": 1816
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4815398564790043e-05,
+      "loss": 1.8613,
+      "step": 1818
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480414028415295e-05,
+      "loss": 1.9096,
+      "step": 1820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4792874081696656e-05,
+      "loss": 1.9117,
+      "step": 1822
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4781599975998648e-05,
+      "loss": 1.8697,
+      "step": 1824
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4770317985649433e-05,
+      "loss": 1.8572,
+      "step": 1826
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4759028129252529e-05,
+      "loss": 1.8467,
+      "step": 1828
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4747730425424412e-05,
+      "loss": 1.9164,
+      "step": 1830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4736424892794511e-05,
+      "loss": 1.7991,
+      "step": 1832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.472511155000516e-05,
+      "loss": 1.883,
+      "step": 1834
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.471379041571157e-05,
+      "loss": 1.8497,
+      "step": 1836
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4702461508581802e-05,
+      "loss": 1.8542,
+      "step": 1838
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4691124847296732e-05,
+      "loss": 1.81,
+      "step": 1840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4679780450550023e-05,
+      "loss": 1.8004,
+      "step": 1842
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4668428337048094e-05,
+      "loss": 1.8258,
+      "step": 1844
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657068525510092e-05,
+      "loss": 1.853,
+      "step": 1846
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4645701034667848e-05,
+      "loss": 1.839,
+      "step": 1848
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634325883265862e-05,
+      "loss": 1.8676,
+      "step": 1850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4622943090061274e-05,
+      "loss": 1.8516,
+      "step": 1852
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.461155267382381e-05,
+      "loss": 1.8357,
+      "step": 1854
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4600154653335772e-05,
+      "loss": 1.8383,
+      "step": 1856
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4588749047392004e-05,
+      "loss": 1.8048,
+      "step": 1858
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457733587479986e-05,
+      "loss": 1.8377,
+      "step": 1860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4565915154379163e-05,
+      "loss": 1.8478,
+      "step": 1862
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4554486904962184e-05,
+      "loss": 1.8374,
+      "step": 1864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4543051145393614e-05,
+      "loss": 1.8672,
+      "step": 1866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4531607894530529e-05,
+      "loss": 1.8479,
+      "step": 1868
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4520157171242346e-05,
+      "loss": 1.8766,
+      "step": 1870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4508698994410817e-05,
+      "loss": 1.8179,
+      "step": 1872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4497233382929976e-05,
+      "loss": 1.8947,
+      "step": 1874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4485760355706125e-05,
+      "loss": 1.8374,
+      "step": 1876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.447427993165778e-05,
+      "loss": 1.8032,
+      "step": 1878
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4462792129715672e-05,
+      "loss": 1.8278,
+      "step": 1880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4451296968822686e-05,
+      "loss": 1.8654,
+      "step": 1882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4439794467933838e-05,
+      "loss": 1.8621,
+      "step": 1884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4428284646016262e-05,
+      "loss": 1.8167,
+      "step": 1886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4416767522049154e-05,
+      "loss": 1.8495,
+      "step": 1888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4405243115023749e-05,
+      "loss": 1.8108,
+      "step": 1890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4393711443943294e-05,
+      "loss": 1.8269,
+      "step": 1892
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4382172527823022e-05,
+      "loss": 1.7924,
+      "step": 1894
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43706263856901e-05,
+      "loss": 1.8068,
+      "step": 1896
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.435907303658362e-05,
+      "loss": 1.7942,
+      "step": 1898
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4347512499554547e-05,
+      "loss": 1.8767,
+      "step": 1900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4335944793665714e-05,
+      "loss": 1.8073,
+      "step": 1902
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4324369937991765e-05,
+      "loss": 1.8657,
+      "step": 1904
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4312787951619132e-05,
+      "loss": 1.7788,
+      "step": 1906
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4301198853646007e-05,
+      "loss": 1.8642,
+      "step": 1908
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428960266318232e-05,
+      "loss": 1.8309,
+      "step": 1910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427799939934968e-05,
+      "loss": 1.8407,
+      "step": 1912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4266389081281364e-05,
+      "loss": 1.8705,
+      "step": 1914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425477172812229e-05,
+      "loss": 1.8459,
+      "step": 1916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4243147359028968e-05,
+      "loss": 1.8991,
+      "step": 1918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231515993169475e-05,
+      "loss": 1.7811,
+      "step": 1920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4219877649723432e-05,
+      "loss": 1.845,
+      "step": 1922
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4208232347881966e-05,
+      "loss": 1.794,
+      "step": 1924
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4196580106847671e-05,
+      "loss": 1.7898,
+      "step": 1926
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.418492094583459e-05,
+      "loss": 1.7987,
+      "step": 1928
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4173254884068173e-05,
+      "loss": 1.7728,
+      "step": 1930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161581940785252e-05,
+      "loss": 1.8395,
+      "step": 1932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4149902135234004e-05,
+      "loss": 1.8482,
+      "step": 1934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.413821548667392e-05,
+      "loss": 1.8233,
+      "step": 1936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4126522014375783e-05,
+      "loss": 1.8391,
+      "step": 1938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4114821737621615e-05,
+      "loss": 1.8511,
+      "step": 1940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410311467570467e-05,
+      "loss": 1.7979,
+      "step": 1942
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4091400847929386e-05,
+      "loss": 1.844,
+      "step": 1944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4079680273611358e-05,
+      "loss": 1.8148,
+      "step": 1946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4067952972077303e-05,
+      "loss": 1.8158,
+      "step": 1948
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4056218962665034e-05,
+      "loss": 1.8625,
+      "step": 1950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404447826472342e-05,
+      "loss": 1.7924,
+      "step": 1952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4032730897612374e-05,
+      "loss": 1.8209,
+      "step": 1954
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4020976880702786e-05,
+      "loss": 1.8038,
+      "step": 1956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4009216233376523e-05,
+      "loss": 1.7336,
+      "step": 1958
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3997448975026382e-05,
+      "loss": 1.7873,
+      "step": 1960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3985675125056069e-05,
+      "loss": 1.8773,
+      "step": 1962
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3973894702880142e-05,
+      "loss": 1.8225,
+      "step": 1964
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3962107727924015e-05,
+      "loss": 1.8429,
+      "step": 1966
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3950314219623892e-05,
+      "loss": 1.8288,
+      "step": 1968
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3938514197426765e-05,
+      "loss": 1.8682,
+      "step": 1970
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3926707680790355e-05,
+      "loss": 1.8379,
+      "step": 1972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3914894689183099e-05,
+      "loss": 1.8541,
+      "step": 1974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.39030752420841e-05,
+      "loss": 1.8831,
+      "step": 1976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3891249358983128e-05,
+      "loss": 1.8214,
+      "step": 1978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387941705938054e-05,
+      "loss": 1.8858,
+      "step": 1980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3867578362787295e-05,
+      "loss": 1.8044,
+      "step": 1982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3855733288724882e-05,
+      "loss": 1.8544,
+      "step": 1984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384388185672532e-05,
+      "loss": 1.7526,
+      "step": 1986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3832024086331104e-05,
+      "loss": 1.8112,
+      "step": 1988
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3820159997095188e-05,
+      "loss": 1.8115,
+      "step": 1990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3808289608580936e-05,
+      "loss": 1.8223,
+      "step": 1992
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3796412940362107e-05,
+      "loss": 1.8356,
+      "step": 1994
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3784530012022815e-05,
+      "loss": 1.8505,
+      "step": 1996
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.377264084315749e-05,
+      "loss": 1.8027,
+      "step": 1998
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3760745453370856e-05,
+      "loss": 1.8296,
+      "step": 2000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.37488438622779e-05,
+      "loss": 1.8194,
+      "step": 2002
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3736936089503826e-05,
+      "loss": 1.8535,
+      "step": 2004
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3725022154684035e-05,
+      "loss": 1.7733,
+      "step": 2006
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3713102077464093e-05,
+      "loss": 1.8151,
+      "step": 2008
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370117587749969e-05,
+      "loss": 1.8246,
+      "step": 2010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3689243574456608e-05,
+      "loss": 1.7683,
+      "step": 2012
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3677305188010705e-05,
+      "loss": 1.8437,
+      "step": 2014
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.366536073784786e-05,
+      "loss": 1.7182,
+      "step": 2016
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 1.8429,
+      "step": 2018
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3641453725164828e-05,
+      "loss": 1.7991,
+      "step": 2020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3629491202066277e-05,
+      "loss": 1.8065,
+      "step": 2022
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3617522694093969e-05,
+      "loss": 1.8192,
+      "step": 2024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3605548220983466e-05,
+      "loss": 1.8119,
+      "step": 2026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3593567802480145e-05,
+      "loss": 1.8319,
+      "step": 2028
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3581581458339208e-05,
+      "loss": 1.8146,
+      "step": 2030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3569589208325612e-05,
+      "loss": 1.7612,
+      "step": 2032
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3557591072214056e-05,
+      "loss": 1.8532,
+      "step": 2034
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3545587069788946e-05,
+      "loss": 1.7898,
+      "step": 2036
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3533577220844367e-05,
+      "loss": 1.8083,
+      "step": 2038
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3521561545184035e-05,
+      "loss": 1.7325,
+      "step": 2040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3509540062621284e-05,
+      "loss": 1.8481,
+      "step": 2042
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3497512792979013e-05,
+      "loss": 1.7899,
+      "step": 2044
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3485479756089668e-05,
+      "loss": 1.7916,
+      "step": 2046
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3473440971795207e-05,
+      "loss": 1.7878,
+      "step": 2048
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3461396459947066e-05,
+      "loss": 1.8372,
+      "step": 2050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3449346240406122e-05,
+      "loss": 1.8075,
+      "step": 2052
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3437290333042666e-05,
+      "loss": 1.8297,
+      "step": 2054
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3425228757736362e-05,
+      "loss": 1.7481,
+      "step": 2056
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3413161534376231e-05,
+      "loss": 1.8172,
+      "step": 2058
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.34010886828606e-05,
+      "loss": 1.7674,
+      "step": 2060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3389010223097075e-05,
+      "loss": 1.8279,
+      "step": 2062
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3376926175002511e-05,
+      "loss": 1.7555,
+      "step": 2064
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3364836558502991e-05,
+      "loss": 1.7728,
+      "step": 2066
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335274139353376e-05,
+      "loss": 1.8201,
+      "step": 2068
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3340640700039222e-05,
+      "loss": 1.8028,
+      "step": 2070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3328534497972893e-05,
+      "loss": 1.8233,
+      "step": 2072
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3316422807297387e-05,
+      "loss": 1.7748,
+      "step": 2074
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3304305647984346e-05,
+      "loss": 1.7748,
+      "step": 2076
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3292183040014446e-05,
+      "loss": 1.7609,
+      "step": 2078
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280055003377338e-05,
+      "loss": 1.7605,
+      "step": 2080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267921558071635e-05,
+      "loss": 1.786,
+      "step": 2082
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325578272410486e-05,
+      "loss": 1.7447,
+      "step": 2084
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3243638521493425e-05,
+      "loss": 1.8691,
+      "step": 2086
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.323148897026259e-05,
+      "loss": 1.7978,
+      "step": 2088
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3219334090446443e-05,
+      "loss": 1.7611,
+      "step": 2090
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3207173902087852e-05,
+      "loss": 1.7974,
+      "step": 2092
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3195008425238444e-05,
+      "loss": 1.8525,
+      "step": 2094
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.318283767995856e-05,
+      "loss": 1.7838,
+      "step": 2096
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317066168631723e-05,
+      "loss": 1.7623,
+      "step": 2098
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3158480464392146e-05,
+      "loss": 1.8042,
+      "step": 2100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3146294034269611e-05,
+      "loss": 1.8428,
+      "step": 2102
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313410241604452e-05,
+      "loss": 1.789,
+      "step": 2104
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3121905629820324e-05,
+      "loss": 1.7368,
+      "step": 2106
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3109703695708997e-05,
+      "loss": 1.7949,
+      "step": 2108
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3097496633830999e-05,
+      "loss": 1.7989,
+      "step": 2110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3085284464315245e-05,
+      "loss": 1.7768,
+      "step": 2112
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3073067207299072e-05,
+      "loss": 1.8308,
+      "step": 2114
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3060844882928212e-05,
+      "loss": 1.7505,
+      "step": 2116
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3048617511356745e-05,
+      "loss": 1.7858,
+      "step": 2118
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3036385112747077e-05,
+      "loss": 1.8728,
+      "step": 2120
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3024147707269905e-05,
+      "loss": 1.8082,
+      "step": 2122
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3011905315104176e-05,
+      "loss": 1.7622,
+      "step": 2124
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2999657956437076e-05,
+      "loss": 1.7773,
+      "step": 2126
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2987405651463954e-05,
+      "loss": 1.8094,
+      "step": 2128
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2975148420388339e-05,
+      "loss": 1.8049,
+      "step": 2130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2962886283421872e-05,
+      "loss": 1.8057,
+      "step": 2132
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2950619260784285e-05,
+      "loss": 1.7688,
+      "step": 2134
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2938347372703368e-05,
+      "loss": 1.8297,
+      "step": 2136
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2926070639414934e-05,
+      "loss": 1.8429,
+      "step": 2138
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2913789081162778e-05,
+      "loss": 1.7641,
+      "step": 2140
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2901502718198664e-05,
+      "loss": 1.7219,
+      "step": 2142
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288921157078227e-05,
+      "loss": 1.7942,
+      "step": 2144
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2876915659181168e-05,
+      "loss": 1.7802,
+      "step": 2146
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.286461500367078e-05,
+      "loss": 1.8482,
+      "step": 2148
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2852309624534353e-05,
+      "loss": 1.8071,
+      "step": 2150
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2839999542062929e-05,
+      "loss": 1.8093,
+      "step": 2152
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2827684776555299e-05,
+      "loss": 1.8156,
+      "step": 2154
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2815365348317976e-05,
+      "loss": 1.7951,
+      "step": 2156
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2803041277665162e-05,
+      "loss": 1.7487,
+      "step": 2158
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2790712584918723e-05,
+      "loss": 1.7973,
+      "step": 2160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2778379290408131e-05,
+      "loss": 1.7714,
+      "step": 2162
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2766041414470458e-05,
+      "loss": 1.7598,
+      "step": 2164
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2753698977450326e-05,
+      "loss": 1.7782,
+      "step": 2166
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2741351999699881e-05,
+      "loss": 1.796,
+      "step": 2168
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272900050157875e-05,
+      "loss": 1.8066,
+      "step": 2170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2716644503454024e-05,
+      "loss": 1.8019,
+      "step": 2172
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2704284025700204e-05,
+      "loss": 1.7,
+      "step": 2174
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2691919088699184e-05,
+      "loss": 1.7832,
+      "step": 2176
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267954971284021e-05,
+      "loss": 1.8352,
+      "step": 2178
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2667175918519846e-05,
+      "loss": 1.8173,
+      "step": 2180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2654797726141945e-05,
+      "loss": 1.7778,
+      "step": 2182
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2642415156117605e-05,
+      "loss": 1.7443,
+      "step": 2184
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2630028228865152e-05,
+      "loss": 1.7725,
+      "step": 2186
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2617636964810091e-05,
+      "loss": 1.8178,
+      "step": 2188
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2605241384385081e-05,
+      "loss": 1.8251,
+      "step": 2190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2592841508029892e-05,
+      "loss": 1.74,
+      "step": 2192
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2580437356191388e-05,
+      "loss": 1.8057,
+      "step": 2194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2568028949323479e-05,
+      "loss": 1.8054,
+      "step": 2196
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2555616307887087e-05,
+      "loss": 1.7365,
+      "step": 2198
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2543199452350118e-05,
+      "loss": 1.7347,
+      "step": 2200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2530778403187434e-05,
+      "loss": 1.8453,
+      "step": 2202
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2518353180880806e-05,
+      "loss": 1.8584,
+      "step": 2204
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2505923805918884e-05,
+      "loss": 1.8355,
+      "step": 2206
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.249349029879717e-05,
+      "loss": 1.7777,
+      "step": 2208
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.248105268001798e-05,
+      "loss": 1.7867,
+      "step": 2210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246861097009041e-05,
+      "loss": 1.7458,
+      "step": 2212
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2456165189530299e-05,
+      "loss": 1.7831,
+      "step": 2214
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2443715358860198e-05,
+      "loss": 1.7584,
+      "step": 2216
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2431261498609337e-05,
+      "loss": 1.8358,
+      "step": 2218
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2418803629313601e-05,
+      "loss": 1.7419,
+      "step": 2220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2406341771515466e-05,
+      "loss": 1.7388,
+      "step": 2222
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2393875945764002e-05,
+      "loss": 1.7788,
+      "step": 2224
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2381406172614811e-05,
+      "loss": 1.7508,
+      "step": 2226
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2368932472630018e-05,
+      "loss": 1.845,
+      "step": 2228
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2356454866378202e-05,
+      "loss": 1.8324,
+      "step": 2230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2343973374434404e-05,
+      "loss": 1.8258,
+      "step": 2232
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2331488017380058e-05,
+      "loss": 1.7988,
+      "step": 2234
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2318998815802981e-05,
+      "loss": 1.756,
+      "step": 2236
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.230650579029732e-05,
+      "loss": 1.7886,
+      "step": 2238
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2294008961463539e-05,
+      "loss": 1.7387,
+      "step": 2240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2281508349908365e-05,
+      "loss": 1.7451,
+      "step": 2242
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226900397624476e-05,
+      "loss": 1.7944,
+      "step": 2244
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2256495861091902e-05,
+      "loss": 1.8012,
+      "step": 2246
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224398402507513e-05,
+      "loss": 1.7951,
+      "step": 2248
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2231468488825913e-05,
+      "loss": 1.7507,
+      "step": 2250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2218949272981834e-05,
+      "loss": 1.7414,
+      "step": 2252
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2206426398186534e-05,
+      "loss": 1.7697,
+      "step": 2254
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.21938998850897e-05,
+      "loss": 1.7321,
+      "step": 2256
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2181369754346994e-05,
+      "loss": 1.8177,
+      "step": 2258
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2168836026620075e-05,
+      "loss": 1.7292,
+      "step": 2260
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2156298722576501e-05,
+      "loss": 1.7524,
+      "step": 2262
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2143757862889756e-05,
+      "loss": 1.7835,
+      "step": 2264
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2131213468239168e-05,
+      "loss": 1.7536,
+      "step": 2266
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2118665559309906e-05,
+      "loss": 1.7644,
+      "step": 2268
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2106114156792916e-05,
+      "loss": 1.7761,
+      "step": 2270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2093559281384933e-05,
+      "loss": 1.8112,
+      "step": 2272
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.208100095378839e-05,
+      "loss": 1.7309,
+      "step": 2274
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2068439194711434e-05,
+      "loss": 1.7684,
+      "step": 2276
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2055874024867854e-05,
+      "loss": 1.8481,
+      "step": 2278
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2043305464977077e-05,
+      "loss": 1.7995,
+      "step": 2280
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2030733535764108e-05,
+      "loss": 1.7843,
+      "step": 2282
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.201815825795952e-05,
+      "loss": 1.7852,
+      "step": 2284
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2005579652299395e-05,
+      "loss": 1.7775,
+      "step": 2286
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1992997739525312e-05,
+      "loss": 1.7753,
+      "step": 2288
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1980412540384299e-05,
+      "loss": 1.7892,
+      "step": 2290
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1967824075628807e-05,
+      "loss": 1.7624,
+      "step": 2292
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1955232366016665e-05,
+      "loss": 1.744,
+      "step": 2294
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.194263743231106e-05,
+      "loss": 1.8016,
+      "step": 2296
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1930039295280489e-05,
+      "loss": 1.7463,
+      "step": 2298
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1917437975698736e-05,
+      "loss": 1.7324,
+      "step": 2300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1904833494344832e-05,
+      "loss": 1.7396,
+      "step": 2302
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1892225872003016e-05,
+      "loss": 1.7611,
+      "step": 2304
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1879615129462718e-05,
+      "loss": 1.7851,
+      "step": 2306
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1867001287518503e-05,
+      "loss": 1.7498,
+      "step": 2308
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1854384366970046e-05,
+      "loss": 1.8009,
+      "step": 2310
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1841764388622108e-05,
+      "loss": 1.7646,
+      "step": 2312
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1829141373284483e-05,
+      "loss": 1.7436,
+      "step": 2314
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.181651534177198e-05,
+      "loss": 1.7687,
+      "step": 2316
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1803886314904374e-05,
+      "loss": 1.7879,
+      "step": 2318
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1791254313506388e-05,
+      "loss": 1.7438,
+      "step": 2320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1778619358407637e-05,
+      "loss": 1.7935,
+      "step": 2322
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1765981470442624e-05,
+      "loss": 1.7744,
+      "step": 2324
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175334067045067e-05,
+      "loss": 1.8138,
+      "step": 2326
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1740696979275917e-05,
+      "loss": 1.7818,
+      "step": 2328
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1728050417767251e-05,
+      "loss": 1.8081,
+      "step": 2330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1715401006778316e-05,
+      "loss": 1.75,
+      "step": 2332
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1702748767167434e-05,
+      "loss": 1.7785,
+      "step": 2334
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1690093719797607e-05,
+      "loss": 1.7871,
+      "step": 2336
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1677435885536453e-05,
+      "loss": 1.7847,
+      "step": 2338
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1664775285256198e-05,
+      "loss": 1.7799,
+      "step": 2340
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.165211193983362e-05,
+      "loss": 1.762,
+      "step": 2342
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163944587015003e-05,
+      "loss": 1.7783,
+      "step": 2344
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1626777097091223e-05,
+      "loss": 1.8216,
+      "step": 2346
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1614105641547462e-05,
+      "loss": 1.7763,
+      "step": 2348
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1601431524413424e-05,
+      "loss": 1.814,
+      "step": 2350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1588754766588188e-05,
+      "loss": 1.7817,
+      "step": 2352
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1576075388975165e-05,
+      "loss": 1.7392,
+      "step": 2354
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1563393412482109e-05,
+      "loss": 1.8065,
+      "step": 2356
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1550708858021046e-05,
+      "loss": 1.726,
+      "step": 2358
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.153802174650826e-05,
+      "loss": 1.7778,
+      "step": 2360
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1525332098864246e-05,
+      "loss": 1.7496,
+      "step": 2362
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1512639936013681e-05,
+      "loss": 1.7715,
+      "step": 2364
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1499945278885396e-05,
+      "loss": 1.7758,
+      "step": 2366
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.148724814841233e-05,
+      "loss": 1.83,
+      "step": 2368
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1474548565531496e-05,
+      "loss": 1.8003,
+      "step": 2370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1461846551183965e-05,
+      "loss": 1.7823,
+      "step": 2372
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1449142126314796e-05,
+      "loss": 1.7752,
+      "step": 2374
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1436435311873047e-05,
+      "loss": 1.742,
+      "step": 2376
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1423726128811698e-05,
+      "loss": 1.7438,
+      "step": 2378
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1411014598087645e-05,
+      "loss": 1.8367,
+      "step": 2380
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1398300740661647e-05,
+      "loss": 1.7874,
+      "step": 2382
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1385584577498308e-05,
+      "loss": 1.7792,
+      "step": 2384
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1372866129566027e-05,
+      "loss": 1.81,
+      "step": 2386
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1360145417836978e-05,
+      "loss": 1.8151,
+      "step": 2388
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1347422463287058e-05,
+      "loss": 1.7891,
+      "step": 2390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1334697286895873e-05,
+      "loss": 1.7095,
+      "step": 2392
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1321969909646684e-05,
+      "loss": 1.7381,
+      "step": 2394
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130924035252639e-05,
+      "loss": 1.7689,
+      "step": 2396
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1296508636525471e-05,
+      "loss": 1.8124,
+      "step": 2398
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.128377478263798e-05,
+      "loss": 1.7631,
+      "step": 2400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.127103881186149e-05,
+      "loss": 1.7579,
+      "step": 2402
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1258300745197066e-05,
+      "loss": 1.7781,
+      "step": 2404
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1245560603649224e-05,
+      "loss": 1.8085,
+      "step": 2406
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1232818408225909e-05,
+      "loss": 1.8277,
+      "step": 2408
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1220074179938452e-05,
+      "loss": 1.7505,
+      "step": 2410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1207327939801526e-05,
+      "loss": 1.7228,
+      "step": 2412
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1194579708833136e-05,
+      "loss": 1.8139,
+      "step": 2414
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1181829508054556e-05,
+      "loss": 1.7713,
+      "step": 2416
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1169077358490322e-05,
+      "loss": 1.8342,
+      "step": 2418
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1156323281168167e-05,
+      "loss": 1.7341,
+      "step": 2420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114356729711902e-05,
+      "loss": 1.7553,
+      "step": 2422
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1130809427376941e-05,
+      "loss": 1.7248,
+      "step": 2424
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.111804969297911e-05,
+      "loss": 1.7809,
+      "step": 2426
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1105288114965772e-05,
+      "loss": 1.7788,
+      "step": 2428
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1092524714380213e-05,
+      "loss": 1.7765,
+      "step": 2430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1079759512268738e-05,
+      "loss": 1.7545,
+      "step": 2432
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.10669925296806e-05,
+      "loss": 1.781,
+      "step": 2434
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1054223787668007e-05,
+      "loss": 1.7976,
+      "step": 2436
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1041453307286064e-05,
+      "loss": 1.7121,
+      "step": 2438
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1028681109592736e-05,
+      "loss": 1.8351,
+      "step": 2440
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1015907215648822e-05,
+      "loss": 1.7499,
+      "step": 2442
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1003131646517922e-05,
+      "loss": 1.7432,
+      "step": 2444
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0990354423266399e-05,
+      "loss": 1.7359,
+      "step": 2446
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0977575566963334e-05,
+      "loss": 1.7206,
+      "step": 2448
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0964795098680512e-05,
+      "loss": 1.7902,
+      "step": 2450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0952013039492369e-05,
+      "loss": 1.8185,
+      "step": 2452
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093922941047597e-05,
+      "loss": 1.7895,
+      "step": 2454
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0926444232710964e-05,
+      "loss": 1.7812,
+      "step": 2456
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0913657527279554e-05,
+      "loss": 1.738,
+      "step": 2458
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0900869315266462e-05,
+      "loss": 1.7006,
+      "step": 2460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.08880796177589e-05,
+      "loss": 1.7481,
+      "step": 2462
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0875288455846521e-05,
+      "loss": 1.7418,
+      "step": 2464
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0862495850621403e-05,
+      "loss": 1.8163,
+      "step": 2466
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0849701823177987e-05,
+      "loss": 1.75,
+      "step": 2468
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0836906394613083e-05,
+      "loss": 1.7809,
+      "step": 2470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0824109586025789e-05,
+      "loss": 1.6806,
+      "step": 2472
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0811311418517491e-05,
+      "loss": 1.7208,
+      "step": 2474
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.079851191319181e-05,
+      "loss": 1.7647,
+      "step": 2476
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.078571109115458e-05,
+      "loss": 1.7786,
+      "step": 2478
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0772908973513801e-05,
+      "loss": 1.813,
+      "step": 2480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076010558137961e-05,
+      "loss": 1.7447,
+      "step": 2482
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 1.7712,
+      "step": 2484
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0734495058082011e-05,
+      "loss": 1.7693,
+      "step": 2486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0721687969149249e-05,
+      "loss": 1.783,
+      "step": 2488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0708879690184291e-05,
+      "loss": 1.7081,
+      "step": 2490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0696070242307433e-05,
+      "loss": 1.7457,
+      "step": 2492
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0683259646640903e-05,
+      "loss": 1.7984,
+      "step": 2494
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0670447924308816e-05,
+      "loss": 1.7477,
+      "step": 2496
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.065763509643715e-05,
+      "loss": 1.7712,
+      "step": 2498
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.06448211841537e-05,
+      "loss": 1.7259,
+      "step": 2500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.063200620858805e-05,
+      "loss": 1.7306,
+      "step": 2502
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0619190190871544e-05,
+      "loss": 1.755,
+      "step": 2504
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0606373152137241e-05,
+      "loss": 1.7509,
+      "step": 2506
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.059355511351988e-05,
+      "loss": 1.785,
+      "step": 2508
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0580736096155853e-05,
+      "loss": 1.7382,
+      "step": 2510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0567916121183157e-05,
+      "loss": 1.7969,
+      "step": 2512
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0555095209741383e-05,
+      "loss": 1.7446,
+      "step": 2514
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054227338297166e-05,
+      "loss": 1.8089,
+      "step": 2516
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0529450662016621e-05,
+      "loss": 1.7457,
+      "step": 2518
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0516627068020373e-05,
+      "loss": 1.7283,
+      "step": 2520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0503802622128476e-05,
+      "loss": 1.7717,
+      "step": 2522
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049097734548788e-05,
+      "loss": 1.8138,
+      "step": 2524
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0478151259246915e-05,
+      "loss": 1.7835,
+      "step": 2526
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0465324384555233e-05,
+      "loss": 1.7515,
+      "step": 2528
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0452496742563808e-05,
+      "loss": 1.7558,
+      "step": 2530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043966835442485e-05,
+      "loss": 1.7672,
+      "step": 2532
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426839241291828e-05,
+      "loss": 1.758,
+      "step": 2534
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0414009424319387e-05,
+      "loss": 1.7799,
+      "step": 2536
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0401178924663344e-05,
+      "loss": 1.7552,
+      "step": 2538
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0388347763480633e-05,
+      "loss": 1.7437,
+      "step": 2540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0375515961929286e-05,
+      "loss": 1.8032,
+      "step": 2542
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0362683541168386e-05,
+      "loss": 1.7835,
+      "step": 2544
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0349850522358038e-05,
+      "loss": 1.783,
+      "step": 2546
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0337016926659333e-05,
+      "loss": 1.7404,
+      "step": 2548
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.032418277523432e-05,
+      "loss": 1.7514,
+      "step": 2550
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0311348089245956e-05,
+      "loss": 1.7429,
+      "step": 2552
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0298512889858082e-05,
+      "loss": 1.7311,
+      "step": 2554
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028567719823538e-05,
+      "loss": 1.8207,
+      "step": 2556
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0272841035543359e-05,
+      "loss": 1.7502,
+      "step": 2558
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0260004422948292e-05,
+      "loss": 1.7645,
+      "step": 2560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0247167381617191e-05,
+      "loss": 1.7704,
+      "step": 2562
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0234329932717788e-05,
+      "loss": 1.7943,
+      "step": 2564
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0221492097418479e-05,
+      "loss": 1.7948,
+      "step": 2566
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0208653896888296e-05,
+      "loss": 1.8134,
+      "step": 2568
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0195815352296879e-05,
+      "loss": 1.7469,
+      "step": 2570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0182976484814423e-05,
+      "loss": 1.855,
+      "step": 2572
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0170137315611677e-05,
+      "loss": 1.7474,
+      "step": 2574
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0157297865859866e-05,
+      "loss": 1.743,
+      "step": 2576
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0150878042394203e-05,
+      "loss": 1.7589,
+      "step": 2578
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.013803821151586e-05,
+      "loss": 1.7765,
+      "step": 2580
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0125198153018495e-05,
+      "loss": 1.8336,
+      "step": 2582
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0112357888074793e-05,
+      "loss": 1.7698,
+      "step": 2584
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0099517437857807e-05,
+      "loss": 1.7144,
+      "step": 2586
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0086676823540874e-05,
+      "loss": 1.7519,
+      "step": 2588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0073836066297611e-05,
+      "loss": 1.7314,
+      "step": 2590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0060995187301877e-05,
+      "loss": 1.6776,
+      "step": 2592
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0048154207727723e-05,
+      "loss": 1.7432,
+      "step": 2594
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0035313148749364e-05,
+      "loss": 1.8257,
+      "step": 2596
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0022472031541153e-05,
+      "loss": 1.7878,
+      "step": 2598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0009630877277536e-05,
+      "loss": 1.7463,
+      "step": 2600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99678970713302e-06,
+      "loss": 1.796,
+      "step": 2602
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983948542282139e-06,
+      "loss": 1.7277,
+      "step": 2604
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.971107403899417e-06,
+      "loss": 1.8215,
+      "step": 2606
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95826631315933e-06,
+      "loss": 1.8132,
+      "step": 2608
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.945425291236287e-06,
+      "loss": 1.7776,
+      "step": 2610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.932584359304571e-06,
+      "loss": 1.7158,
+      "step": 2612
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.91974353853833e-06,
+      "loss": 1.794,
+      "step": 2614
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.906902850111516e-06,
+      "loss": 1.6947,
+      "step": 2616
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894062315197872e-06,
+      "loss": 1.792,
+      "step": 2618
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.881221954970876e-06,
+      "loss": 1.7501,
+      "step": 2620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.868381790603734e-06,
+      "loss": 1.7712,
+      "step": 2622
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.855541843269313e-06,
+      "loss": 1.7685,
+      "step": 2624
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.842702134140136e-06,
+      "loss": 1.7444,
+      "step": 2626
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.829862684388323e-06,
+      "loss": 1.7449,
+      "step": 2628
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.817023515185577e-06,
+      "loss": 1.7404,
+      "step": 2630
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.804184647703125e-06,
+      "loss": 1.7476,
+      "step": 2632
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.791346103111707e-06,
+      "loss": 1.7726,
+      "step": 2634
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.778507902581523e-06,
+      "loss": 1.7875,
+      "step": 2636
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.765670067282216e-06,
+      "loss": 1.769,
+      "step": 2638
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.752832618382812e-06,
+      "loss": 1.8019,
+      "step": 2640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.739995577051715e-06,
+      "loss": 1.7486,
+      "step": 2642
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.727158964456644e-06,
+      "loss": 1.7915,
+      "step": 2644
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.71432280176462e-06,
+      "loss": 1.7486,
+      "step": 2646
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.70148711014192e-06,
+      "loss": 1.7629,
+      "step": 2648
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.688651910754047e-06,
+      "loss": 1.7383,
+      "step": 2650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.675817224765681e-06,
+      "loss": 1.7491,
+      "step": 2652
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662983073340669e-06,
+      "loss": 1.7495,
+      "step": 2654
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.650149477641965e-06,
+      "loss": 1.7128,
+      "step": 2656
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63731645883162e-06,
+      "loss": 1.7907,
+      "step": 2658
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.624484038070718e-06,
+      "loss": 1.7352,
+      "step": 2660
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.611652236519368e-06,
+      "loss": 1.8447,
+      "step": 2662
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.598821075336656e-06,
+      "loss": 1.7524,
+      "step": 2664
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.585990575680613e-06,
+      "loss": 1.7794,
+      "step": 2666
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573160758708174e-06,
+      "loss": 1.8178,
+      "step": 2668
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.560331645575152e-06,
+      "loss": 1.7415,
+      "step": 2670
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.547503257436197e-06,
+      "loss": 1.7102,
+      "step": 2672
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53467561544477e-06,
+      "loss": 1.7375,
+      "step": 2674
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52184874075309e-06,
+      "loss": 1.7768,
+      "step": 2676
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.509022654512125e-06,
+      "loss": 1.7092,
+      "step": 2678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.496197377871524e-06,
+      "loss": 1.789,
+      "step": 2680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.483372931979628e-06,
+      "loss": 1.726,
+      "step": 2682
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.470549337983382e-06,
+      "loss": 1.7759,
+      "step": 2684
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.457726617028344e-06,
+      "loss": 1.7104,
+      "step": 2686
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.444904790258618e-06,
+      "loss": 1.7524,
+      "step": 2688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.432083878816846e-06,
+      "loss": 1.7065,
+      "step": 2690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.419263903844154e-06,
+      "loss": 1.8173,
+      "step": 2692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.406444886480124e-06,
+      "loss": 1.7298,
+      "step": 2694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393626847862762e-06,
+      "loss": 1.7542,
+      "step": 2696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.380809809128455e-06,
+      "loss": 1.7389,
+      "step": 2698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.36799379141195e-06,
+      "loss": 1.747,
+      "step": 2700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.355178815846303e-06,
+      "loss": 1.7669,
+      "step": 2702
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.342364903562853e-06,
+      "loss": 1.7004,
+      "step": 2704
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.329552075691188e-06,
+      "loss": 1.7553,
+      "step": 2706
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.316740353359099e-06,
+      "loss": 1.6849,
+      "step": 2708
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.30392975769257e-06,
+      "loss": 1.7339,
+      "step": 2710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.291120309815714e-06,
+      "loss": 1.7595,
+      "step": 2712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.278312030850753e-06,
+      "loss": 1.779,
+      "step": 2714
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.265504941917992e-06,
+      "loss": 1.7351,
+      "step": 2716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 1.7455,
+      "step": 2718
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.239894418620392e-06,
+      "loss": 1.7857,
+      "step": 2720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.227091026486202e-06,
+      "loss": 1.7343,
+      "step": 2722
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.214288908845422e-06,
+      "loss": 1.7319,
+      "step": 2724
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201488086808194e-06,
+      "loss": 1.7392,
+      "step": 2726
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.188688581482514e-06,
+      "loss": 1.7911,
+      "step": 2728
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175890413974218e-06,
+      "loss": 1.7123,
+      "step": 2730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.16309360538692e-06,
+      "loss": 1.7875,
+      "step": 2732
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.150298176822013e-06,
+      "loss": 1.7639,
+      "step": 2734
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.1375041493786e-06,
+      "loss": 1.7093,
+      "step": 2736
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.12471154415348e-06,
+      "loss": 1.7514,
+      "step": 2738
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.111920382241102e-06,
+      "loss": 1.7498,
+      "step": 2740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.099130684733541e-06,
+      "loss": 1.714,
+      "step": 2742
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.086342472720451e-06,
+      "loss": 1.7036,
+      "step": 2744
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.073555767289042e-06,
+      "loss": 1.7382,
+      "step": 2746
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.060770589524031e-06,
+      "loss": 1.7491,
+      "step": 2748
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.047986960507633e-06,
+      "loss": 1.7033,
+      "step": 2750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.035204901319493e-06,
+      "loss": 1.7809,
+      "step": 2752
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.022424433036669e-06,
+      "loss": 1.7679,
+      "step": 2754
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.009645576733604e-06,
+      "loss": 1.7468,
+      "step": 2756
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99686835348208e-06,
+      "loss": 1.7357,
+      "step": 2758
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98409278435118e-06,
+      "loss": 1.7597,
+      "step": 2760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.97131889040727e-06,
+      "loss": 1.7333,
+      "step": 2762
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.958546692713939e-06,
+      "loss": 1.7172,
+      "step": 2764
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94577621233199e-06,
+      "loss": 1.7551,
+      "step": 2766
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.933007470319403e-06,
+      "loss": 1.7344,
+      "step": 2768
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.920240487731267e-06,
+      "loss": 1.7727,
+      "step": 2770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.90747528561979e-06,
+      "loss": 1.7703,
+      "step": 2772
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.894711885034233e-06,
+      "loss": 1.6647,
+      "step": 2774
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.881950307020894e-06,
+      "loss": 1.7818,
+      "step": 2776
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.869190572623062e-06,
+      "loss": 1.7615,
+      "step": 2778
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.856432702880984e-06,
+      "loss": 1.7353,
+      "step": 2780
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.843676718831836e-06,
+      "loss": 1.7476,
+      "step": 2782
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.830922641509683e-06,
+      "loss": 1.7845,
+      "step": 2784
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.818170491945444e-06,
+      "loss": 1.7123,
+      "step": 2786
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.805420291166868e-06,
+      "loss": 1.719,
+      "step": 2788
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.792672060198475e-06,
+      "loss": 1.7755,
+      "step": 2790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.77992582006155e-06,
+      "loss": 1.694,
+      "step": 2792
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.767181591774093e-06,
+      "loss": 1.7741,
+      "step": 2794
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.754439396350777e-06,
+      "loss": 1.7179,
+      "step": 2796
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74169925480294e-06,
+      "loss": 1.6854,
+      "step": 2798
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.728961188138512e-06,
+      "loss": 1.6929,
+      "step": 2800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.71622521736202e-06,
+      "loss": 1.7152,
+      "step": 2802
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.703491363474532e-06,
+      "loss": 1.7698,
+      "step": 2804
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.690759647473614e-06,
+      "loss": 1.7353,
+      "step": 2806
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.678030090353317e-06,
+      "loss": 1.7117,
+      "step": 2808
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.665302713104128e-06,
+      "loss": 1.7484,
+      "step": 2810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.652577536712943e-06,
+      "loss": 1.7439,
+      "step": 2812
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.639854582163027e-06,
+      "loss": 1.7331,
+      "step": 2814
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.627133870433975e-06,
+      "loss": 1.7582,
+      "step": 2816
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620774362182389e-06,
+      "loss": 1.79,
+      "step": 2818
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.608057054013293e-06,
+      "loss": 1.7458,
+      "step": 2820
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.595342041097812e-06,
+      "loss": 1.7907,
+      "step": 2822
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.582629344402449e-06,
+      "loss": 1.7571,
+      "step": 2824
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.569918984889896e-06,
+      "loss": 1.7103,
+      "step": 2826
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.557210983518977e-06,
+      "loss": 1.6854,
+      "step": 2828
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.544505361244642e-06,
+      "loss": 1.7454,
+      "step": 2830
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.53180213901791e-06,
+      "loss": 1.7574,
+      "step": 2832
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.519101337785844e-06,
+      "loss": 1.6922,
+      "step": 2834
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50640297849151e-06,
+      "loss": 1.7222,
+      "step": 2836
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.49370708207396e-06,
+      "loss": 1.7379,
+      "step": 2838
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.481013669468171e-06,
+      "loss": 1.772,
+      "step": 2840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.46832276160504e-06,
+      "loss": 1.8053,
+      "step": 2842
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.455634379411314e-06,
+      "loss": 1.7427,
+      "step": 2844
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.442948543809592e-06,
+      "loss": 1.8353,
+      "step": 2846
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.430265275718263e-06,
+      "loss": 1.7224,
+      "step": 2848
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41758459605149e-06,
+      "loss": 1.7413,
+      "step": 2850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.404906525719159e-06,
+      "loss": 1.7279,
+      "step": 2852
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.392231085626861e-06,
+      "loss": 1.6568,
+      "step": 2854
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.379558296675855e-06,
+      "loss": 1.7546,
+      "step": 2856
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.366888179763012e-06,
+      "loss": 1.7959,
+      "step": 2858
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.354220755780809e-06,
+      "loss": 1.7552,
+      "step": 2860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.34155604561728e-06,
+      "loss": 1.7453,
+      "step": 2862
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.328894070155981e-06,
+      "loss": 1.7786,
+      "step": 2864
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.316234850275967e-06,
+      "loss": 1.7281,
+      "step": 2866
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.303578406851733e-06,
+      "loss": 1.7656,
+      "step": 2868
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.290924760753212e-06,
+      "loss": 1.7311,
+      "step": 2870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.27827393284572e-06,
+      "loss": 1.6912,
+      "step": 2872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.265625943989921e-06,
+      "loss": 1.7453,
+      "step": 2874
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.252980815041803e-06,
+      "loss": 1.7382,
+      "step": 2876
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.240338566852634e-06,
+      "loss": 1.7499,
+      "step": 2878
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.227699220268934e-06,
+      "loss": 1.7347,
+      "step": 2880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.21506279613244e-06,
+      "loss": 1.7771,
+      "step": 2882
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.202429315280061e-06,
+      "loss": 1.706,
+      "step": 2884
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18979879854387e-06,
+      "loss": 1.7935,
+      "step": 2886
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.17717126675103e-06,
+      "loss": 1.741,
+      "step": 2888
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.164546740723802e-06,
+      "loss": 1.773,
+      "step": 2890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.151925241279486e-06,
+      "loss": 1.7348,
+      "step": 2892
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.139306789230377e-06,
+      "loss": 1.6834,
+      "step": 2894
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.126691405383763e-06,
+      "loss": 1.7802,
+      "step": 2896
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.114079110541865e-06,
+      "loss": 1.7219,
+      "step": 2898
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.101469925501809e-06,
+      "loss": 1.773,
+      "step": 2900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.088863871055596e-06,
+      "loss": 1.6986,
+      "step": 2902
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.07626096799006e-06,
+      "loss": 1.7483,
+      "step": 2904
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.063661237086845e-06,
+      "loss": 1.7567,
+      "step": 2906
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.051064699122365e-06,
+      "loss": 1.7786,
+      "step": 2908
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03847137486776e-06,
+      "loss": 1.7074,
+      "step": 2910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.02588128508888e-06,
+      "loss": 1.7351,
+      "step": 2912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.013294450546232e-06,
+      "loss": 1.7444,
+      "step": 2914
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.000710891994967e-06,
+      "loss": 1.7653,
+      "step": 2916
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.988130630184826e-06,
+      "loss": 1.742,
+      "step": 2918
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.975553685860113e-06,
+      "loss": 1.8059,
+      "step": 2920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.962980079759665e-06,
+      "loss": 1.6869,
+      "step": 2922
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.950409832616818e-06,
+      "loss": 1.7694,
+      "step": 2924
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.937842965159363e-06,
+      "loss": 1.7488,
+      "step": 2926
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92527949810952e-06,
+      "loss": 1.714,
+      "step": 2928
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.912719452183897e-06,
+      "loss": 1.7032,
+      "step": 2930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.900162848093472e-06,
+      "loss": 1.7632,
+      "step": 2932
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.887609706543538e-06,
+      "loss": 1.7582,
+      "step": 2934
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.875060048233682e-06,
+      "loss": 1.7309,
+      "step": 2936
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.862513893857749e-06,
+      "loss": 1.7053,
+      "step": 2938
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.849971264103797e-06,
+      "loss": 1.7604,
+      "step": 2940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.837432179654087e-06,
+      "loss": 1.7344,
+      "step": 2942
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.824896661185026e-06,
+      "loss": 1.7125,
+      "step": 2944
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.812364729367136e-06,
+      "loss": 1.726,
+      "step": 2946
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799836404865035e-06,
+      "loss": 1.75,
+      "step": 2948
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.78731170833738e-06,
+      "loss": 1.7155,
+      "step": 2950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 1.6769,
+      "step": 2952
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.762273281810135e-06,
+      "loss": 1.6764,
+      "step": 2954
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74975959309782e-06,
+      "loss": 1.7795,
+      "step": 2956
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.737249614934453e-06,
+      "loss": 1.6796,
+      "step": 2958
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72474336794844e-06,
+      "loss": 1.8113,
+      "step": 2960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.712240872762042e-06,
+      "loss": 1.6718,
+      "step": 2962
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.699742149991332e-06,
+      "loss": 1.7547,
+      "step": 2964
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687247220246161e-06,
+      "loss": 1.7054,
+      "step": 2966
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67475610413013e-06,
+      "loss": 1.7756,
+      "step": 2968
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.662268822240546e-06,
+      "loss": 1.7461,
+      "step": 2970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.649785395168391e-06,
+      "loss": 1.8237,
+      "step": 2972
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.637305843498303e-06,
+      "loss": 1.7544,
+      "step": 2974
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6248301878085215e-06,
+      "loss": 1.7074,
+      "step": 2976
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.612358448670859e-06,
+      "loss": 1.724,
+      "step": 2978
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.599890646650676e-06,
+      "loss": 1.691,
+      "step": 2980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5874268023068335e-06,
+      "loss": 1.734,
+      "step": 2982
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.574966936191676e-06,
+      "loss": 1.7382,
+      "step": 2984
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562511068850978e-06,
+      "loss": 1.7528,
+      "step": 2986
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.550059220823927e-06,
+      "loss": 1.7086,
+      "step": 2988
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.537611412643085e-06,
+      "loss": 1.7194,
+      "step": 2990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.525167664834341e-06,
+      "loss": 1.7357,
+      "step": 2992
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.5127279979168995e-06,
+      "loss": 1.7294,
+      "step": 2994
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.5002924324032365e-06,
+      "loss": 1.7491,
+      "step": 2996
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.487860988799055e-06,
+      "loss": 1.7811,
+      "step": 2998
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.475433687603271e-06,
+      "loss": 1.764,
+      "step": 3000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.46301054930796e-06,
+      "loss": 1.7314,
+      "step": 3002
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.450591594398348e-06,
+      "loss": 1.7087,
+      "step": 3004
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.438176843352744e-06,
+      "loss": 1.7543,
+      "step": 3006
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.425766316642535e-06,
+      "loss": 1.7635,
+      "step": 3008
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.413360034732151e-06,
+      "loss": 1.7508,
+      "step": 3010
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.400958018079009e-06,
+      "loss": 1.7588,
+      "step": 3012
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.388560287133493e-06,
+      "loss": 1.72,
+      "step": 3014
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.37616686233893e-06,
+      "loss": 1.6976,
+      "step": 3016
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3637777641315365e-06,
+      "loss": 1.7438,
+      "step": 3018
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.351393012940402e-06,
+      "loss": 1.7463,
+      "step": 3020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.339012629187441e-06,
+      "loss": 1.7578,
+      "step": 3022
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.326636633287374e-06,
+      "loss": 1.7353,
+      "step": 3024
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.314265045647674e-06,
+      "loss": 1.7575,
+      "step": 3026
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3018978866685645e-06,
+      "loss": 1.7299,
+      "step": 3028
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.289535176742948e-06,
+      "loss": 1.6735,
+      "step": 3030
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.277176936256399e-06,
+      "loss": 1.7195,
+      "step": 3032
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.26482318558712e-06,
+      "loss": 1.7136,
+      "step": 3034
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.252473945105913e-06,
+      "loss": 1.7057,
+      "step": 3036
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.240129235176138e-06,
+      "loss": 1.7523,
+      "step": 3038
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2277890761536885e-06,
+      "loss": 1.7228,
+      "step": 3040
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.215453488386945e-06,
+      "loss": 1.7321,
+      "step": 3042
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.20312249221677e-06,
+      "loss": 1.7216,
+      "step": 3044
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.190796107976432e-06,
+      "loss": 1.7161,
+      "step": 3046
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1784743559916115e-06,
+      "loss": 1.744,
+      "step": 3048
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.166157256580339e-06,
+      "loss": 1.7208,
+      "step": 3050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.15384483005298e-06,
+      "loss": 1.7066,
+      "step": 3052
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1415370967121896e-06,
+      "loss": 1.7108,
+      "step": 3054
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.129234076852891e-06,
+      "loss": 1.7175,
+      "step": 3056
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.116935790762226e-06,
+      "loss": 1.6687,
+      "step": 3058
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.104642258719536e-06,
+      "loss": 1.669,
+      "step": 3060
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.09235350099633e-06,
+      "loss": 1.7317,
+      "step": 3062
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.080069537856231e-06,
+      "loss": 1.7176,
+      "step": 3064
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0677903895549625e-06,
+      "loss": 1.7129,
+      "step": 3066
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.055516076340311e-06,
+      "loss": 1.7681,
+      "step": 3068
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.043246618452082e-06,
+      "loss": 1.7726,
+      "step": 3070
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.030982036122086e-06,
+      "loss": 1.7202,
+      "step": 3072
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.018722349574083e-06,
+      "loss": 1.7677,
+      "step": 3074
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.006467579023768e-06,
+      "loss": 1.7238,
+      "step": 3076
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.99421774467872e-06,
+      "loss": 1.6808,
+      "step": 3078
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.981972866738397e-06,
+      "loss": 1.7812,
+      "step": 3080
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.969732965394064e-06,
+      "loss": 1.7176,
+      "step": 3082
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.957498060828791e-06,
+      "loss": 1.6742,
+      "step": 3084
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.945268173217404e-06,
+      "loss": 1.7832,
+      "step": 3086
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.93304332272646e-06,
+      "loss": 1.7012,
+      "step": 3088
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.920823529514205e-06,
+      "loss": 1.7512,
+      "step": 3090
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9086088137305506e-06,
+      "loss": 1.6759,
+      "step": 3092
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.896399195517029e-06,
+      "loss": 1.7283,
+      "step": 3094
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.884194695006779e-06,
+      "loss": 1.7325,
+      "step": 3096
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.871995332324489e-06,
+      "loss": 1.7148,
+      "step": 3098
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.859801127586383e-06,
+      "loss": 1.7332,
+      "step": 3100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.847612100900173e-06,
+      "loss": 1.6571,
+      "step": 3102
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.835428272365039e-06,
+      "loss": 1.7508,
+      "step": 3104
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.82324966207158e-06,
+      "loss": 1.7754,
+      "step": 3106
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.811076290101806e-06,
+      "loss": 1.73,
+      "step": 3108
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.798908176529071e-06,
+      "loss": 1.6937,
+      "step": 3110
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.786745341418071e-06,
+      "loss": 1.714,
+      "step": 3112
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.774587804824796e-06,
+      "loss": 1.6884,
+      "step": 3114
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.762435586796494e-06,
+      "loss": 1.722,
+      "step": 3116
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.750288707371649e-06,
+      "loss": 1.6668,
+      "step": 3118
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.738147186579939e-06,
+      "loss": 1.7087,
+      "step": 3120
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.726011044442202e-06,
+      "loss": 1.6427,
+      "step": 3122
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.713880300970415e-06,
+      "loss": 1.7338,
+      "step": 3124
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.701754976167644e-06,
+      "loss": 1.7626,
+      "step": 3126
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6896350900280236e-06,
+      "loss": 1.7272,
+      "step": 3128
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.67752066253673e-06,
+      "loss": 1.6758,
+      "step": 3130
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.66541171366992e-06,
+      "loss": 1.669,
+      "step": 3132
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.653308263394729e-06,
+      "loss": 1.6627,
+      "step": 3134
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.641210331669226e-06,
+      "loss": 1.7601,
+      "step": 3136
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6291179384423674e-06,
+      "loss": 1.6956,
+      "step": 3138
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.617031103653994e-06,
+      "loss": 1.7636,
+      "step": 3140
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.604949847234767e-06,
+      "loss": 1.7448,
+      "step": 3142
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.592874189106161e-06,
+      "loss": 1.749,
+      "step": 3144
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.580804149180404e-06,
+      "loss": 1.7156,
+      "step": 3146
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.568739747360477e-06,
+      "loss": 1.6888,
+      "step": 3148
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.556681003540054e-06,
+      "loss": 1.773,
+      "step": 3150
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.54462793760348e-06,
+      "loss": 1.6866,
+      "step": 3152
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5325805694257375e-06,
+      "loss": 1.756,
+      "step": 3154
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.520538918872419e-06,
+      "loss": 1.6848,
+      "step": 3156
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.508503005799679e-06,
+      "loss": 1.6354,
+      "step": 3158
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.496472850054222e-06,
+      "loss": 1.6916,
+      "step": 3160
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.484448471473248e-06,
+      "loss": 1.7072,
+      "step": 3162
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.472429889884439e-06,
+      "loss": 1.7441,
+      "step": 3164
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.460417125105919e-06,
+      "loss": 1.7271,
+      "step": 3166
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.448410196946211e-06,
+      "loss": 1.6829,
+      "step": 3168
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.436409125204225e-06,
+      "loss": 1.7156,
+      "step": 3170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4244139296692025e-06,
+      "loss": 1.7673,
+      "step": 3172
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4124246301207035e-06,
+      "loss": 1.7121,
+      "step": 3174
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.400441246328566e-06,
+      "loss": 1.7334,
+      "step": 3176
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3884637980528685e-06,
+      "loss": 1.7087,
+      "step": 3178
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3764923050439e-06,
+      "loss": 1.7468,
+      "step": 3180
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.364526787042145e-06,
+      "loss": 1.7383,
+      "step": 3182
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3525672637782135e-06,
+      "loss": 1.7301,
+      "step": 3184
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3406137549728474e-06,
+      "loss": 1.6833,
+      "step": 3186
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.328666280336861e-06,
+      "loss": 1.7198,
+      "step": 3188
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.316724859571127e-06,
+      "loss": 1.7081,
+      "step": 3190
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.304789512366526e-06,
+      "loss": 1.724,
+      "step": 3192
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29286025840393e-06,
+      "loss": 1.6357,
+      "step": 3194
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.280937117354164e-06,
+      "loss": 1.7317,
+      "step": 3196
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.269020108877967e-06,
+      "loss": 1.6897,
+      "step": 3198
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.257109252625973e-06,
+      "loss": 1.6535,
+      "step": 3200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24520456823867e-06,
+      "loss": 1.7209,
+      "step": 3202
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.233306075346362e-06,
+      "loss": 1.7269,
+      "step": 3204
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.221413793569153e-06,
+      "loss": 1.6897,
+      "step": 3206
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.209527742516894e-06,
+      "loss": 1.708,
+      "step": 3208
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.197647941789173e-06,
+      "loss": 1.7405,
+      "step": 3210
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.185774410975268e-06,
+      "loss": 1.7663,
+      "step": 3212
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.173907169654112e-06,
+      "loss": 1.7076,
+      "step": 3214
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.162046237394272e-06,
+      "loss": 1.7101,
+      "step": 3216
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.150191633753919e-06,
+      "loss": 1.7113,
+      "step": 3218
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.138343378280773e-06,
+      "loss": 1.6908,
+      "step": 3220
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1265014905121e-06,
+      "loss": 1.731,
+      "step": 3222
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.114665989974653e-06,
+      "loss": 1.728,
+      "step": 3224
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.102836896184666e-06,
+      "loss": 1.7388,
+      "step": 3226
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.091014228647798e-06,
+      "loss": 1.7684,
+      "step": 3228
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.079198006859115e-06,
+      "loss": 1.6657,
+      "step": 3230
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.067388250303057e-06,
+      "loss": 1.7267,
+      "step": 3232
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.055584978453404e-06,
+      "loss": 1.7587,
+      "step": 3234
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.043788210773233e-06,
+      "loss": 1.6884,
+      "step": 3236
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.031997966714911e-06,
+      "loss": 1.7389,
+      "step": 3238
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.020214265720031e-06,
+      "loss": 1.7156,
+      "step": 3240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.008437127219413e-06,
+      "loss": 1.7046,
+      "step": 3242
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.996666570633042e-06,
+      "loss": 1.7168,
+      "step": 3244
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.984902615370061e-06,
+      "loss": 1.6875,
+      "step": 3246
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.973145280828718e-06,
+      "loss": 1.6607,
+      "step": 3248
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.961394586396348e-06,
+      "loss": 1.7148,
+      "step": 3250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9496505514493415e-06,
+      "loss": 1.7092,
+      "step": 3252
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.937913195353104e-06,
+      "loss": 1.7488,
+      "step": 3254
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9261825374620215e-06,
+      "loss": 1.6945,
+      "step": 3256
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.914458597119448e-06,
+      "loss": 1.696,
+      "step": 3258
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.902741393657645e-06,
+      "loss": 1.7215,
+      "step": 3260
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.89103094639778e-06,
+      "loss": 1.7364,
+      "step": 3262
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.879327274649868e-06,
+      "loss": 1.7026,
+      "step": 3264
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.867630397712757e-06,
+      "loss": 1.7747,
+      "step": 3266
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.855940334874095e-06,
+      "loss": 1.7406,
+      "step": 3268
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.844257105410288e-06,
+      "loss": 1.6833,
+      "step": 3270
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.832580728586474e-06,
+      "loss": 1.6937,
+      "step": 3272
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.820911223656491e-06,
+      "loss": 1.7058,
+      "step": 3274
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.809248609862848e-06,
+      "loss": 1.7217,
+      "step": 3276
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.797592906436691e-06,
+      "loss": 1.6833,
+      "step": 3278
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.785944132597772e-06,
+      "loss": 1.7193,
+      "step": 3280
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.774302307554409e-06,
+      "loss": 1.6748,
+      "step": 3282
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7626674505034656e-06,
+      "loss": 1.7815,
+      "step": 3284
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.751039580630324e-06,
+      "loss": 1.6722,
+      "step": 3286
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.73941871710884e-06,
+      "loss": 1.7531,
+      "step": 3288
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.727804879101306e-06,
+      "loss": 1.7253,
+      "step": 3290
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.716198085758441e-06,
+      "loss": 1.6982,
+      "step": 3292
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.704598356219344e-06,
+      "loss": 1.6648,
+      "step": 3294
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6930057096114695e-06,
+      "loss": 1.736,
+      "step": 3296
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.681420165050584e-06,
+      "loss": 1.6754,
+      "step": 3298
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.669841741640752e-06,
+      "loss": 1.7281,
+      "step": 3300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.658270458474287e-06,
+      "loss": 1.6913,
+      "step": 3302
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.646706334631745e-06,
+      "loss": 1.716,
+      "step": 3304
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.635149389181855e-06,
+      "loss": 1.686,
+      "step": 3306
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.623599641181523e-06,
+      "loss": 1.7215,
+      "step": 3308
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.612057109675783e-06,
+      "loss": 1.7375,
+      "step": 3310
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.600521813697773e-06,
+      "loss": 1.755,
+      "step": 3312
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.588993772268689e-06,
+      "loss": 1.6676,
+      "step": 3314
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5774730043977754e-06,
+      "loss": 1.7118,
+      "step": 3316
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.565959529082278e-06,
+      "loss": 1.7041,
+      "step": 3318
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5544533653074205e-06,
+      "loss": 1.578,
+      "step": 3320
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.542954532046368e-06,
+      "loss": 1.6882,
+      "step": 3322
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.531463048260199e-06,
+      "loss": 1.6585,
+      "step": 3324
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.519978932897872e-06,
+      "loss": 1.6987,
+      "step": 3326
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.508502204896199e-06,
+      "loss": 1.7256,
+      "step": 3328
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4970328831798005e-06,
+      "loss": 1.7598,
+      "step": 3330
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.485570986661093e-06,
+      "loss": 1.6903,
+      "step": 3332
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.474116534240251e-06,
+      "loss": 1.678,
+      "step": 3334
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.462669544805166e-06,
+      "loss": 1.6743,
+      "step": 3336
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.45123003723143e-06,
+      "loss": 1.7381,
+      "step": 3338
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.439798030382296e-06,
+      "loss": 1.7339,
+      "step": 3340
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.428373543108647e-06,
+      "loss": 1.7239,
+      "step": 3342
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.416956594248971e-06,
+      "loss": 1.7397,
+      "step": 3344
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.405547202629315e-06,
+      "loss": 1.7075,
+      "step": 3346
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.394145387063274e-06,
+      "loss": 1.7025,
+      "step": 3348
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.382751166351952e-06,
+      "loss": 1.7341,
+      "step": 3350
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.37136455928392e-06,
+      "loss": 1.6631,
+      "step": 3352
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.359985584635203e-06,
+      "loss": 1.6961,
+      "step": 3354
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.348614261169236e-06,
+      "loss": 1.7443,
+      "step": 3356
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.33725060763684e-06,
+      "loss": 1.6763,
+      "step": 3358
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3258946427761905e-06,
+      "loss": 1.7395,
+      "step": 3360
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.314546385312773e-06,
+      "loss": 1.6788,
+      "step": 3362
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.303205853959377e-06,
+      "loss": 1.6859,
+      "step": 3364
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.291873067416048e-06,
+      "loss": 1.7126,
+      "step": 3366
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.280548044370063e-06,
+      "loss": 1.664,
+      "step": 3368
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.269230803495883e-06,
+      "loss": 1.7647,
+      "step": 3370
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.257921363455159e-06,
+      "loss": 1.7033,
+      "step": 3372
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.246619742896669e-06,
+      "loss": 1.7287,
+      "step": 3374
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.235325960456288e-06,
+      "loss": 1.7137,
+      "step": 3376
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.224040034756975e-06,
+      "loss": 1.6515,
+      "step": 3378
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.212761984408736e-06,
+      "loss": 1.6676,
+      "step": 3380
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.201491828008586e-06,
+      "loss": 1.7372,
+      "step": 3382
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.190229584140527e-06,
+      "loss": 1.7793,
+      "step": 3384
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.178975271375506e-06,
+      "loss": 1.715,
+      "step": 3386
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.167728908271397e-06,
+      "loss": 1.71,
+      "step": 3388
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.156490513372976e-06,
+      "loss": 1.7222,
+      "step": 3390
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.145260105211857e-06,
+      "loss": 1.6985,
+      "step": 3392
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134037702306502e-06,
+      "loss": 1.7443,
+      "step": 3394
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122823323162166e-06,
+      "loss": 1.7298,
+      "step": 3396
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.111616986270875e-06,
+      "loss": 1.7298,
+      "step": 3398
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.100418710111398e-06,
+      "loss": 1.6558,
+      "step": 3400
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.089228513149198e-06,
+      "loss": 1.7461,
+      "step": 3402
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.078046413836424e-06,
+      "loss": 1.6995,
+      "step": 3404
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.066872430611888e-06,
+      "loss": 1.7031,
+      "step": 3406
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0557065819009896e-06,
+      "loss": 1.744,
+      "step": 3408
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0445488861157354e-06,
+      "loss": 1.7055,
+      "step": 3410
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.033399361654682e-06,
+      "loss": 1.7081,
+      "step": 3412
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.022258026902911e-06,
+      "loss": 1.749,
+      "step": 3414
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.011124900232006e-06,
+      "loss": 1.7324,
+      "step": 3416
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.6734,
+      "step": 3418
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.988883344551383e-06,
+      "loss": 1.7253,
+      "step": 3420
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.977774952217029e-06,
+      "loss": 1.6955,
+      "step": 3422
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.966674841314201e-06,
+      "loss": 1.6978,
+      "step": 3424
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.955583030146499e-06,
+      "loss": 1.6691,
+      "step": 3426
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.944499537003841e-06,
+      "loss": 1.6756,
+      "step": 3428
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.933424380162428e-06,
+      "loss": 1.7525,
+      "step": 3430
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.922357577884709e-06,
+      "loss": 1.7249,
+      "step": 3432
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.911299148419363e-06,
+      "loss": 1.7172,
+      "step": 3434
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.900249110001263e-06,
+      "loss": 1.7199,
+      "step": 3436
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.889207480851442e-06,
+      "loss": 1.6727,
+      "step": 3438
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.87817427917707e-06,
+      "loss": 1.7453,
+      "step": 3440
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.867149523171416e-06,
+      "loss": 1.7051,
+      "step": 3442
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.856133231013828e-06,
+      "loss": 1.7334,
+      "step": 3444
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.845125420869696e-06,
+      "loss": 1.6893,
+      "step": 3446
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8341261108904155e-06,
+      "loss": 1.7094,
+      "step": 3448
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.823135319213376e-06,
+      "loss": 1.7141,
+      "step": 3450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.812153063961915e-06,
+      "loss": 1.6983,
+      "step": 3452
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.801179363245299e-06,
+      "loss": 1.7055,
+      "step": 3454
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.790214235158682e-06,
+      "loss": 1.6991,
+      "step": 3456
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.779257697783087e-06,
+      "loss": 1.67,
+      "step": 3458
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.768309769185369e-06,
+      "loss": 1.6783,
+      "step": 3460
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.757370467418191e-06,
+      "loss": 1.6837,
+      "step": 3462
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74643981051998e-06,
+      "loss": 1.6924,
+      "step": 3464
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735517816514921e-06,
+      "loss": 1.726,
+      "step": 3466
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.724604503412909e-06,
+      "loss": 1.6536,
+      "step": 3468
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.713699889209521e-06,
+      "loss": 1.7332,
+      "step": 3470
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.702803991886001e-06,
+      "loss": 1.7704,
+      "step": 3472
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6919168294092e-06,
+      "loss": 1.7276,
+      "step": 3474
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.681038419731587e-06,
+      "loss": 1.6783,
+      "step": 3476
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6701687807911885e-06,
+      "loss": 1.6687,
+      "step": 3478
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6593079305115615e-06,
+      "loss": 1.697,
+      "step": 3480
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.64845588680178e-06,
+      "loss": 1.7164,
+      "step": 3482
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.637612667556397e-06,
+      "loss": 1.7573,
+      "step": 3484
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.626778290655411e-06,
+      "loss": 1.7073,
+      "step": 3486
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.615952773964235e-06,
+      "loss": 1.7193,
+      "step": 3488
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.605136135333676e-06,
+      "loss": 1.7629,
+      "step": 3490
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.594328392599912e-06,
+      "loss": 1.6917,
+      "step": 3492
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.58352956358444e-06,
+      "loss": 1.6744,
+      "step": 3494
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5727396660940545e-06,
+      "loss": 1.6927,
+      "step": 3496
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.561958717920834e-06,
+      "loss": 1.6888,
+      "step": 3498
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.551186736842094e-06,
+      "loss": 1.7117,
+      "step": 3500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.540423740620368e-06,
+      "loss": 1.7026,
+      "step": 3502
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.529669747003362e-06,
+      "loss": 1.6325,
+      "step": 3504
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.518924773723953e-06,
+      "loss": 1.6608,
+      "step": 3506
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.508188838500128e-06,
+      "loss": 1.7004,
+      "step": 3508
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.497461959034995e-06,
+      "loss": 1.6976,
+      "step": 3510
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4867441530166986e-06,
+      "loss": 1.6689,
+      "step": 3512
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.476035438118442e-06,
+      "loss": 1.7208,
+      "step": 3514
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.465335831998432e-06,
+      "loss": 1.7226,
+      "step": 3516
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.454645352299859e-06,
+      "loss": 1.6693,
+      "step": 3518
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.443964016650853e-06,
+      "loss": 1.7041,
+      "step": 3520
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.433291842664479e-06,
+      "loss": 1.7228,
+      "step": 3522
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.422628847938687e-06,
+      "loss": 1.6999,
+      "step": 3524
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4119750500562965e-06,
+      "loss": 1.7468,
+      "step": 3526
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.401330466584956e-06,
+      "loss": 1.715,
+      "step": 3528
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3906951150771235e-06,
+      "loss": 1.6822,
+      "step": 3530
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.380069013070033e-06,
+      "loss": 1.6671,
+      "step": 3532
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.36945217808567e-06,
+      "loss": 1.7151,
+      "step": 3534
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3588446276307286e-06,
+      "loss": 1.7124,
+      "step": 3536
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.348246379196602e-06,
+      "loss": 1.6857,
+      "step": 3538
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.337657450259345e-06,
+      "loss": 1.72,
+      "step": 3540
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3270778582796416e-06,
+      "loss": 1.7263,
+      "step": 3542
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.316507620702779e-06,
+      "loss": 1.6945,
+      "step": 3544
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305946754958623e-06,
+      "loss": 1.6743,
+      "step": 3546
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2953952784615826e-06,
+      "loss": 1.6788,
+      "step": 3548
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.284853208610591e-06,
+      "loss": 1.6946,
+      "step": 3550
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.274320562789055e-06,
+      "loss": 1.6837,
+      "step": 3552
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.263797358364857e-06,
+      "loss": 1.6873,
+      "step": 3554
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253283612690302e-06,
+      "loss": 1.6974,
+      "step": 3556
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.242779343102108e-06,
+      "loss": 1.7481,
+      "step": 3558
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2322845669213485e-06,
+      "loss": 1.7223,
+      "step": 3560
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.221799301453467e-06,
+      "loss": 1.7276,
+      "step": 3562
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.211323563988204e-06,
+      "loss": 1.7289,
+      "step": 3564
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2008573717996045e-06,
+      "loss": 1.7162,
+      "step": 3566
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.190400742145959e-06,
+      "loss": 1.7411,
+      "step": 3568
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.179953692269798e-06,
+      "loss": 1.6959,
+      "step": 3570
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.169516239397855e-06,
+      "loss": 1.7402,
+      "step": 3572
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.15908840074104e-06,
+      "loss": 1.7238,
+      "step": 3574
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1486701934944e-06,
+      "loss": 1.6771,
+      "step": 3576
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.138261634837114e-06,
+      "loss": 1.6555,
+      "step": 3578
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.127862741932444e-06,
+      "loss": 1.6674,
+      "step": 3580
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.117473531927715e-06,
+      "loss": 1.6888,
+      "step": 3582
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.107094021954279e-06,
+      "loss": 1.7023,
+      "step": 3584
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0967242291275e-06,
+      "loss": 1.714,
+      "step": 3586
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.086364170546718e-06,
+      "loss": 1.7046,
+      "step": 3588
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.076013863295225e-06,
+      "loss": 1.7409,
+      "step": 3590
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.065673324440221e-06,
+      "loss": 1.6997,
+      "step": 3592
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.055342571032808e-06,
+      "loss": 1.7561,
+      "step": 3594
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.045021620107956e-06,
+      "loss": 1.6366,
+      "step": 3596
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.034710488684468e-06,
+      "loss": 1.7556,
+      "step": 3598
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.024409193764944e-06,
+      "loss": 1.7527,
+      "step": 3600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.014117752335778e-06,
+      "loss": 1.6664,
+      "step": 3602
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.00383618136711e-06,
+      "loss": 1.7273,
+      "step": 3604
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.99356449781281e-06,
+      "loss": 1.6775,
+      "step": 3606
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98330271861043e-06,
+      "loss": 1.7207,
+      "step": 3608
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.973050860681204e-06,
+      "loss": 1.6676,
+      "step": 3610
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.962808940930003e-06,
+      "loss": 1.7502,
+      "step": 3612
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.952576976245307e-06,
+      "loss": 1.7088,
+      "step": 3614
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.942354983499182e-06,
+      "loss": 1.6985,
+      "step": 3616
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.932142979547255e-06,
+      "loss": 1.6376,
+      "step": 3618
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.921940981228673e-06,
+      "loss": 1.7089,
+      "step": 3620
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.911749005366095e-06,
+      "loss": 1.6919,
+      "step": 3622
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.901567068765643e-06,
+      "loss": 1.6936,
+      "step": 3624
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.891395188216889e-06,
+      "loss": 1.743,
+      "step": 3626
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8812333804928255e-06,
+      "loss": 1.7154,
+      "step": 3628
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.871081662349833e-06,
+      "loss": 1.7057,
+      "step": 3630
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.860940050527654e-06,
+      "loss": 1.6757,
+      "step": 3632
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850808561749367e-06,
+      "loss": 1.6905,
+      "step": 3634
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8406872127213605e-06,
+      "loss": 1.7503,
+      "step": 3636
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.830576020133301e-06,
+      "loss": 1.7559,
+      "step": 3638
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8204750006581014e-06,
+      "loss": 1.6708,
+      "step": 3640
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8103841709519087e-06,
+      "loss": 1.6499,
+      "step": 3642
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.800303547654064e-06,
+      "loss": 1.6913,
+      "step": 3644
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7902331473870778e-06,
+      "loss": 1.6909,
+      "step": 3646
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.780172986756604e-06,
+      "loss": 1.715,
+      "step": 3648
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.770123082351412e-06,
+      "loss": 1.6852,
+      "step": 3650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7600834507433626e-06,
+      "loss": 1.7525,
+      "step": 3652
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7500541084873665e-06,
+      "loss": 1.673,
+      "step": 3654
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7400350721213764e-06,
+      "loss": 1.6967,
+      "step": 3656
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.730026358166351e-06,
+      "loss": 1.6646,
+      "step": 3658
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7200279831262244e-06,
+      "loss": 1.6362,
+      "step": 3660
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.710039963487889e-06,
+      "loss": 1.7247,
+      "step": 3662
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7000623157211458e-06,
+      "loss": 1.7023,
+      "step": 3664
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.690095056278713e-06,
+      "loss": 1.7301,
+      "step": 3666
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6801382015961707e-06,
+      "loss": 1.6812,
+      "step": 3668
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.670191768091933e-06,
+      "loss": 1.6189,
+      "step": 3670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6602557721672416e-06,
+      "loss": 1.6973,
+      "step": 3672
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.650330230206124e-06,
+      "loss": 1.672,
+      "step": 3674
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.640415158575368e-06,
+      "loss": 1.6815,
+      "step": 3676
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6305105736245017e-06,
+      "loss": 1.6858,
+      "step": 3678
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6206164916857455e-06,
+      "loss": 1.7371,
+      "step": 3680
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.610732929074022e-06,
+      "loss": 1.7136,
+      "step": 3682
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6008599020868985e-06,
+      "loss": 1.689,
+      "step": 3684
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.59099742700456e-06,
+      "loss": 1.7184,
+      "step": 3686
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.581145520089807e-06,
+      "loss": 1.7299,
+      "step": 3688
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.571304197588006e-06,
+      "loss": 1.712,
+      "step": 3690
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5614734757270695e-06,
+      "loss": 1.6681,
+      "step": 3692
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5516533707174385e-06,
+      "loss": 1.7441,
+      "step": 3694
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.541843898752032e-06,
+      "loss": 1.6771,
+      "step": 3696
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.532045076006244e-06,
+      "loss": 1.6812,
+      "step": 3698
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5222569186379208e-06,
+      "loss": 1.6989,
+      "step": 3700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5124794427872976e-06,
+      "loss": 1.733,
+      "step": 3702
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5027126645770125e-06,
+      "loss": 1.6811,
+      "step": 3704
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4929566001120607e-06,
+      "loss": 1.6739,
+      "step": 3706
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4832112654797722e-06,
+      "loss": 1.7706,
+      "step": 3708
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4734766767497753e-06,
+      "loss": 1.6636,
+      "step": 3710
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.463752849973987e-06,
+      "loss": 1.7613,
+      "step": 3712
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.454039801186575e-06,
+      "loss": 1.7144,
+      "step": 3714
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.444337546403945e-06,
+      "loss": 1.6643,
+      "step": 3716
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.434646101624686e-06,
+      "loss": 1.7702,
+      "step": 3718
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.424965482829572e-06,
+      "loss": 1.6968,
+      "step": 3720
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.415295705981525e-06,
+      "loss": 1.7,
+      "step": 3722
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.405636787025591e-06,
+      "loss": 1.6739,
+      "step": 3724
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.395988741888904e-06,
+      "loss": 1.6437,
+      "step": 3726
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.386351586480673e-06,
+      "loss": 1.6887,
+      "step": 3728
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3767253366921515e-06,
+      "loss": 1.7369,
+      "step": 3730
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.367110008396608e-06,
+      "loss": 1.7072,
+      "step": 3732
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3575056174493027e-06,
+      "loss": 1.7198,
+      "step": 3734
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.347912179687459e-06,
+      "loss": 1.6896,
+      "step": 3736
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3383297109302415e-06,
+      "loss": 1.6986,
+      "step": 3738
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.328758226978729e-06,
+      "loss": 1.7217,
+      "step": 3740
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3191977436158763e-06,
+      "loss": 1.6343,
+      "step": 3742
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3096482766065106e-06,
+      "loss": 1.6583,
+      "step": 3744
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.300109841697289e-06,
+      "loss": 1.6993,
+      "step": 3746
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2905824546166764e-06,
+      "loss": 1.6916,
+      "step": 3748
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2810661310749237e-06,
+      "loss": 1.7258,
+      "step": 3750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2715608867640347e-06,
+      "loss": 1.6964,
+      "step": 3752
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.262066737357744e-06,
+      "loss": 1.7522,
+      "step": 3754
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.252583698511499e-06,
+      "loss": 1.6674,
+      "step": 3756
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.243111785862412e-06,
+      "loss": 1.711,
+      "step": 3758
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.233651015029259e-06,
+      "loss": 1.7383,
+      "step": 3760
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2242014016124424e-06,
+      "loss": 1.7193,
+      "step": 3762
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2147629611939666e-06,
+      "loss": 1.7363,
+      "step": 3764
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.205335709337405e-06,
+      "loss": 1.676,
+      "step": 3766
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.195919661587894e-06,
+      "loss": 1.6596,
+      "step": 3768
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.186514833472085e-06,
+      "loss": 1.6935,
+      "step": 3770
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1771212404981375e-06,
+      "loss": 1.7308,
+      "step": 3772
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1677388981556724e-06,
+      "loss": 1.6727,
+      "step": 3774
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1583678219157686e-06,
+      "loss": 1.672,
+      "step": 3776
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1490080272309243e-06,
+      "loss": 1.6796,
+      "step": 3778
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1396595295350383e-06,
+      "loss": 1.7015,
+      "step": 3780
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130322344243373e-06,
+      "loss": 1.7093,
+      "step": 3782
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.120996486752542e-06,
+      "loss": 1.6717,
+      "step": 3784
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1116819724404877e-06,
+      "loss": 1.7212,
+      "step": 3786
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1023788166664383e-06,
+      "loss": 1.667,
+      "step": 3788
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.093087034770891e-06,
+      "loss": 1.7073,
+      "step": 3790
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.083806642075592e-06,
+      "loss": 1.7223,
+      "step": 3792
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0745376538835092e-06,
+      "loss": 1.6782,
+      "step": 3794
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0652800854788046e-06,
+      "loss": 1.692,
+      "step": 3796
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0560339521268e-06,
+      "loss": 1.6957,
+      "step": 3798
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0467992690739714e-06,
+      "loss": 1.6837,
+      "step": 3800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.037576051547916e-06,
+      "loss": 1.6785,
+      "step": 3802
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0283643147573195e-06,
+      "loss": 1.7033,
+      "step": 3804
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.01916407389193e-06,
+      "loss": 1.6981,
+      "step": 3806
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.009975344122552e-06,
+      "loss": 1.6983,
+      "step": 3808
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.000798140600999e-06,
+      "loss": 1.7058,
+      "step": 3810
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.991632478460088e-06,
+      "loss": 1.6837,
+      "step": 3812
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.982478372813592e-06,
+      "loss": 1.6645,
+      "step": 3814
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.973335838756237e-06,
+      "loss": 1.7346,
+      "step": 3816
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.964204891363662e-06,
+      "loss": 1.763,
+      "step": 3818
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9550855456924154e-06,
+      "loss": 1.7203,
+      "step": 3820
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9459778167798914e-06,
+      "loss": 1.7042,
+      "step": 3822
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9368817196443444e-06,
+      "loss": 1.7017,
+      "step": 3824
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9277972692848444e-06,
+      "loss": 1.736,
+      "step": 3826
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.918724480681259e-06,
+      "loss": 1.7306,
+      "step": 3828
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9096633687942187e-06,
+      "loss": 1.7035,
+      "step": 3830
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9006139485651065e-06,
+      "loss": 1.6413,
+      "step": 3832
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.891576234916025e-06,
+      "loss": 1.6867,
+      "step": 3834
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.882550242749771e-06,
+      "loss": 1.6727,
+      "step": 3836
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.873535986949817e-06,
+      "loss": 1.667,
+      "step": 3838
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8645334823802784e-06,
+      "loss": 1.6604,
+      "step": 3840
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.855542743885895e-06,
+      "loss": 1.6662,
+      "step": 3842
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8465637862920092e-06,
+      "loss": 1.6988,
+      "step": 3844
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.837596624404527e-06,
+      "loss": 1.6451,
+      "step": 3846
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8286412730099133e-06,
+      "loss": 1.7142,
+      "step": 3848
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.819697746875153e-06,
+      "loss": 1.7108,
+      "step": 3850
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.810766060747734e-06,
+      "loss": 1.6933,
+      "step": 3852
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.801846229355619e-06,
+      "loss": 1.6906,
+      "step": 3854
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7929382674072246e-06,
+      "loss": 1.7149,
+      "step": 3856
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.784042189591395e-06,
+      "loss": 1.6991,
+      "step": 3858
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.775158010577379e-06,
+      "loss": 1.721,
+      "step": 3860
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7662857450147963e-06,
+      "loss": 1.7201,
+      "step": 3862
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7574254075336316e-06,
+      "loss": 1.6912,
+      "step": 3864
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.748577012744198e-06,
+      "loss": 1.6594,
+      "step": 3866
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7397405752371185e-06,
+      "loss": 1.692,
+      "step": 3868
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.730916109583287e-06,
+      "loss": 1.6464,
+      "step": 3870
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.722103630333872e-06,
+      "loss": 1.6939,
+      "step": 3872
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.713303152020269e-06,
+      "loss": 1.6617,
+      "step": 3874
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7045146891540883e-06,
+      "loss": 1.6905,
+      "step": 3876
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6957382562271185e-06,
+      "loss": 1.6843,
+      "step": 3878
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.686973867711319e-06,
+      "loss": 1.6941,
+      "step": 3880
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.678221538058787e-06,
+      "loss": 1.6748,
+      "step": 3882
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 1.7123,
+      "step": 3884
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6607531130524733e-06,
+      "loss": 1.6903,
+      "step": 3886
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.652037046503361e-06,
+      "loss": 1.6681,
+      "step": 3888
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6433330964268266e-06,
+      "loss": 1.7041,
+      "step": 3890
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6346412771752984e-06,
+      "loss": 1.7106,
+      "step": 3892
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6259616030812128e-06,
+      "loss": 1.6703,
+      "step": 3894
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6172940884569765e-06,
+      "loss": 1.7233,
+      "step": 3896
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.608638747594946e-06,
+      "loss": 1.6705,
+      "step": 3898
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.599995594767406e-06,
+      "loss": 1.7194,
+      "step": 3900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.591364644226534e-06,
+      "loss": 1.6988,
+      "step": 3902
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.582745910204395e-06,
+      "loss": 1.6966,
+      "step": 3904
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5741394069129165e-06,
+      "loss": 1.7032,
+      "step": 3906
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.565545148543841e-06,
+      "loss": 1.7128,
+      "step": 3908
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5569631492687295e-06,
+      "loss": 1.6933,
+      "step": 3910
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5483934232389298e-06,
+      "loss": 1.7039,
+      "step": 3912
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5398359845855457e-06,
+      "loss": 1.6688,
+      "step": 3914
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.531290847419429e-06,
+      "loss": 1.689,
+      "step": 3916
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5227580258311323e-06,
+      "loss": 1.6852,
+      "step": 3918
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.514237533890913e-06,
+      "loss": 1.6775,
+      "step": 3920
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.505729385648692e-06,
+      "loss": 1.7075,
+      "step": 3922
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4972335951340387e-06,
+      "loss": 1.7188,
+      "step": 3924
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4887501763561417e-06,
+      "loss": 1.6703,
+      "step": 3926
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.480279143303793e-06,
+      "loss": 1.7228,
+      "step": 3928
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4718205099453586e-06,
+      "loss": 1.6877,
+      "step": 3930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4633742902287596e-06,
+      "loss": 1.6504,
+      "step": 3932
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.454940498081443e-06,
+      "loss": 1.658,
+      "step": 3934
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.446519147410368e-06,
+      "loss": 1.7203,
+      "step": 3936
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.438110252101977e-06,
+      "loss": 1.6388,
+      "step": 3938
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4297138260221764e-06,
+      "loss": 1.6593,
+      "step": 3940
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4213298830163055e-06,
+      "loss": 1.7502,
+      "step": 3942
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4129584369091264e-06,
+      "loss": 1.7081,
+      "step": 3944
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4045995015047906e-06,
+      "loss": 1.7114,
+      "step": 3946
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3962530905868174e-06,
+      "loss": 1.6988,
+      "step": 3948
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.387919217918079e-06,
+      "loss": 1.743,
+      "step": 3950
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3795978972407708e-06,
+      "loss": 1.7257,
+      "step": 3952
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3712891422763884e-06,
+      "loss": 1.6917,
+      "step": 3954
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.36299296672571e-06,
+      "loss": 1.6811,
+      "step": 3956
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3547093842687685e-06,
+      "loss": 1.7271,
+      "step": 3958
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.346438408564835e-06,
+      "loss": 1.6958,
+      "step": 3960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3381800532523903e-06,
+      "loss": 1.7317,
+      "step": 3962
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.329934331949101e-06,
+      "loss": 1.7416,
+      "step": 3964
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.321701258251805e-06,
+      "loss": 1.7144,
+      "step": 3966
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3134808457364865e-06,
+      "loss": 1.7364,
+      "step": 3968
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.305273107958248e-06,
+      "loss": 1.6821,
+      "step": 3970
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2970780584512964e-06,
+      "loss": 1.7358,
+      "step": 3972
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2888957107289046e-06,
+      "loss": 1.7143,
+      "step": 3974
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2807260782834185e-06,
+      "loss": 1.7406,
+      "step": 3976
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.272569174586209e-06,
+      "loss": 1.7027,
+      "step": 3978
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.26442501308765e-06,
+      "loss": 1.6409,
+      "step": 3980
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2562936072171137e-06,
+      "loss": 1.706,
+      "step": 3982
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.248174970382937e-06,
+      "loss": 1.7064,
+      "step": 3984
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.240069115972403e-06,
+      "loss": 1.7357,
+      "step": 3986
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.231976057351708e-06,
+      "loss": 1.6777,
+      "step": 3988
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.223895807865956e-06,
+      "loss": 1.7153,
+      "step": 3990
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2158283808391323e-06,
+      "loss": 1.6937,
+      "step": 3992
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2077737895740757e-06,
+      "loss": 1.7011,
+      "step": 3994
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.199732047352454e-06,
+      "loss": 1.6676,
+      "step": 3996
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1917031674347535e-06,
+      "loss": 1.6669,
+      "step": 3998
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1836871630602474e-06,
+      "loss": 1.6645,
+      "step": 4000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1756840474469852e-06,
+      "loss": 1.6953,
+      "step": 4002
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.167693833791751e-06,
+      "loss": 1.6863,
+      "step": 4004
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.159716535270063e-06,
+      "loss": 1.6519,
+      "step": 4006
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1517521650361407e-06,
+      "loss": 1.7179,
+      "step": 4008
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1438007362228906e-06,
+      "loss": 1.686,
+      "step": 4010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.135862261941869e-06,
+      "loss": 1.6834,
+      "step": 4012
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.127936755283276e-06,
+      "loss": 1.7057,
+      "step": 4014
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.120024229315931e-06,
+      "loss": 1.6986,
+      "step": 4016
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.112124697087248e-06,
+      "loss": 1.6942,
+      "step": 4018
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1042381716232085e-06,
+      "loss": 1.6366,
+      "step": 4020
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0963646659283545e-06,
+      "loss": 1.6976,
+      "step": 4022
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0885041929857506e-06,
+      "loss": 1.6788,
+      "step": 4024
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0806567657569874e-06,
+      "loss": 1.7089,
+      "step": 4026
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0728223971821236e-06,
+      "loss": 1.6973,
+      "step": 4028
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0650011001796954e-06,
+      "loss": 1.6983,
+      "step": 4030
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.057192887646683e-06,
+      "loss": 1.7225,
+      "step": 4032
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0493977724584924e-06,
+      "loss": 1.7349,
+      "step": 4034
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0416157674689274e-06,
+      "loss": 1.6811,
+      "step": 4036
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0338468855101777e-06,
+      "loss": 1.6924,
+      "step": 4038
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.026091139392795e-06,
+      "loss": 1.6511,
+      "step": 4040
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.018348541905668e-06,
+      "loss": 1.6885,
+      "step": 4042
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0106191058160028e-06,
+      "loss": 1.6799,
+      "step": 4044
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0029028438693075e-06,
+      "loss": 1.6833,
+      "step": 4046
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9951997687893623e-06,
+      "loss": 1.7172,
+      "step": 4048
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9875098932782065e-06,
+      "loss": 1.6572,
+      "step": 4050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9798332300161073e-06,
+      "loss": 1.7592,
+      "step": 4052
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9721697916615524e-06,
+      "loss": 1.6929,
+      "step": 4054
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9645195908512183e-06,
+      "loss": 1.676,
+      "step": 4056
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9568826401999586e-06,
+      "loss": 1.6516,
+      "step": 4058
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9492589523007665e-06,
+      "loss": 1.711,
+      "step": 4060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9416485397247796e-06,
+      "loss": 1.6535,
+      "step": 4062
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9340514150212375e-06,
+      "loss": 1.6974,
+      "step": 4064
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9264675907174713e-06,
+      "loss": 1.6661,
+      "step": 4066
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9188970793188732e-06,
+      "loss": 1.6867,
+      "step": 4068
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.911339893308892e-06,
+      "loss": 1.7158,
+      "step": 4070
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9037960451490001e-06,
+      "loss": 1.7097,
+      "step": 4072
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8962655472786772e-06,
+      "loss": 1.7105,
+      "step": 4074
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8887484121153831e-06,
+      "loss": 1.6908,
+      "step": 4076
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8812446520545534e-06,
+      "loss": 1.6994,
+      "step": 4078
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8737542794695608e-06,
+      "loss": 1.6628,
+      "step": 4080
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8662773067117102e-06,
+      "loss": 1.6573,
+      "step": 4082
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8588137461101973e-06,
+      "loss": 1.6519,
+      "step": 4084
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8513636099721156e-06,
+      "loss": 1.7173,
+      "step": 4086
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8439269105824142e-06,
+      "loss": 1.7113,
+      "step": 4088
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8365036602038932e-06,
+      "loss": 1.691,
+      "step": 4090
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8290938710771644e-06,
+      "loss": 1.6722,
+      "step": 4092
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8216975554206495e-06,
+      "loss": 1.7265,
+      "step": 4094
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8143147254305582e-06,
+      "loss": 1.6499,
+      "step": 4096
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8069453932808556e-06,
+      "loss": 1.6807,
+      "step": 4098
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.799589571123248e-06,
+      "loss": 1.657,
+      "step": 4100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7922472710871686e-06,
+      "loss": 1.629,
+      "step": 4102
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7849185052797525e-06,
+      "loss": 1.6466,
+      "step": 4104
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7776032857858205e-06,
+      "loss": 1.6842,
+      "step": 4106
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.770301624667846e-06,
+      "loss": 1.7095,
+      "step": 4108
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7630135339659549e-06,
+      "loss": 1.6502,
+      "step": 4110
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7557390256978946e-06,
+      "loss": 1.7131,
+      "step": 4112
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.748478111859011e-06,
+      "loss": 1.6945,
+      "step": 4114
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7412308044222392e-06,
+      "loss": 1.6525,
+      "step": 4116
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7339971153380742e-06,
+      "loss": 1.6918,
+      "step": 4118
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7267770565345554e-06,
+      "loss": 1.6592,
+      "step": 4120
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7195706399172496e-06,
+      "loss": 1.6943,
+      "step": 4122
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7123778773692213e-06,
+      "loss": 1.6851,
+      "step": 4124
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7051987807510262e-06,
+      "loss": 1.6928,
+      "step": 4126
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.698033361900685e-06,
+      "loss": 1.6884,
+      "step": 4128
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6908816326336607e-06,
+      "loss": 1.6203,
+      "step": 4130
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.683743604742847e-06,
+      "loss": 1.767,
+      "step": 4132
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.67661928999854e-06,
+      "loss": 1.6674,
+      "step": 4134
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.66950870014843e-06,
+      "loss": 1.7176,
+      "step": 4136
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.662411846917571e-06,
+      "loss": 1.6641,
+      "step": 4138
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6553287420083618e-06,
+      "loss": 1.7207,
+      "step": 4140
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.64825939710054e-06,
+      "loss": 1.7016,
+      "step": 4142
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6412038238511475e-06,
+      "loss": 1.6533,
+      "step": 4144
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6341620338945187e-06,
+      "loss": 1.7716,
+      "step": 4146
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6271340388422606e-06,
+      "loss": 1.6804,
+      "step": 4148
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6201198502832327e-06,
+      "loss": 1.6459,
+      "step": 4150
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.613119479783528e-06,
+      "loss": 1.6907,
+      "step": 4152
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.6061329388864577e-06,
+      "loss": 1.7065,
+      "step": 4154
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5991602391125173e-06,
+      "loss": 1.7075,
+      "step": 4156
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5922013919593925e-06,
+      "loss": 1.6951,
+      "step": 4158
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5852564089019185e-06,
+      "loss": 1.6751,
+      "step": 4160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5783253013920718e-06,
+      "loss": 1.6838,
+      "step": 4162
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5714080808589494e-06,
+      "loss": 1.6825,
+      "step": 4164
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5645047587087469e-06,
+      "loss": 1.6849,
+      "step": 4166
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5576153463247468e-06,
+      "loss": 1.6737,
+      "step": 4168
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.550739855067287e-06,
+      "loss": 1.6975,
+      "step": 4170
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5438782962737564e-06,
+      "loss": 1.6889,
+      "step": 4172
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.537030681258569e-06,
+      "loss": 1.6583,
+      "step": 4174
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5301970213131446e-06,
+      "loss": 1.641,
+      "step": 4176
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.523377327705895e-06,
+      "loss": 1.662,
+      "step": 4178
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5165716116821928e-06,
+      "loss": 1.6627,
+      "step": 4180
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.509779884464374e-06,
+      "loss": 1.6778,
+      "step": 4182
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.5030021572517062e-06,
+      "loss": 1.709,
+      "step": 4184
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.49623844122036e-06,
+      "loss": 1.7209,
+      "step": 4186
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4894887475234132e-06,
+      "loss": 1.6737,
+      "step": 4188
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4827530872908192e-06,
+      "loss": 1.6546,
+      "step": 4190
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.47603147162939e-06,
+      "loss": 1.6376,
+      "step": 4192
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4693239116227798e-06,
+      "loss": 1.7199,
+      "step": 4194
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4626304183314599e-06,
+      "loss": 1.68,
+      "step": 4196
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4559510027927115e-06,
+      "loss": 1.7021,
+      "step": 4198
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4492856760206087e-06,
+      "loss": 1.6845,
+      "step": 4200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4426344490059785e-06,
+      "loss": 1.69,
+      "step": 4202
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.435997332716409e-06,
+      "loss": 1.6867,
+      "step": 4204
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4293743380962178e-06,
+      "loss": 1.6813,
+      "step": 4206
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4227654760664344e-06,
+      "loss": 1.7228,
+      "step": 4208
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4161707575247907e-06,
+      "loss": 1.7068,
+      "step": 4210
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4095901933456857e-06,
+      "loss": 1.6901,
+      "step": 4212
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.4030237943801862e-06,
+      "loss": 1.641,
+      "step": 4214
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.396471571456004e-06,
+      "loss": 1.6804,
+      "step": 4216
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3899335353774656e-06,
+      "loss": 1.6894,
+      "step": 4218
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3834096969255118e-06,
+      "loss": 1.6826,
+      "step": 4220
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3769000668576682e-06,
+      "loss": 1.6736,
+      "step": 4222
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3704046559080364e-06,
+      "loss": 1.6661,
+      "step": 4224
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3639234747872631e-06,
+      "loss": 1.7033,
+      "step": 4226
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3574565341825362e-06,
+      "loss": 1.685,
+      "step": 4228
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3510038447575613e-06,
+      "loss": 1.692,
+      "step": 4230
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3445654171525435e-06,
+      "loss": 1.6571,
+      "step": 4232
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3381412619841716e-06,
+      "loss": 1.6703,
+      "step": 4234
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3317313898455997e-06,
+      "loss": 1.6652,
+      "step": 4236
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3253358113064273e-06,
+      "loss": 1.6959,
+      "step": 4238
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3189545369126888e-06,
+      "loss": 1.7001,
+      "step": 4240
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3125875771868267e-06,
+      "loss": 1.6994,
+      "step": 4242
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3062349426276832e-06,
+      "loss": 1.7067,
+      "step": 4244
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2998966437104765e-06,
+      "loss": 1.7069,
+      "step": 4246
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2935726908867874e-06,
+      "loss": 1.6534,
+      "step": 4248
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2872630945845399e-06,
+      "loss": 1.6984,
+      "step": 4250
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.280967865207986e-06,
+      "loss": 1.6811,
+      "step": 4252
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2746870131376855e-06,
+      "loss": 1.6826,
+      "step": 4254
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2684205487304935e-06,
+      "loss": 1.6856,
+      "step": 4256
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2621684823195334e-06,
+      "loss": 1.702,
+      "step": 4258
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2559308242141955e-06,
+      "loss": 1.6667,
+      "step": 4260
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2497075847001073e-06,
+      "loss": 1.7403,
+      "step": 4262
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2434987740391203e-06,
+      "loss": 1.7048,
+      "step": 4264
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2373044024692938e-06,
+      "loss": 1.6566,
+      "step": 4266
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2311244802048806e-06,
+      "loss": 1.7021,
+      "step": 4268
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2249590174363024e-06,
+      "loss": 1.6777,
+      "step": 4270
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2188080243301438e-06,
+      "loss": 1.6937,
+      "step": 4272
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2126715110291199e-06,
+      "loss": 1.6675,
+      "step": 4274
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2065494876520801e-06,
+      "loss": 1.6821,
+      "step": 4276
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2004419642939735e-06,
+      "loss": 1.6434,
+      "step": 4278
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1943489510258444e-06,
+      "loss": 1.721,
+      "step": 4280
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1882704578948045e-06,
+      "loss": 1.6389,
+      "step": 4282
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.182206494924023e-06,
+      "loss": 1.6688,
+      "step": 4284
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1761570721127191e-06,
+      "loss": 1.6982,
+      "step": 4286
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1701221994361277e-06,
+      "loss": 1.7247,
+      "step": 4288
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1641018868454878e-06,
+      "loss": 1.6867,
+      "step": 4290
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1580961442680372e-06,
+      "loss": 1.7177,
+      "step": 4292
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.152104981606984e-06,
+      "loss": 1.6641,
+      "step": 4294
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1461284087414993e-06,
+      "loss": 1.6726,
+      "step": 4296
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1401664355266872e-06,
+      "loss": 1.6823,
+      "step": 4298
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1342190717935853e-06,
+      "loss": 1.6865,
+      "step": 4300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1282863273491408e-06,
+      "loss": 1.6226,
+      "step": 4302
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1223682119761926e-06,
+      "loss": 1.7109,
+      "step": 4304
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1164647354334536e-06,
+      "loss": 1.6268,
+      "step": 4306
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1105759074555012e-06,
+      "loss": 1.7408,
+      "step": 4308
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.104701737752759e-06,
+      "loss": 1.6519,
+      "step": 4310
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0988422360114814e-06,
+      "loss": 1.7205,
+      "step": 4312
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0929974118937258e-06,
+      "loss": 1.6745,
+      "step": 4314
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0871672750373585e-06,
+      "loss": 1.634,
+      "step": 4316
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0813518350560192e-06,
+      "loss": 1.7084,
+      "step": 4318
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0755511015391228e-06,
+      "loss": 1.7412,
+      "step": 4320
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0697650840518226e-06,
+      "loss": 1.6887,
+      "step": 4322
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0639937921350118e-06,
+      "loss": 1.6502,
+      "step": 4324
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0582372353053005e-06,
+      "loss": 1.6919,
+      "step": 4326
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0524954230550043e-06,
+      "loss": 1.6986,
+      "step": 4328
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0467683648521188e-06,
+      "loss": 1.6331,
+      "step": 4330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.041056070140316e-06,
+      "loss": 1.7176,
+      "step": 4332
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0353585483389227e-06,
+      "loss": 1.6957,
+      "step": 4334
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.029675808842906e-06,
+      "loss": 1.656,
+      "step": 4336
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0240078610228566e-06,
+      "loss": 1.6799,
+      "step": 4338
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.018354714224976e-06,
+      "loss": 1.6579,
+      "step": 4340
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.012716377771057e-06,
+      "loss": 1.6799,
+      "step": 4342
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0070928609584752e-06,
+      "loss": 1.7249,
+      "step": 4344
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.001484173060161e-06,
+      "loss": 1.708,
+      "step": 4346
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.958903233245998e-07,
+      "loss": 1.684,
+      "step": 4348
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 1.7048,
+      "step": 4350
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.847471752133208e-07,
+      "loss": 1.7328,
+      "step": 4352
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.791978952121705e-07,
+      "loss": 1.6642,
+      "step": 4354
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.736634901228814e-07,
+      "loss": 1.6788,
+      "step": 4356
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.681439690714456e-07,
+      "loss": 1.638,
+      "step": 4358
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.626393411593171e-07,
+      "loss": 1.7067,
+      "step": 4360
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.571496154633863e-07,
+      "loss": 1.6852,
+      "step": 4362
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.516748010359733e-07,
+      "loss": 1.7521,
+      "step": 4364
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.46214906904811e-07,
+      "loss": 1.7091,
+      "step": 4366
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.407699420730287e-07,
+      "loss": 1.6925,
+      "step": 4368
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.353399155191334e-07,
+      "loss": 1.7111,
+      "step": 4370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.299248361970092e-07,
+      "loss": 1.6441,
+      "step": 4372
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.245247130358837e-07,
+      "loss": 1.6535,
+      "step": 4374
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.191395549403304e-07,
+      "loss": 1.652,
+      "step": 4376
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.137693707902362e-07,
+      "loss": 1.64,
+      "step": 4378
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.084141694408044e-07,
+      "loss": 1.6707,
+      "step": 4380
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.030739597225302e-07,
+      "loss": 1.7144,
+      "step": 4382
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.977487504411897e-07,
+      "loss": 1.6946,
+      "step": 4384
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.924385503778166e-07,
+      "loss": 1.6865,
+      "step": 4386
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.871433682887076e-07,
+      "loss": 1.7206,
+      "step": 4388
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.818632129053861e-07,
+      "loss": 1.6914,
+      "step": 4390
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.765980929345985e-07,
+      "loss": 1.6411,
+      "step": 4392
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.713480170582999e-07,
+      "loss": 1.6902,
+      "step": 4394
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.661129939336377e-07,
+      "loss": 1.6516,
+      "step": 4396
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.608930321929398e-07,
+      "loss": 1.6931,
+      "step": 4398
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.556881404436967e-07,
+      "loss": 1.6475,
+      "step": 4400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.504983272685474e-07,
+      "loss": 1.6893,
+      "step": 4402
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.453236012252686e-07,
+      "loss": 1.6824,
+      "step": 4404
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.427418985455859e-07,
+      "loss": 1.6853,
+      "step": 4406
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.375898191915244e-07,
+      "loss": 1.7274,
+      "step": 4408
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.324528482549055e-07,
+      "loss": 1.6576,
+      "step": 4410
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.273309942063745e-07,
+      "loss": 1.6874,
+      "step": 4412
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.222242654916413e-07,
+      "loss": 1.6864,
+      "step": 4414
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.17132670531483e-07,
+      "loss": 1.7053,
+      "step": 4416
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.120562177217206e-07,
+      "loss": 1.6266,
+      "step": 4418
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.069949154332047e-07,
+      "loss": 1.6783,
+      "step": 4420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.01948772011798e-07,
+      "loss": 1.6454,
+      "step": 4422
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.969177957783791e-07,
+      "loss": 1.7116,
+      "step": 4424
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.919019950288076e-07,
+      "loss": 1.6624,
+      "step": 4426
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.86901378033924e-07,
+      "loss": 1.6788,
+      "step": 4428
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819159530395238e-07,
+      "loss": 1.712,
+      "step": 4430
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.769457282663595e-07,
+      "loss": 1.7096,
+      "step": 4432
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.719907119101178e-07,
+      "loss": 1.6761,
+      "step": 4434
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.670509121414049e-07,
+      "loss": 1.7139,
+      "step": 4436
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.621263371057352e-07,
+      "loss": 1.7139,
+      "step": 4438
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.572169949235198e-07,
+      "loss": 1.7088,
+      "step": 4440
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.523228936900539e-07,
+      "loss": 1.6932,
+      "step": 4442
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.47444041475498e-07,
+      "loss": 1.6889,
+      "step": 4444
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.425804463248654e-07,
+      "loss": 1.6598,
+      "step": 4446
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.377321162580153e-07,
+      "loss": 1.6555,
+      "step": 4448
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.32899059269635e-07,
+      "loss": 1.6567,
+      "step": 4450
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.28081283329225e-07,
+      "loss": 1.6951,
+      "step": 4452
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.232787963810884e-07,
+      "loss": 1.7267,
+      "step": 4454
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.18491606344317e-07,
+      "loss": 1.6765,
+      "step": 4456
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.137197211127822e-07,
+      "loss": 1.674,
+      "step": 4458
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.089631485551184e-07,
+      "loss": 1.7166,
+      "step": 4460
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.042218965147029e-07,
+      "loss": 1.69,
+      "step": 4462
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.994959728096562e-07,
+      "loss": 1.697,
+      "step": 4464
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.947853852328224e-07,
+      "loss": 1.6717,
+      "step": 4466
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.90090141551758e-07,
+      "loss": 1.6852,
+      "step": 4468
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.854102495087123e-07,
+      "loss": 1.6443,
+      "step": 4470
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.807457168206266e-07,
+      "loss": 1.6351,
+      "step": 4472
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.760965511791151e-07,
+      "loss": 1.6775,
+      "step": 4474
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.714627602504476e-07,
+      "loss": 1.6713,
+      "step": 4476
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.668443516755463e-07,
+      "loss": 1.7247,
+      "step": 4478
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.622413330699662e-07,
+      "loss": 1.6939,
+      "step": 4480
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.576537120238857e-07,
+      "loss": 1.7095,
+      "step": 4482
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.530814961020948e-07,
+      "loss": 1.6632,
+      "step": 4484
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.485246928439759e-07,
+      "loss": 1.651,
+      "step": 4486
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.439833097635029e-07,
+      "loss": 1.6496,
+      "step": 4488
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.394573543492188e-07,
+      "loss": 1.6899,
+      "step": 4490
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.349468340642284e-07,
+      "loss": 1.6627,
+      "step": 4492
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.304517563461832e-07,
+      "loss": 1.7039,
+      "step": 4494
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.259721286072729e-07,
+      "loss": 1.6799,
+      "step": 4496
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.215079582342099e-07,
+      "loss": 1.7124,
+      "step": 4498
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.17059252588218e-07,
+      "loss": 1.6938,
+      "step": 4500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.126260190050171e-07,
+      "loss": 1.6587,
+      "step": 4502
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.082082647948196e-07,
+      "loss": 1.6976,
+      "step": 4504
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.038059972423105e-07,
+      "loss": 1.6126,
+      "step": 4506
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.994192236066376e-07,
+      "loss": 1.704,
+      "step": 4508
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.950479511214025e-07,
+      "loss": 1.6969,
+      "step": 4510
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.906921869946425e-07,
+      "loss": 1.7187,
+      "step": 4512
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.863519384088235e-07,
+      "loss": 1.6232,
+      "step": 4514
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.820272125208304e-07,
+      "loss": 1.6623,
+      "step": 4516
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.777180164619445e-07,
+      "loss": 1.7064,
+      "step": 4518
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.734243573378462e-07,
+      "loss": 1.6498,
+      "step": 4520
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.691462422285921e-07,
+      "loss": 1.6293,
+      "step": 4522
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.648836781886113e-07,
+      "loss": 1.7038,
+      "step": 4524
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.606366722466816e-07,
+      "loss": 1.6831,
+      "step": 4526
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.564052314059376e-07,
+      "loss": 1.6717,
+      "step": 4528
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.521893626438412e-07,
+      "loss": 1.6384,
+      "step": 4530
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.47989072912174e-07,
+      "loss": 1.6866,
+      "step": 4532
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.438043691370343e-07,
+      "loss": 1.7289,
+      "step": 4534
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.396352582188158e-07,
+      "loss": 1.7279,
+      "step": 4536
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.354817470322038e-07,
+      "loss": 1.6376,
+      "step": 4538
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.313438424261574e-07,
+      "loss": 1.7013,
+      "step": 4540
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.272215512238999e-07,
+      "loss": 1.6658,
+      "step": 4542
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.231148802229147e-07,
+      "loss": 1.6671,
+      "step": 4544
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.190238361949229e-07,
+      "loss": 1.6846,
+      "step": 4546
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.149484258858772e-07,
+      "loss": 1.7105,
+      "step": 4548
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.108886560159543e-07,
+      "loss": 1.6776,
+      "step": 4550
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.06844533279538e-07,
+      "loss": 1.706,
+      "step": 4552
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.02816064345213e-07,
+      "loss": 1.7162,
+      "step": 4554
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.988032558557488e-07,
+      "loss": 1.7033,
+      "step": 4556
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.948061144280924e-07,
+      "loss": 1.6739,
+      "step": 4558
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.908246466533562e-07,
+      "loss": 1.6754,
+      "step": 4560
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.868588590968115e-07,
+      "loss": 1.6763,
+      "step": 4562
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.829087582978664e-07,
+      "loss": 1.7312,
+      "step": 4564
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.789743507700684e-07,
+      "loss": 1.6977,
+      "step": 4566
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.750556430010833e-07,
+      "loss": 1.7283,
+      "step": 4568
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.7115264145269123e-07,
+      "loss": 1.7249,
+      "step": 4570
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.672653525607751e-07,
+      "loss": 1.6424,
+      "step": 4572
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6339378273530076e-07,
+      "loss": 1.6499,
+      "step": 4574
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5953793836032266e-07,
+      "loss": 1.6256,
+      "step": 4576
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.556978257939604e-07,
+      "loss": 1.6947,
+      "step": 4578
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.518734513683931e-07,
+      "loss": 1.7562,
+      "step": 4580
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4806482138984863e-07,
+      "loss": 1.6536,
+      "step": 4582
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 1.7332,
+      "step": 4584
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.404948198689196e-07,
+      "loss": 1.6487,
+      "step": 4586
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3673346080913894e-07,
+      "loss": 1.7043,
+      "step": 4588
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.329878711615698e-07,
+      "loss": 1.6823,
+      "step": 4590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.292580571025273e-07,
+      "loss": 1.6815,
+      "step": 4592
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.255440247823117e-07,
+      "loss": 1.7014,
+      "step": 4594
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.218457803252041e-07,
+      "loss": 1.689,
+      "step": 4596
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.181633298294474e-07,
+      "loss": 1.6404,
+      "step": 4598
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1449667936724335e-07,
+      "loss": 1.6571,
+      "step": 4600
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1084583498474105e-07,
+      "loss": 1.6508,
+      "step": 4602
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.072108027020216e-07,
+      "loss": 1.6943,
+      "step": 4604
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.035915885130981e-07,
+      "loss": 1.7096,
+      "step": 4606
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9998819838589773e-07,
+      "loss": 1.6743,
+      "step": 4608
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.964006382622543e-07,
+      "loss": 1.6699,
+      "step": 4610
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9282891405790003e-07,
+      "loss": 1.6594,
+      "step": 4612
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.892730316624516e-07,
+      "loss": 1.6932,
+      "step": 4614
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8573299693940635e-07,
+      "loss": 1.6982,
+      "step": 4616
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.8220881572612833e-07,
+      "loss": 1.687,
+      "step": 4618
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.78700493833839e-07,
+      "loss": 1.7376,
+      "step": 4620
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7520803704760855e-07,
+      "loss": 1.6929,
+      "step": 4622
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.717314511263481e-07,
+      "loss": 1.7304,
+      "step": 4624
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6827074180279643e-07,
+      "loss": 1.6991,
+      "step": 4626
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.6482591478351423e-07,
+      "loss": 1.6367,
+      "step": 4628
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.613969757488711e-07,
+      "loss": 1.6283,
+      "step": 4630
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5798393035303856e-07,
+      "loss": 1.6867,
+      "step": 4632
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.545867842239836e-07,
+      "loss": 1.6911,
+      "step": 4634
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.512055429634498e-07,
+      "loss": 1.6823,
+      "step": 4636
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.478402121469571e-07,
+      "loss": 1.6709,
+      "step": 4638
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.444907973237932e-07,
+      "loss": 1.6824,
+      "step": 4640
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4115730401699686e-07,
+      "loss": 1.7071,
+      "step": 4642
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.378397377233522e-07,
+      "loss": 1.6824,
+      "step": 4644
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3453810391338324e-07,
+      "loss": 1.6902,
+      "step": 4646
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.312524080313428e-07,
+      "loss": 1.6503,
+      "step": 4648
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2798265549520034e-07,
+      "loss": 1.6265,
+      "step": 4650
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2472885169663514e-07,
+      "loss": 1.684,
+      "step": 4652
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2149100200102865e-07,
+      "loss": 1.7013,
+      "step": 4654
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1826911174745566e-07,
+      "loss": 1.6736,
+      "step": 4656
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.1506318624867525e-07,
+      "loss": 1.6726,
+      "step": 4658
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.1187323079111545e-07,
+      "loss": 1.6752,
+      "step": 4660
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.086992506348774e-07,
+      "loss": 1.6772,
+      "step": 4662
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.055412510137157e-07,
+      "loss": 1.6843,
+      "step": 4664
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0239923713503706e-07,
+      "loss": 1.6491,
+      "step": 4666
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9927321417988377e-07,
+      "loss": 1.6988,
+      "step": 4668
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.961631873029336e-07,
+      "loss": 1.6461,
+      "step": 4670
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.930691616324854e-07,
+      "loss": 1.6879,
+      "step": 4672
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.899911422704549e-07,
+      "loss": 1.7102,
+      "step": 4674
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.869291342923608e-07,
+      "loss": 1.6824,
+      "step": 4676
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8388314274732207e-07,
+      "loss": 1.6492,
+      "step": 4678
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8085317265804655e-07,
+      "loss": 1.6898,
+      "step": 4680
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.778392290208265e-07,
+      "loss": 1.6967,
+      "step": 4682
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.748413168055208e-07,
+      "loss": 1.689,
+      "step": 4684
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.718594409555575e-07,
+      "loss": 1.6954,
+      "step": 4686
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6889360638792216e-07,
+      "loss": 1.6827,
+      "step": 4688
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.659438179931473e-07,
+      "loss": 1.6579,
+      "step": 4690
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.630100806353053e-07,
+      "loss": 1.6822,
+      "step": 4692
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6009239915200214e-07,
+      "loss": 1.6219,
+      "step": 4694
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.57190778354367e-07,
+      "loss": 1.6541,
+      "step": 4696
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.543052230270493e-07,
+      "loss": 1.6682,
+      "step": 4698
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5143573792820287e-07,
+      "loss": 1.6877,
+      "step": 4700
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.485823277894861e-07,
+      "loss": 1.6712,
+      "step": 4702
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.457449973160464e-07,
+      "loss": 1.6634,
+      "step": 4704
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4292375118652233e-07,
+      "loss": 1.6844,
+      "step": 4706
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.401185940530226e-07,
+      "loss": 1.6844,
+      "step": 4708
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3732953054113161e-07,
+      "loss": 1.6439,
+      "step": 4710
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.345565652498938e-07,
+      "loss": 1.6426,
+      "step": 4712
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.317997027518104e-07,
+      "loss": 1.7018,
+      "step": 4714
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.29058947592824e-07,
+      "loss": 1.6842,
+      "step": 4716
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2633430429232606e-07,
+      "loss": 1.6846,
+      "step": 4718
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2362577734313385e-07,
+      "loss": 1.7205,
+      "step": 4720
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.209333712114903e-07,
+      "loss": 1.6764,
+      "step": 4722
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1825709033705622e-07,
+      "loss": 1.6167,
+      "step": 4724
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1692499825141654e-07,
+      "loss": 1.6302,
+      "step": 4726
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1427291352899447e-07,
+      "loss": 1.6741,
+      "step": 4728
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.116369650465888e-07,
+      "loss": 1.7534,
+      "step": 4730
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0901715715076264e-07,
+      "loss": 1.67,
+      "step": 4732
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0641349416146817e-07,
+      "loss": 1.6788,
+      "step": 4734
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0382598037203217e-07,
+      "loss": 1.6955,
+      "step": 4736
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0125462004915164e-07,
+      "loss": 1.6924,
+      "step": 4738
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9869941743288935e-07,
+      "loss": 1.6893,
+      "step": 4740
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9616037673666378e-07,
+      "loss": 1.6298,
+      "step": 4742
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9363750214724254e-07,
+      "loss": 1.6908,
+      "step": 4744
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9113079782473676e-07,
+      "loss": 1.6628,
+      "step": 4746
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8864026790259226e-07,
+      "loss": 1.6284,
+      "step": 4748
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.861659164875873e-07,
+      "loss": 1.7271,
+      "step": 4750
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8370774765981813e-07,
+      "loss": 1.6778,
+      "step": 4752
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8126576547269902e-07,
+      "loss": 1.658,
+      "step": 4754
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7883997395295448e-07,
+      "loss": 1.6874,
+      "step": 4756
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7643037710060818e-07,
+      "loss": 1.6754,
+      "step": 4758
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.740369788889795e-07,
+      "loss": 1.6754,
+      "step": 4760
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.7165978326468048e-07,
+      "loss": 1.6646,
+      "step": 4762
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.692987941476032e-07,
+      "loss": 1.6493,
+      "step": 4764
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6695401543091682e-07,
+      "loss": 1.6583,
+      "step": 4766
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6462545098105963e-07,
+      "loss": 1.6365,
+      "step": 4768
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6231310463773242e-07,
+      "loss": 1.6684,
+      "step": 4770
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6001698021389512e-07,
+      "loss": 1.6427,
+      "step": 4772
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5773708149575905e-07,
+      "loss": 1.6601,
+      "step": 4774
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.554734122427748e-07,
+      "loss": 1.7091,
+      "step": 4776
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5322597618763756e-07,
+      "loss": 1.6305,
+      "step": 4778
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5099477703627074e-07,
+      "loss": 1.7016,
+      "step": 4780
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4877981846782685e-07,
+      "loss": 1.6369,
+      "step": 4782
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.465811041346743e-07,
+      "loss": 1.6551,
+      "step": 4784
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.443986376623996e-07,
+      "loss": 1.6988,
+      "step": 4786
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.42232422649794e-07,
+      "loss": 1.6768,
+      "step": 4788
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.400824626688535e-07,
+      "loss": 1.6815,
+      "step": 4790
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3794876126476785e-07,
+      "loss": 1.7146,
+      "step": 4792
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.358313219559193e-07,
+      "loss": 1.6187,
+      "step": 4794
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3373014823387154e-07,
+      "loss": 1.643,
+      "step": 4796
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3164524356337193e-07,
+      "loss": 1.7074,
+      "step": 4798
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2957661138233602e-07,
+      "loss": 1.6628,
+      "step": 4800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.275242551018485e-07,
+      "loss": 1.6454,
+      "step": 4802
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2548817810615676e-07,
+      "loss": 1.716,
+      "step": 4804
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2346838375266402e-07,
+      "loss": 1.6828,
+      "step": 4806
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2146487537192275e-07,
+      "loss": 1.6452,
+      "step": 4808
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1947765626763142e-07,
+      "loss": 1.6995,
+      "step": 4810
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.175067297166288e-07,
+      "loss": 1.6592,
+      "step": 4812
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1555209896888741e-07,
+      "loss": 1.6575,
+      "step": 4814
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1361376724750683e-07,
+      "loss": 1.6953,
+      "step": 4816
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 1.6579,
+      "step": 4818
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0978601364185271e-07,
+      "loss": 1.6241,
+      "step": 4820
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0789659806937913e-07,
+      "loss": 1.7087,
+      "step": 4822
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0602349414685631e-07,
+      "loss": 1.6812,
+      "step": 4824
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0416670496295244e-07,
+      "loss": 1.6668,
+      "step": 4826
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0232623357943283e-07,
+      "loss": 1.7054,
+      "step": 4828
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0050208303115539e-07,
+      "loss": 1.6249,
+      "step": 4830
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.869425632606288e-08,
+      "loss": 1.6125,
+      "step": 4832
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.690275644518521e-08,
+      "loss": 1.6892,
+      "step": 4834
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.51275863426282e-08,
+      "loss": 1.6877,
+      "step": 4836
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.336874894557035e-08,
+      "loss": 1.6876,
+      "step": 4838
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.162624715425727e-08,
+      "loss": 1.6551,
+      "step": 4840
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.990008384199944e-08,
+      "loss": 1.7193,
+      "step": 4842
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.819026185516667e-08,
+      "loss": 1.6894,
+      "step": 4844
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.649678401318029e-08,
+      "loss": 1.6861,
+      "step": 4846
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.481965310851103e-08,
+      "loss": 1.6662,
+      "step": 4848
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.31588719066756e-08,
+      "loss": 1.6273,
+      "step": 4850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.151444314623114e-08,
+      "loss": 1.7157,
+      "step": 4852
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.98863695387686e-08,
+      "loss": 1.6811,
+      "step": 4854
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.827465376891052e-08,
+      "loss": 1.697,
+      "step": 4856
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.667929849430767e-08,
+      "loss": 1.6701,
+      "step": 4858
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.510030634563125e-08,
+      "loss": 1.6494,
+      "step": 4860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.353767992657079e-08,
+      "loss": 1.6894,
+      "step": 4862
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.199142181382956e-08,
+      "loss": 1.694,
+      "step": 4864
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.046153455711913e-08,
+      "loss": 1.7133,
+      "step": 4866
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.894802067915929e-08,
+      "loss": 1.6999,
+      "step": 4868
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.745088267566702e-08,
+      "loss": 1.6855,
+      "step": 4870
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.597012301535976e-08,
+      "loss": 1.7207,
+      "step": 4872
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.450574413994437e-08,
+      "loss": 1.6981,
+      "step": 4874
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.305774846411927e-08,
+      "loss": 1.6912,
+      "step": 4876
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.162613837556674e-08,
+      "loss": 1.7138,
+      "step": 4878
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.021091623494846e-08,
+      "loss": 1.7084,
+      "step": 4880
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.8812084375905464e-08,
+      "loss": 1.7082,
+      "step": 4882
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.7429645105050444e-08,
+      "loss": 1.7096,
+      "step": 4884
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.60636007019677e-08,
+      "loss": 1.6792,
+      "step": 4886
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.471395341920205e-08,
+      "loss": 1.7069,
+      "step": 4888
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.338070548226548e-08,
+      "loss": 1.615,
+      "step": 4890
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.206385908962608e-08,
+      "loss": 1.7502,
+      "step": 4892
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.076341641270577e-08,
+      "loss": 1.7245,
+      "step": 4894
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.947937959587812e-08,
+      "loss": 1.6174,
+      "step": 4896
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8211750756465e-08,
+      "loss": 1.6765,
+      "step": 4898
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.696053198473105e-08,
+      "loss": 1.6386,
+      "step": 4900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.572572534388253e-08,
+      "loss": 1.7684,
+      "step": 4902
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.4507332870059594e-08,
+      "loss": 1.6886,
+      "step": 4904
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.330535657234181e-08,
+      "loss": 1.6844,
+      "step": 4906
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2119798432733725e-08,
+      "loss": 1.6571,
+      "step": 4908
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.095066040617046e-08,
+      "loss": 1.7035,
+      "step": 4910
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.979794442050988e-08,
+      "loss": 1.6507,
+      "step": 4912
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.8661652376531524e-08,
+      "loss": 1.6743,
+      "step": 4914
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7541786147929914e-08,
+      "loss": 1.6558,
+      "step": 4916
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.643834758131681e-08,
+      "loss": 1.7059,
+      "step": 4918
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5351338496214524e-08,
+      "loss": 1.6817,
+      "step": 4920
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.42807606850537e-08,
+      "loss": 1.6801,
+      "step": 4922
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.322661591317222e-08,
+      "loss": 1.73,
+      "step": 4924
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.218890591880741e-08,
+      "loss": 1.694,
+      "step": 4926
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.116763241309828e-08,
+      "loss": 1.6608,
+      "step": 4928
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.016279708008218e-08,
+      "loss": 1.6694,
+      "step": 4930
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.917440157668705e-08,
+      "loss": 1.6606,
+      "step": 4932
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.820244753273582e-08,
+      "loss": 1.7309,
+      "step": 4934
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7246936550936466e-08,
+      "loss": 1.6596,
+      "step": 4936
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.630787020688752e-08,
+      "loss": 1.6344,
+      "step": 4938
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.538525004906811e-08,
+      "loss": 1.673,
+      "step": 4940
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4479077598837942e-08,
+      "loss": 1.6836,
+      "step": 4942
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.35893543504373e-08,
+      "loss": 1.6181,
+      "step": 4944
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2716081770980392e-08,
+      "loss": 1.7039,
+      "step": 4946
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1859261300456458e-08,
+      "loss": 1.6639,
+      "step": 4948
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1018894351725326e-08,
+      "loss": 1.6495,
+      "step": 4950
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.01949823105152e-08,
+      "loss": 1.6508,
+      "step": 4952
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.938752653542153e-08,
+      "loss": 1.7073,
+      "step": 4954
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.859652835790371e-08,
+      "loss": 1.6456,
+      "step": 4956
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7821989082283942e-08,
+      "loss": 1.7032,
+      "step": 4958
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.706390998574392e-08,
+      "loss": 1.711,
+      "step": 4960
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.632229231832372e-08,
+      "loss": 1.7041,
+      "step": 4962
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5597137302917343e-08,
+      "loss": 1.6802,
+      "step": 4964
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4888446135274959e-08,
+      "loss": 1.7243,
+      "step": 4966
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4196219983996228e-08,
+      "loss": 1.7065,
+      "step": 4968
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3520459990534752e-08,
+      "loss": 1.6566,
+      "step": 4970
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2861167269186959e-08,
+      "loss": 1.6964,
+      "step": 4972
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2218342907099889e-08,
+      "loss": 1.6947,
+      "step": 4974
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1591987964261198e-08,
+      "loss": 1.6369,
+      "step": 4976
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0982103473504702e-08,
+      "loss": 1.6791,
+      "step": 4978
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0388690440502613e-08,
+      "loss": 1.7309,
+      "step": 4980
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.811749843767759e-09,
+      "loss": 1.6665,
+      "step": 4982
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.251282634649139e-09,
+      "loss": 1.7215,
+      "step": 4984
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.707289737334146e-09,
+      "loss": 1.6596,
+      "step": 4986
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.179772048843016e-09,
+      "loss": 1.6664,
+      "step": 4988
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.668730439029937e-09,
+      "loss": 1.7287,
+      "step": 4990
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.1741657505797206e-09,
+      "loss": 1.6738,
+      "step": 4992
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.696078799008909e-09,
+      "loss": 1.658,
+      "step": 4994
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.234470372661338e-09,
+      "loss": 1.6769,
+      "step": 4996
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.7893412327092445e-09,
+      "loss": 1.6286,
+      "step": 4998
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.360692113152155e-09,
+      "loss": 1.6605,
+      "step": 5000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.948523720813558e-09,
+      "loss": 1.7059,
+      "step": 5002
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.552836735340904e-09,
+      "loss": 1.6603,
+      "step": 5004
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.173631809205603e-09,
+      "loss": 1.6746,
+      "step": 5006
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.810909567699694e-09,
+      "loss": 1.6709,
+      "step": 5008
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4646706089358493e-09,
+      "loss": 1.6987,
+      "step": 5010
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.1349155038484793e-09,
+      "loss": 1.6765,
+      "step": 5012
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.821644796188183e-09,
+      "loss": 1.6781,
+      "step": 5014
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5248590025250817e-09,
+      "loss": 1.6501,
+      "step": 5016
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2445586122465946e-09,
+      "loss": 1.6948,
+      "step": 5018
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.98074408755633e-09,
+      "loss": 1.6914,
+      "step": 5020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7334158634718656e-09,
+      "loss": 1.7532,
+      "step": 5022
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5025743478269683e-09,
+      "loss": 1.6991,
+      "step": 5024
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2882199212693736e-09,
+      "loss": 1.6498,
+      "step": 5026
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0903529372607857e-09,
+      "loss": 1.7046,
+      "step": 5028
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.089737220746575e-10,
+      "loss": 1.6628,
+      "step": 5030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.440825747984104e-10,
+      "loss": 1.66,
+      "step": 5032
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.956797673289937e-10,
+      "loss": 1.6812,
+      "step": 5034
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.637655443773259e-10,
+      "loss": 1.6565,
+      "step": 5036
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.483401234627426e-10,
+      "loss": 1.679,
+      "step": 5038
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.4940369491854857e-10,
+      "loss": 1.6856,
+      "step": 5040
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.669564218864661e-10,
+      "loss": 1.6809,
+      "step": 5042
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0099844031774553e-10,
+      "loss": 1.6455,
+      "step": 5044
+    },
+    {
+      "epoch": 5.0,
+      "step": 5045,
+      "total_flos": 1.88606135857578e+17,
+      "train_loss": 1.8637235274045507,
+      "train_runtime": 40193.4201,
+      "train_samples_per_second": 8.035,
+      "train_steps_per_second": 0.126
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 5045,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 3000,
+  "total_flos": 1.88606135857578e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}