diff --git "a/checkpoint-67458/trainer_state.json" "b/checkpoint-67458/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-67458/trainer_state.json"
@@ -0,0 +1,6345 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 67458,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.990957336416734e-07,
+      "loss": 0.1541,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.981469951673633e-07,
+      "loss": 0.0788,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971982566930536e-07,
+      "loss": 0.2255,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.962495182187434e-07,
+      "loss": 0.1454,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.953007797444335e-07,
+      "loss": 0.081,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943520412701236e-07,
+      "loss": 0.1782,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.934033027958137e-07,
+      "loss": 0.1119,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.924545643215036e-07,
+      "loss": 0.1648,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.915058258471937e-07,
+      "loss": 0.1536,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.905570873728838e-07,
+      "loss": 0.1527,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.896083488985739e-07,
+      "loss": 0.1153,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.88659610424264e-07,
+      "loss": 0.0826,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.87710871949954e-07,
+      "loss": 0.135,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.867621334756441e-07,
+      "loss": 0.1657,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.85813395001334e-07,
+      "loss": 0.1777,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848646565270243e-07,
+      "loss": 0.0975,
+      "step": 1024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.839159180527142e-07,
+      "loss": 0.1336,
+      "step": 1088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.829671795784043e-07,
+      "loss": 0.0785,
+      "step": 1152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.820184411040944e-07,
+      "loss": 0.164,
+      "step": 1216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.810697026297845e-07,
+      "loss": 0.0864,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.801209641554746e-07,
+      "loss": 0.128,
+      "step": 1344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.791722256811644e-07,
+      "loss": 0.0897,
+      "step": 1408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782234872068545e-07,
+      "loss": 0.0989,
+      "step": 1472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.772747487325446e-07,
+      "loss": 0.1309,
+      "step": 1536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.763408342968958e-07,
+      "loss": 0.2144,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.75392095822586e-07,
+      "loss": 0.0691,
+      "step": 1664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.744433573482758e-07,
+      "loss": 0.1476,
+      "step": 1728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.73494618873966e-07,
+      "loss": 0.1541,
+      "step": 1792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.72545880399656e-07,
+      "loss": 0.1108,
+      "step": 1856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.71597141925346e-07,
+      "loss": 0.112,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706484034510362e-07,
+      "loss": 0.1901,
+      "step": 1984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.696996649767262e-07,
+      "loss": 0.1207,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.687509265024163e-07,
+      "loss": 0.128,
+      "step": 2112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.678021880281062e-07,
+      "loss": 0.2029,
+      "step": 2176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.668534495537965e-07,
+      "loss": 0.1629,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.659047110794864e-07,
+      "loss": 0.1662,
+      "step": 2304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.649559726051765e-07,
+      "loss": 0.1392,
+      "step": 2368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.640072341308666e-07,
+      "loss": 0.081,
+      "step": 2432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.630584956565567e-07,
+      "loss": 0.0894,
+      "step": 2496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.621097571822468e-07,
+      "loss": 0.1017,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.611610187079368e-07,
+      "loss": 0.1476,
+      "step": 2624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.602122802336267e-07,
+      "loss": 0.1592,
+      "step": 2688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.592635417593168e-07,
+      "loss": 0.146,
+      "step": 2752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58314803285007e-07,
+      "loss": 0.1552,
+      "step": 2816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57366064810697e-07,
+      "loss": 0.0498,
+      "step": 2880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56417326336387e-07,
+      "loss": 0.0999,
+      "step": 2944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.55468587862077e-07,
+      "loss": 0.1196,
+      "step": 3008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545198493877673e-07,
+      "loss": 0.1172,
+      "step": 3072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.535711109134573e-07,
+      "loss": 0.1107,
+      "step": 3136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526223724391472e-07,
+      "loss": 0.182,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.516736339648373e-07,
+      "loss": 0.0687,
+      "step": 3264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.507248954905273e-07,
+      "loss": 0.1274,
+      "step": 3328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497761570162175e-07,
+      "loss": 0.1374,
+      "step": 3392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.488274185419075e-07,
+      "loss": 0.159,
+      "step": 3456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.478786800675975e-07,
+      "loss": 0.0687,
+      "step": 3520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.469299415932877e-07,
+      "loss": 0.1446,
+      "step": 3584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.459812031189777e-07,
+      "loss": 0.1047,
+      "step": 3648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.450324646446678e-07,
+      "loss": 0.1445,
+      "step": 3712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.440837261703578e-07,
+      "loss": 0.1033,
+      "step": 3776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.431349876960478e-07,
+      "loss": 0.1514,
+      "step": 3840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.421862492217379e-07,
+      "loss": 0.1245,
+      "step": 3904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41237510747428e-07,
+      "loss": 0.2032,
+      "step": 3968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402887722731181e-07,
+      "loss": 0.0935,
+      "step": 4032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393400337988081e-07,
+      "loss": 0.1492,
+      "step": 4096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383912953244981e-07,
+      "loss": 0.1011,
+      "step": 4160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374425568501883e-07,
+      "loss": 0.1279,
+      "step": 4224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364938183758783e-07,
+      "loss": 0.0842,
+      "step": 4288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355599039402295e-07,
+      "loss": 0.1482,
+      "step": 4352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.346111654659194e-07,
+      "loss": 0.1444,
+      "step": 4416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.336624269916095e-07,
+      "loss": 0.1124,
+      "step": 4480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.327285125559607e-07,
+      "loss": 0.1552,
+      "step": 4544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.317797740816507e-07,
+      "loss": 0.113,
+      "step": 4608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.308310356073408e-07,
+      "loss": 0.1406,
+      "step": 4672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.298822971330309e-07,
+      "loss": 0.1464,
+      "step": 4736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.289335586587209e-07,
+      "loss": 0.0889,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279848201844111e-07,
+      "loss": 0.1028,
+      "step": 4864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270360817101011e-07,
+      "loss": 0.1816,
+      "step": 4928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26087343235791e-07,
+      "loss": 0.0867,
+      "step": 4992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.251386047614812e-07,
+      "loss": 0.1426,
+      "step": 5056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.241898662871712e-07,
+      "loss": 0.1404,
+      "step": 5120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.232411278128613e-07,
+      "loss": 0.0599,
+      "step": 5184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.222923893385513e-07,
+      "loss": 0.0753,
+      "step": 5248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.213436508642414e-07,
+      "loss": 0.1109,
+      "step": 5312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.203949123899315e-07,
+      "loss": 0.1836,
+      "step": 5376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.194461739156215e-07,
+      "loss": 0.0633,
+      "step": 5440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.184974354413117e-07,
+      "loss": 0.1392,
+      "step": 5504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.175486969670017e-07,
+      "loss": 0.11,
+      "step": 5568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.165999584926916e-07,
+      "loss": 0.1901,
+      "step": 5632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.156512200183818e-07,
+      "loss": 0.1399,
+      "step": 5696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.147024815440718e-07,
+      "loss": 0.1312,
+      "step": 5760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.137537430697619e-07,
+      "loss": 0.0884,
+      "step": 5824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.12805004595452e-07,
+      "loss": 0.1663,
+      "step": 5888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.11856266121142e-07,
+      "loss": 0.0728,
+      "step": 5952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.109075276468321e-07,
+      "loss": 0.1709,
+      "step": 6016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.099587891725221e-07,
+      "loss": 0.1476,
+      "step": 6080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.090100506982122e-07,
+      "loss": 0.0778,
+      "step": 6144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.080613122239022e-07,
+      "loss": 0.1081,
+      "step": 6208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.071125737495922e-07,
+      "loss": 0.1059,
+      "step": 6272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.061638352752824e-07,
+      "loss": 0.1143,
+      "step": 6336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.052150968009724e-07,
+      "loss": 0.1205,
+      "step": 6400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.042663583266624e-07,
+      "loss": 0.1479,
+      "step": 6464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.033176198523526e-07,
+      "loss": 0.0817,
+      "step": 6528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.023688813780426e-07,
+      "loss": 0.1398,
+      "step": 6592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.014201429037327e-07,
+      "loss": 0.1608,
+      "step": 6656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.004714044294228e-07,
+      "loss": 0.1345,
+      "step": 6720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.995226659551127e-07,
+      "loss": 0.1773,
+      "step": 6784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.985739274808028e-07,
+      "loss": 0.1769,
+      "step": 6848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.976251890064928e-07,
+      "loss": 0.1345,
+      "step": 6912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.96676450532183e-07,
+      "loss": 0.1293,
+      "step": 6976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.95727712057873e-07,
+      "loss": 0.1388,
+      "step": 7040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.94778973583563e-07,
+      "loss": 0.1004,
+      "step": 7104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.938302351092532e-07,
+      "loss": 0.1363,
+      "step": 7168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.928814966349432e-07,
+      "loss": 0.0933,
+      "step": 7232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.919327581606333e-07,
+      "loss": 0.1617,
+      "step": 7296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.909840196863233e-07,
+      "loss": 0.1085,
+      "step": 7360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.900352812120133e-07,
+      "loss": 0.1836,
+      "step": 7424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.890865427377034e-07,
+      "loss": 0.1464,
+      "step": 7488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.881378042633935e-07,
+      "loss": 0.1166,
+      "step": 7552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.871890657890836e-07,
+      "loss": 0.1071,
+      "step": 7616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.862403273147736e-07,
+      "loss": 0.1564,
+      "step": 7680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.852915888404636e-07,
+      "loss": 0.0937,
+      "step": 7744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.843428503661538e-07,
+      "loss": 0.1095,
+      "step": 7808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.833941118918438e-07,
+      "loss": 0.1313,
+      "step": 7872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.824453734175337e-07,
+      "loss": 0.0819,
+      "step": 7936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.814966349432239e-07,
+      "loss": 0.1273,
+      "step": 8000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.805478964689139e-07,
+      "loss": 0.1079,
+      "step": 8064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.79599157994604e-07,
+      "loss": 0.0566,
+      "step": 8128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.786504195202941e-07,
+      "loss": 0.0882,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.777016810459841e-07,
+      "loss": 0.1801,
+      "step": 8256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.767529425716742e-07,
+      "loss": 0.1006,
+      "step": 8320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.758042040973643e-07,
+      "loss": 0.1033,
+      "step": 8384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.748554656230544e-07,
+      "loss": 0.0907,
+      "step": 8448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.739067271487443e-07,
+      "loss": 0.0548,
+      "step": 8512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.729579886744343e-07,
+      "loss": 0.1173,
+      "step": 8576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.720092502001245e-07,
+      "loss": 0.1415,
+      "step": 8640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.710605117258145e-07,
+      "loss": 0.1035,
+      "step": 8704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.701117732515046e-07,
+      "loss": 0.1016,
+      "step": 8768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.691630347771947e-07,
+      "loss": 0.111,
+      "step": 8832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.682291203415458e-07,
+      "loss": 0.1427,
+      "step": 8896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.672803818672359e-07,
+      "loss": 0.1181,
+      "step": 8960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.66331643392926e-07,
+      "loss": 0.097,
+      "step": 9024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.65382904918616e-07,
+      "loss": 0.1367,
+      "step": 9088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.64434166444306e-07,
+      "loss": 0.1212,
+      "step": 9152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.634854279699961e-07,
+      "loss": 0.1393,
+      "step": 9216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.625366894956861e-07,
+      "loss": 0.1322,
+      "step": 9280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.615879510213763e-07,
+      "loss": 0.0786,
+      "step": 9344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.606392125470663e-07,
+      "loss": 0.1263,
+      "step": 9408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.596904740727563e-07,
+      "loss": 0.1537,
+      "step": 9472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.587417355984465e-07,
+      "loss": 0.1267,
+      "step": 9536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.577929971241365e-07,
+      "loss": 0.1087,
+      "step": 9600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.568442586498266e-07,
+      "loss": 0.1097,
+      "step": 9664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.558955201755165e-07,
+      "loss": 0.1423,
+      "step": 9728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.549467817012066e-07,
+      "loss": 0.0911,
+      "step": 9792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.539980432268967e-07,
+      "loss": 0.0748,
+      "step": 9856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.530493047525867e-07,
+      "loss": 0.112,
+      "step": 9920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.521005662782769e-07,
+      "loss": 0.1765,
+      "step": 9984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.511518278039669e-07,
+      "loss": 0.1539,
+      "step": 10048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.502030893296569e-07,
+      "loss": 0.1382,
+      "step": 10112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.492543508553471e-07,
+      "loss": 0.1146,
+      "step": 10176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.48305612381037e-07,
+      "loss": 0.1589,
+      "step": 10240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.473568739067271e-07,
+      "loss": 0.1184,
+      "step": 10304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.464081354324172e-07,
+      "loss": 0.0892,
+      "step": 10368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.454593969581072e-07,
+      "loss": 0.1883,
+      "step": 10432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.445106584837973e-07,
+      "loss": 0.1441,
+      "step": 10496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.435619200094873e-07,
+      "loss": 0.1233,
+      "step": 10560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.426131815351774e-07,
+      "loss": 0.1376,
+      "step": 10624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.416644430608675e-07,
+      "loss": 0.057,
+      "step": 10688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.407157045865575e-07,
+      "loss": 0.0966,
+      "step": 10752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.397669661122477e-07,
+      "loss": 0.1026,
+      "step": 10816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.388182276379376e-07,
+      "loss": 0.1161,
+      "step": 10880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.378694891636276e-07,
+      "loss": 0.1889,
+      "step": 10944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.369207506893178e-07,
+      "loss": 0.0808,
+      "step": 11008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.359868362536689e-07,
+      "loss": 0.8512,
+      "step": 11072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.35038097779359e-07,
+      "loss": 0.1572,
+      "step": 11136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.34089359305049e-07,
+      "loss": 0.1883,
+      "step": 11200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.331406208307391e-07,
+      "loss": 0.0772,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.321918823564291e-07,
+      "loss": 0.0924,
+      "step": 11328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.312431438821193e-07,
+      "loss": 0.1308,
+      "step": 11392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.302944054078093e-07,
+      "loss": 0.2317,
+      "step": 11456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.293456669334992e-07,
+      "loss": 0.1581,
+      "step": 11520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.283969284591894e-07,
+      "loss": 0.1068,
+      "step": 11584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.274481899848794e-07,
+      "loss": 0.0793,
+      "step": 11648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.264994515105695e-07,
+      "loss": 0.1407,
+      "step": 11712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.255507130362596e-07,
+      "loss": 0.2219,
+      "step": 11776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.246019745619496e-07,
+      "loss": 0.1364,
+      "step": 11840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.236532360876397e-07,
+      "loss": 0.0842,
+      "step": 11904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.227193216519909e-07,
+      "loss": 0.0893,
+      "step": 11968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.217705831776809e-07,
+      "loss": 0.1798,
+      "step": 12032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.20821844703371e-07,
+      "loss": 0.1263,
+      "step": 12096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.19873106229061e-07,
+      "loss": 0.1289,
+      "step": 12160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.18924367754751e-07,
+      "loss": 0.1712,
+      "step": 12224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.179756292804412e-07,
+      "loss": 0.2207,
+      "step": 12288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.170268908061312e-07,
+      "loss": 0.1219,
+      "step": 12352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.160781523318212e-07,
+      "loss": 0.1446,
+      "step": 12416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.151294138575113e-07,
+      "loss": 0.076,
+      "step": 12480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.141806753832014e-07,
+      "loss": 0.1448,
+      "step": 12544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.132319369088915e-07,
+      "loss": 0.1613,
+      "step": 12608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.122831984345814e-07,
+      "loss": 0.1241,
+      "step": 12672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113344599602715e-07,
+      "loss": 0.1833,
+      "step": 12736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103857214859616e-07,
+      "loss": 0.2071,
+      "step": 12800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094369830116516e-07,
+      "loss": 0.1467,
+      "step": 12864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084882445373418e-07,
+      "loss": 0.0889,
+      "step": 12928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.075395060630318e-07,
+      "loss": 0.141,
+      "step": 12992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065907675887218e-07,
+      "loss": 0.106,
+      "step": 13056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05642029114412e-07,
+      "loss": 0.1359,
+      "step": 13120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.04693290640102e-07,
+      "loss": 0.2273,
+      "step": 13184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.03744552165792e-07,
+      "loss": 0.1071,
+      "step": 13248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.02795813691482e-07,
+      "loss": 0.1106,
+      "step": 13312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.018470752171721e-07,
+      "loss": 0.1035,
+      "step": 13376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.008983367428622e-07,
+      "loss": 0.1459,
+      "step": 13440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.999495982685522e-07,
+      "loss": 0.1739,
+      "step": 13504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.990008597942423e-07,
+      "loss": 0.116,
+      "step": 13568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.980521213199324e-07,
+      "loss": 0.1017,
+      "step": 13632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.971033828456224e-07,
+      "loss": 0.1414,
+      "step": 13696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.961546443713126e-07,
+      "loss": 0.2619,
+      "step": 13760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.952059058970026e-07,
+      "loss": 0.1166,
+      "step": 13824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.942571674226925e-07,
+      "loss": 0.142,
+      "step": 13888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.933084289483827e-07,
+      "loss": 0.2068,
+      "step": 13952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.923596904740727e-07,
+      "loss": 0.1119,
+      "step": 14016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.914109519997628e-07,
+      "loss": 0.143,
+      "step": 14080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.904770375641139e-07,
+      "loss": 0.1718,
+      "step": 14144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.89528299089804e-07,
+      "loss": 0.0871,
+      "step": 14208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.88579560615494e-07,
+      "loss": 0.1059,
+      "step": 14272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.876308221411842e-07,
+      "loss": 0.1276,
+      "step": 14336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.866820836668742e-07,
+      "loss": 0.1503,
+      "step": 14400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.857333451925641e-07,
+      "loss": 0.2201,
+      "step": 14464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.847846067182543e-07,
+      "loss": 0.1737,
+      "step": 14528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.838358682439443e-07,
+      "loss": 0.1133,
+      "step": 14592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.828871297696344e-07,
+      "loss": 0.1378,
+      "step": 14656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.819383912953245e-07,
+      "loss": 0.166,
+      "step": 14720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.809896528210145e-07,
+      "loss": 0.2001,
+      "step": 14784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.800409143467046e-07,
+      "loss": 0.158,
+      "step": 14848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.790921758723946e-07,
+      "loss": 0.1009,
+      "step": 14912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.781434373980848e-07,
+      "loss": 0.2009,
+      "step": 14976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.771946989237747e-07,
+      "loss": 0.1437,
+      "step": 15040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.762459604494647e-07,
+      "loss": 0.1653,
+      "step": 15104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.752972219751549e-07,
+      "loss": 0.1736,
+      "step": 15168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.743484835008449e-07,
+      "loss": 0.2137,
+      "step": 15232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.73399745026535e-07,
+      "loss": 0.1282,
+      "step": 15296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.724510065522251e-07,
+      "loss": 0.1766,
+      "step": 15360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.715022680779151e-07,
+      "loss": 0.0876,
+      "step": 15424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.705535296036052e-07,
+      "loss": 0.1338,
+      "step": 15488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.696047911292953e-07,
+      "loss": 0.2006,
+      "step": 15552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.686560526549854e-07,
+      "loss": 0.1505,
+      "step": 15616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.677073141806753e-07,
+      "loss": 0.1223,
+      "step": 15680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.667585757063653e-07,
+      "loss": 0.1693,
+      "step": 15744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.658098372320555e-07,
+      "loss": 0.1872,
+      "step": 15808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.648610987577455e-07,
+      "loss": 0.1109,
+      "step": 15872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.639123602834355e-07,
+      "loss": 0.1351,
+      "step": 15936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.629636218091257e-07,
+      "loss": 0.1336,
+      "step": 16000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.620148833348157e-07,
+      "loss": 0.2153,
+      "step": 16064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.610661448605058e-07,
+      "loss": 0.2274,
+      "step": 16128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.601174063861959e-07,
+      "loss": 0.105,
+      "step": 16192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.591686679118858e-07,
+      "loss": 0.1016,
+      "step": 16256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.582199294375759e-07,
+      "loss": 0.116,
+      "step": 16320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.57271190963266e-07,
+      "loss": 0.1086,
+      "step": 16384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.563224524889561e-07,
+      "loss": 0.1153,
+      "step": 16448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.553737140146461e-07,
+      "loss": 0.1044,
+      "step": 16512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.544249755403361e-07,
+      "loss": 0.2228,
+      "step": 16576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.534762370660263e-07,
+      "loss": 0.1002,
+      "step": 16640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.525274985917163e-07,
+      "loss": 0.1919,
+      "step": 16704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.515787601174063e-07,
+      "loss": 0.13,
+      "step": 16768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.506300216430964e-07,
+      "loss": 0.1165,
+      "step": 16832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.496812831687864e-07,
+      "loss": 0.1385,
+      "step": 16896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.487325446944765e-07,
+      "loss": 0.1836,
+      "step": 16960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.477838062201666e-07,
+      "loss": 0.1852,
+      "step": 17024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.468498917845177e-07,
+      "loss": 0.1721,
+      "step": 17088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.459011533102078e-07,
+      "loss": 0.1164,
+      "step": 17152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.449524148358979e-07,
+      "loss": 0.0989,
+      "step": 17216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.440036763615879e-07,
+      "loss": 0.1524,
+      "step": 17280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.430549378872781e-07,
+      "loss": 0.174,
+      "step": 17344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.42106199412968e-07,
+      "loss": 0.1509,
+      "step": 17408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.41157460938658e-07,
+      "loss": 0.1264,
+      "step": 17472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.402087224643482e-07,
+      "loss": 0.1148,
+      "step": 17536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.392599839900382e-07,
+      "loss": 0.1227,
+      "step": 17600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.383112455157283e-07,
+      "loss": 0.1462,
+      "step": 17664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.373625070414183e-07,
+      "loss": 0.1673,
+      "step": 17728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.364137685671084e-07,
+      "loss": 0.1274,
+      "step": 17792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.354650300927985e-07,
+      "loss": 0.162,
+      "step": 17856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.345162916184885e-07,
+      "loss": 0.148,
+      "step": 17920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.335675531441787e-07,
+      "loss": 0.157,
+      "step": 17984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.326188146698686e-07,
+      "loss": 0.1419,
+      "step": 18048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.316700761955586e-07,
+      "loss": 0.1916,
+      "step": 18112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.307213377212488e-07,
+      "loss": 0.1772,
+      "step": 18176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.297725992469388e-07,
+      "loss": 0.1243,
+      "step": 18240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.288238607726288e-07,
+      "loss": 0.1738,
+      "step": 18304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.27875122298319e-07,
+      "loss": 0.165,
+      "step": 18368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.26926383824009e-07,
+      "loss": 0.0869,
+      "step": 18432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.259776453496991e-07,
+      "loss": 0.1613,
+      "step": 18496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.25028906875389e-07,
+      "loss": 0.1646,
+      "step": 18560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.240801684010791e-07,
+      "loss": 0.1232,
+      "step": 18624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.231314299267692e-07,
+      "loss": 0.1695,
+      "step": 18688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.221826914524592e-07,
+      "loss": 0.1305,
+      "step": 18752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.212339529781494e-07,
+      "loss": 0.2047,
+      "step": 18816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.202852145038394e-07,
+      "loss": 0.1881,
+      "step": 18880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.193364760295294e-07,
+      "loss": 0.1513,
+      "step": 18944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.183877375552196e-07,
+      "loss": 0.1355,
+      "step": 19008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.174389990809096e-07,
+      "loss": 0.1733,
+      "step": 19072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.164902606065996e-07,
+      "loss": 0.1647,
+      "step": 19136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.155415221322897e-07,
+      "loss": 0.132,
+      "step": 19200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.145927836579797e-07,
+      "loss": 0.2464,
+      "step": 19264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.136440451836698e-07,
+      "loss": 0.1954,
+      "step": 19328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.126953067093598e-07,
+      "loss": 0.1751,
+      "step": 19392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.1174656823505e-07,
+      "loss": 0.1182,
+      "step": 19456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.1079782976074e-07,
+      "loss": 0.0678,
+      "step": 19520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.0984909128643e-07,
+      "loss": 0.1656,
+      "step": 19584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.089003528121202e-07,
+      "loss": 0.219,
+      "step": 19648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079516143378101e-07,
+      "loss": 0.15,
+      "step": 19712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.070028758635002e-07,
+      "loss": 0.1845,
+      "step": 19776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060541373891903e-07,
+      "loss": 0.1766,
+      "step": 19840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.051053989148803e-07,
+      "loss": 0.1355,
+      "step": 19904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.041566604405704e-07,
+      "loss": 0.1268,
+      "step": 19968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.032079219662605e-07,
+      "loss": 0.191,
+      "step": 20032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.022591834919506e-07,
+      "loss": 0.2032,
+      "step": 20096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.013104450176406e-07,
+      "loss": 0.2632,
+      "step": 20160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.003617065433306e-07,
+      "loss": 0.2028,
+      "step": 20224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.994129680690208e-07,
+      "loss": 0.135,
+      "step": 20288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.984642295947107e-07,
+      "loss": 0.1635,
+      "step": 20352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.975154911204007e-07,
+      "loss": 0.1986,
+      "step": 20416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.965667526460909e-07,
+      "loss": 0.1786,
+      "step": 20480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.956180141717809e-07,
+      "loss": 0.116,
+      "step": 20544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.94669275697471e-07,
+      "loss": 0.2206,
+      "step": 20608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.937205372231611e-07,
+      "loss": 0.0995,
+      "step": 20672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.927717987488511e-07,
+      "loss": 0.1634,
+      "step": 20736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.918230602745412e-07,
+      "loss": 0.1429,
+      "step": 20800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.908743218002312e-07,
+      "loss": 0.2244,
+      "step": 20864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.899255833259213e-07,
+      "loss": 0.1666,
+      "step": 20928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.889768448516113e-07,
+      "loss": 0.1413,
+      "step": 20992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.880281063773013e-07,
+      "loss": 0.1824,
+      "step": 21056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.870941919416525e-07,
+      "loss": 0.1737,
+      "step": 21120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.861454534673426e-07,
+      "loss": 0.171,
+      "step": 21184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.851967149930327e-07,
+      "loss": 0.1571,
+      "step": 21248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.842479765187227e-07,
+      "loss": 0.115,
+      "step": 21312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.832992380444128e-07,
+      "loss": 0.2192,
+      "step": 21376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.823504995701029e-07,
+      "loss": 0.146,
+      "step": 21440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.81401761095793e-07,
+      "loss": 0.1586,
+      "step": 21504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.804530226214829e-07,
+      "loss": 0.1035,
+      "step": 21568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.795191081858341e-07,
+      "loss": 0.1898,
+      "step": 21632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.785703697115241e-07,
+      "loss": 0.2901,
+      "step": 21696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.776216312372143e-07,
+      "loss": 0.1986,
+      "step": 21760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.766728927629043e-07,
+      "loss": 0.1774,
+      "step": 21824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.757241542885943e-07,
+      "loss": 0.1945,
+      "step": 21888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.747754158142845e-07,
+      "loss": 0.1453,
+      "step": 21952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.738266773399745e-07,
+      "loss": 0.1882,
+      "step": 22016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.728779388656646e-07,
+      "loss": 0.2266,
+      "step": 22080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.719292003913545e-07,
+      "loss": 0.2192,
+      "step": 22144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.709804619170446e-07,
+      "loss": 0.1191,
+      "step": 22208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.700317234427347e-07,
+      "loss": 0.1323,
+      "step": 22272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.690829849684247e-07,
+      "loss": 0.2345,
+      "step": 22336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.681342464941149e-07,
+      "loss": 0.122,
+      "step": 22400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.671855080198049e-07,
+      "loss": 0.1643,
+      "step": 22464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.662367695454949e-07,
+      "loss": 0.1756,
+      "step": 22528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.653028551098461e-07,
+      "loss": 0.0794,
+      "step": 22592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.643541166355362e-07,
+      "loss": 0.0636,
+      "step": 22656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.634053781612263e-07,
+      "loss": 0.1252,
+      "step": 22720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.624566396869162e-07,
+      "loss": 0.1114,
+      "step": 22784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.615079012126063e-07,
+      "loss": 0.1799,
+      "step": 22848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.605591627382963e-07,
+      "loss": 0.1088,
+      "step": 22912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.596104242639865e-07,
+      "loss": 0.0468,
+      "step": 22976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.586616857896765e-07,
+      "loss": 0.0891,
+      "step": 23040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.577129473153665e-07,
+      "loss": 0.0967,
+      "step": 23104
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.567642088410567e-07,
+      "loss": 0.102,
+      "step": 23168
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.558154703667467e-07,
+      "loss": 0.0713,
+      "step": 23232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.548667318924368e-07,
+      "loss": 0.0656,
+      "step": 23296
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.539179934181268e-07,
+      "loss": 0.1185,
+      "step": 23360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.529692549438168e-07,
+      "loss": 0.0787,
+      "step": 23424
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.520205164695069e-07,
+      "loss": 0.0442,
+      "step": 23488
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.51071777995197e-07,
+      "loss": 0.0793,
+      "step": 23552
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.501230395208871e-07,
+      "loss": 0.1479,
+      "step": 23616
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.491743010465771e-07,
+      "loss": 0.0891,
+      "step": 23680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.482255625722671e-07,
+      "loss": 0.0879,
+      "step": 23744
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.472768240979573e-07,
+      "loss": 0.0888,
+      "step": 23808
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.463429096623084e-07,
+      "loss": 0.1321,
+      "step": 23872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.453941711879985e-07,
+      "loss": 0.0842,
+      "step": 23936
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.444454327136884e-07,
+      "loss": 0.1128,
+      "step": 24000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.434966942393785e-07,
+      "loss": 0.1089,
+      "step": 24064
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.425479557650686e-07,
+      "loss": 0.078,
+      "step": 24128
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.415992172907586e-07,
+      "loss": 0.1481,
+      "step": 24192
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.406504788164487e-07,
+      "loss": 0.0639,
+      "step": 24256
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.397017403421388e-07,
+      "loss": 0.1182,
+      "step": 24320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.387530018678289e-07,
+      "loss": 0.0864,
+      "step": 24384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.378042633935189e-07,
+      "loss": 0.0626,
+      "step": 24448
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.36855524919209e-07,
+      "loss": 0.0763,
+      "step": 24512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.35906786444899e-07,
+      "loss": 0.1031,
+      "step": 24576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.34958047970589e-07,
+      "loss": 0.0772,
+      "step": 24640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.340093094962791e-07,
+      "loss": 0.1173,
+      "step": 24704
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.330605710219692e-07,
+      "loss": 0.1254,
+      "step": 24768
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.321118325476592e-07,
+      "loss": 0.1646,
+      "step": 24832
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.311630940733493e-07,
+      "loss": 0.1065,
+      "step": 24896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.302143555990394e-07,
+      "loss": 0.0916,
+      "step": 24960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.292656171247295e-07,
+      "loss": 0.0769,
+      "step": 25024
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.283168786504195e-07,
+      "loss": 0.05,
+      "step": 25088
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.273681401761095e-07,
+      "loss": 0.0748,
+      "step": 25152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.264194017017996e-07,
+      "loss": 0.089,
+      "step": 25216
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.254706632274896e-07,
+      "loss": 0.0951,
+      "step": 25280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.245219247531798e-07,
+      "loss": 0.0955,
+      "step": 25344
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.235731862788698e-07,
+      "loss": 0.1024,
+      "step": 25408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.226244478045598e-07,
+      "loss": 0.0898,
+      "step": 25472
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.216757093302499e-07,
+      "loss": 0.1351,
+      "step": 25536
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.2072697085594e-07,
+      "loss": 0.0797,
+      "step": 25600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.197782323816301e-07,
+      "loss": 0.0459,
+      "step": 25664
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.1882949390732e-07,
+      "loss": 0.0463,
+      "step": 25728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.178807554330101e-07,
+      "loss": 0.0487,
+      "step": 25792
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.169320169587002e-07,
+      "loss": 0.0971,
+      "step": 25856
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.159832784843902e-07,
+      "loss": 0.0793,
+      "step": 25920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.150345400100804e-07,
+      "loss": 0.0507,
+      "step": 25984
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.140858015357704e-07,
+      "loss": 0.1244,
+      "step": 26048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.131370630614604e-07,
+      "loss": 0.1191,
+      "step": 26112
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.121883245871506e-07,
+      "loss": 0.1716,
+      "step": 26176
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.112395861128406e-07,
+      "loss": 0.1052,
+      "step": 26240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.102908476385305e-07,
+      "loss": 0.1227,
+      "step": 26304
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.093421091642206e-07,
+      "loss": 0.0686,
+      "step": 26368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.083933706899107e-07,
+      "loss": 0.1174,
+      "step": 26432
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.074446322156008e-07,
+      "loss": 0.0606,
+      "step": 26496
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.064958937412908e-07,
+      "loss": 0.1176,
+      "step": 26560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.055471552669809e-07,
+      "loss": 0.0385,
+      "step": 26624
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.04598416792671e-07,
+      "loss": 0.0468,
+      "step": 26688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.03649678318361e-07,
+      "loss": 0.1169,
+      "step": 26752
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.027009398440512e-07,
+      "loss": 0.1115,
+      "step": 26816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.017522013697411e-07,
+      "loss": 0.1258,
+      "step": 26880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.008034628954311e-07,
+      "loss": 0.1025,
+      "step": 26944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.998547244211213e-07,
+      "loss": 0.0834,
+      "step": 27008
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.989059859468113e-07,
+      "loss": 0.0421,
+      "step": 27072
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.979572474725014e-07,
+      "loss": 0.0936,
+      "step": 27136
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.970085089981914e-07,
+      "loss": 0.1253,
+      "step": 27200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.960597705238815e-07,
+      "loss": 0.1037,
+      "step": 27264
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.951110320495716e-07,
+      "loss": 0.0553,
+      "step": 27328
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.941622935752616e-07,
+      "loss": 0.0932,
+      "step": 27392
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.932135551009518e-07,
+      "loss": 0.0359,
+      "step": 27456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.922648166266417e-07,
+      "loss": 0.0564,
+      "step": 27520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.913160781523317e-07,
+      "loss": 0.0475,
+      "step": 27584
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.903673396780219e-07,
+      "loss": 0.1187,
+      "step": 27648
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.894186012037119e-07,
+      "loss": 0.102,
+      "step": 27712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.88469862729402e-07,
+      "loss": 0.1048,
+      "step": 27776
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.875211242550921e-07,
+      "loss": 0.1066,
+      "step": 27840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.865723857807821e-07,
+      "loss": 0.1039,
+      "step": 27904
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.856236473064722e-07,
+      "loss": 0.1027,
+      "step": 27968
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.846749088321621e-07,
+      "loss": 0.0535,
+      "step": 28032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.837261703578523e-07,
+      "loss": 0.0737,
+      "step": 28096
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.827774318835423e-07,
+      "loss": 0.094,
+      "step": 28160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.818286934092323e-07,
+      "loss": 0.1002,
+      "step": 28224
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.808799549349225e-07,
+      "loss": 0.0618,
+      "step": 28288
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.799312164606125e-07,
+      "loss": 0.1356,
+      "step": 28352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.789824779863025e-07,
+      "loss": 0.0802,
+      "step": 28416
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.780337395119927e-07,
+      "loss": 0.1333,
+      "step": 28480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.770850010376827e-07,
+      "loss": 0.0687,
+      "step": 28544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.761362625633727e-07,
+      "loss": 0.0562,
+      "step": 28608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.751875240890628e-07,
+      "loss": 0.0955,
+      "step": 28672
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.742387856147528e-07,
+      "loss": 0.0915,
+      "step": 28736
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.732900471404429e-07,
+      "loss": 0.0632,
+      "step": 28800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.723413086661329e-07,
+      "loss": 0.1047,
+      "step": 28864
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.713925701918231e-07,
+      "loss": 0.0981,
+      "step": 28928
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.704438317175131e-07,
+      "loss": 0.0954,
+      "step": 28992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.694950932432031e-07,
+      "loss": 0.0572,
+      "step": 29056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.685463547688933e-07,
+      "loss": 0.1177,
+      "step": 29120
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.675976162945832e-07,
+      "loss": 0.0748,
+      "step": 29184
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.666488778202733e-07,
+      "loss": 0.0783,
+      "step": 29248
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.657001393459634e-07,
+      "loss": 0.0556,
+      "step": 29312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.647514008716534e-07,
+      "loss": 0.0592,
+      "step": 29376
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.638026623973435e-07,
+      "loss": 0.0917,
+      "step": 29440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.628539239230336e-07,
+      "loss": 0.0847,
+      "step": 29504
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.619051854487237e-07,
+      "loss": 0.1253,
+      "step": 29568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.609564469744137e-07,
+      "loss": 0.1362,
+      "step": 29632
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.600077085001037e-07,
+      "loss": 0.0799,
+      "step": 29696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.590589700257938e-07,
+      "loss": 0.1235,
+      "step": 29760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.581102315514838e-07,
+      "loss": 0.056,
+      "step": 29824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.571614930771739e-07,
+      "loss": 0.0952,
+      "step": 29888
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.56212754602864e-07,
+      "loss": 0.102,
+      "step": 29952
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.55264016128554e-07,
+      "loss": 0.105,
+      "step": 30016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.543152776542441e-07,
+      "loss": 0.1049,
+      "step": 30080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.533665391799342e-07,
+      "loss": 0.0722,
+      "step": 30144
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.524178007056243e-07,
+      "loss": 0.0389,
+      "step": 30208
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.514690622313143e-07,
+      "loss": 0.0589,
+      "step": 30272
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.505203237570043e-07,
+      "loss": 0.1049,
+      "step": 30336
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.495715852826944e-07,
+      "loss": 0.1013,
+      "step": 30400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.486228468083844e-07,
+      "loss": 0.0866,
+      "step": 30464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.476741083340744e-07,
+      "loss": 0.1195,
+      "step": 30528
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.467253698597646e-07,
+      "loss": 0.049,
+      "step": 30592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.457766313854546e-07,
+      "loss": 0.0521,
+      "step": 30656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.448278929111447e-07,
+      "loss": 0.0658,
+      "step": 30720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.438791544368348e-07,
+      "loss": 0.0735,
+      "step": 30784
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.429304159625248e-07,
+      "loss": 0.1463,
+      "step": 30848
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.419816774882148e-07,
+      "loss": 0.1333,
+      "step": 30912
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.410329390139049e-07,
+      "loss": 0.0795,
+      "step": 30976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.40084200539595e-07,
+      "loss": 0.0599,
+      "step": 31040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.39135462065285e-07,
+      "loss": 0.1246,
+      "step": 31104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.381867235909751e-07,
+      "loss": 0.0802,
+      "step": 31168
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.372379851166652e-07,
+      "loss": 0.0565,
+      "step": 31232
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.362892466423552e-07,
+      "loss": 0.0646,
+      "step": 31296
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.353405081680454e-07,
+      "loss": 0.0712,
+      "step": 31360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.343917696937354e-07,
+      "loss": 0.1214,
+      "step": 31424
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.334430312194253e-07,
+      "loss": 0.0654,
+      "step": 31488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.324942927451154e-07,
+      "loss": 0.1136,
+      "step": 31552
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.315455542708055e-07,
+      "loss": 0.0907,
+      "step": 31616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.305968157964956e-07,
+      "loss": 0.0797,
+      "step": 31680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.296480773221856e-07,
+      "loss": 0.0739,
+      "step": 31744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.286993388478757e-07,
+      "loss": 0.0763,
+      "step": 31808
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.277654244122268e-07,
+      "loss": 0.0724,
+      "step": 31872
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.26816685937917e-07,
+      "loss": 0.0855,
+      "step": 31936
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.25867947463607e-07,
+      "loss": 0.0873,
+      "step": 32000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.24919208989297e-07,
+      "loss": 0.0688,
+      "step": 32064
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.239704705149871e-07,
+      "loss": 0.0945,
+      "step": 32128
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.230217320406771e-07,
+      "loss": 0.1052,
+      "step": 32192
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.220729935663672e-07,
+      "loss": 0.057,
+      "step": 32256
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.211242550920573e-07,
+      "loss": 0.1135,
+      "step": 32320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.201755166177473e-07,
+      "loss": 0.0806,
+      "step": 32384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.192267781434374e-07,
+      "loss": 0.0931,
+      "step": 32448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.182780396691274e-07,
+      "loss": 0.0863,
+      "step": 32512
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.173293011948176e-07,
+      "loss": 0.0623,
+      "step": 32576
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.163805627205076e-07,
+      "loss": 0.0643,
+      "step": 32640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.154318242461975e-07,
+      "loss": 0.0693,
+      "step": 32704
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.144830857718877e-07,
+      "loss": 0.0433,
+      "step": 32768
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.135343472975777e-07,
+      "loss": 0.0808,
+      "step": 32832
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.125856088232677e-07,
+      "loss": 0.0939,
+      "step": 32896
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.116368703489579e-07,
+      "loss": 0.0852,
+      "step": 32960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.106881318746479e-07,
+      "loss": 0.155,
+      "step": 33024
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.09739393400338e-07,
+      "loss": 0.127,
+      "step": 33088
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.087906549260281e-07,
+      "loss": 0.0432,
+      "step": 33152
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.078419164517181e-07,
+      "loss": 0.1161,
+      "step": 33216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.068931779774081e-07,
+      "loss": 0.0877,
+      "step": 33280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.059444395030981e-07,
+      "loss": 0.0579,
+      "step": 33344
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.049957010287883e-07,
+      "loss": 0.0408,
+      "step": 33408
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.040469625544783e-07,
+      "loss": 0.062,
+      "step": 33472
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.030982240801683e-07,
+      "loss": 0.1147,
+      "step": 33536
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.021494856058585e-07,
+      "loss": 0.1064,
+      "step": 33600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.012007471315485e-07,
+      "loss": 0.1084,
+      "step": 33664
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.002520086572386e-07,
+      "loss": 0.1422,
+      "step": 33728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.993032701829287e-07,
+      "loss": 0.1086,
+      "step": 33792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.983545317086186e-07,
+      "loss": 0.0744,
+      "step": 33856
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.974057932343087e-07,
+      "loss": 0.1086,
+      "step": 33920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.964570547599988e-07,
+      "loss": 0.1106,
+      "step": 33984
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.955083162856888e-07,
+      "loss": 0.0761,
+      "step": 34048
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.945595778113789e-07,
+      "loss": 0.1153,
+      "step": 34112
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.93610839337069e-07,
+      "loss": 0.0666,
+      "step": 34176
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.92662100862759e-07,
+      "loss": 0.095,
+      "step": 34240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.917133623884491e-07,
+      "loss": 0.1187,
+      "step": 34304
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.907646239141392e-07,
+      "loss": 0.0593,
+      "step": 34368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.898158854398292e-07,
+      "loss": 0.0832,
+      "step": 34432
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.888671469655192e-07,
+      "loss": 0.0724,
+      "step": 34496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.879184084912093e-07,
+      "loss": 0.0552,
+      "step": 34560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.869696700168994e-07,
+      "loss": 0.0912,
+      "step": 34624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.860209315425894e-07,
+      "loss": 0.0644,
+      "step": 34688
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.850721930682795e-07,
+      "loss": 0.0813,
+      "step": 34752
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.841234545939696e-07,
+      "loss": 0.1482,
+      "step": 34816
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.831747161196596e-07,
+      "loss": 0.1192,
+      "step": 34880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.822259776453497e-07,
+      "loss": 0.09,
+      "step": 34944
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.812772391710397e-07,
+      "loss": 0.0527,
+      "step": 35008
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.803285006967298e-07,
+      "loss": 0.0867,
+      "step": 35072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.793797622224199e-07,
+      "loss": 0.0483,
+      "step": 35136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.784310237481099e-07,
+      "loss": 0.1053,
+      "step": 35200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.774822852738e-07,
+      "loss": 0.1272,
+      "step": 35264
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7653354679949004e-07,
+      "loss": 0.1117,
+      "step": 35328
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7558480832518013e-07,
+      "loss": 0.1487,
+      "step": 35392
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.746360698508701e-07,
+      "loss": 0.0671,
+      "step": 35456
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736873313765602e-07,
+      "loss": 0.0657,
+      "step": 35520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7273859290225024e-07,
+      "loss": 0.1076,
+      "step": 35584
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7178985442794033e-07,
+      "loss": 0.1425,
+      "step": 35648
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.708411159536304e-07,
+      "loss": 0.088,
+      "step": 35712
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.698923774793204e-07,
+      "loss": 0.1325,
+      "step": 35776
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.689436390050105e-07,
+      "loss": 0.1102,
+      "step": 35840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.679949005307006e-07,
+      "loss": 0.126,
+      "step": 35904
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.6704616205639063e-07,
+      "loss": 0.074,
+      "step": 35968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.661122476207417e-07,
+      "loss": 0.1062,
+      "step": 36032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.651635091464318e-07,
+      "loss": 0.0812,
+      "step": 36096
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.642147706721219e-07,
+      "loss": 0.1062,
+      "step": 36160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.6326603219781194e-07,
+      "loss": 0.0861,
+      "step": 36224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.6231729372350203e-07,
+      "loss": 0.0879,
+      "step": 36288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.6136855524919207e-07,
+      "loss": 0.0649,
+      "step": 36352
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.604198167748821e-07,
+      "loss": 0.0695,
+      "step": 36416
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.594710783005722e-07,
+      "loss": 0.0705,
+      "step": 36480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.5852233982626224e-07,
+      "loss": 0.1642,
+      "step": 36544
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.5757360135195233e-07,
+      "loss": 0.0944,
+      "step": 36608
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.5662486287764237e-07,
+      "loss": 0.0872,
+      "step": 36672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.556909484419935e-07,
+      "loss": 0.105,
+      "step": 36736
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.547422099676836e-07,
+      "loss": 0.0825,
+      "step": 36800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5379347149337364e-07,
+      "loss": 0.1063,
+      "step": 36864
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.528447330190637e-07,
+      "loss": 0.0384,
+      "step": 36928
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.518959945447537e-07,
+      "loss": 0.0765,
+      "step": 36992
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.509472560704438e-07,
+      "loss": 0.1533,
+      "step": 37056
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.499985175961339e-07,
+      "loss": 0.0634,
+      "step": 37120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.490497791218239e-07,
+      "loss": 0.074,
+      "step": 37184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.48101040647514e-07,
+      "loss": 0.0641,
+      "step": 37248
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4715230217320407e-07,
+      "loss": 0.0927,
+      "step": 37312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.462035636988941e-07,
+      "loss": 0.0837,
+      "step": 37376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.452548252245842e-07,
+      "loss": 0.1142,
+      "step": 37440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.443060867502742e-07,
+      "loss": 0.0672,
+      "step": 37504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4335734827596427e-07,
+      "loss": 0.0726,
+      "step": 37568
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4240860980165436e-07,
+      "loss": 0.1104,
+      "step": 37632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.414598713273444e-07,
+      "loss": 0.1157,
+      "step": 37696
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.405111328530345e-07,
+      "loss": 0.0694,
+      "step": 37760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3956239437872447e-07,
+      "loss": 0.0905,
+      "step": 37824
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3861365590441456e-07,
+      "loss": 0.134,
+      "step": 37888
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.3766491743010465e-07,
+      "loss": 0.0478,
+      "step": 37952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.367161789557947e-07,
+      "loss": 0.1028,
+      "step": 38016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.357674404814848e-07,
+      "loss": 0.1023,
+      "step": 38080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.348187020071748e-07,
+      "loss": 0.1258,
+      "step": 38144
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.3386996353286486e-07,
+      "loss": 0.0619,
+      "step": 38208
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.3292122505855495e-07,
+      "loss": 0.1135,
+      "step": 38272
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.31972486584245e-07,
+      "loss": 0.062,
+      "step": 38336
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.31023748109935e-07,
+      "loss": 0.0513,
+      "step": 38400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.300750096356251e-07,
+      "loss": 0.0451,
+      "step": 38464
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2912627116131515e-07,
+      "loss": 0.1072,
+      "step": 38528
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2817753268700524e-07,
+      "loss": 0.0962,
+      "step": 38592
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2722879421269533e-07,
+      "loss": 0.0748,
+      "step": 38656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.262800557383853e-07,
+      "loss": 0.1036,
+      "step": 38720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.253313172640754e-07,
+      "loss": 0.0929,
+      "step": 38784
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2438257878976545e-07,
+      "loss": 0.145,
+      "step": 38848
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2343384031545554e-07,
+      "loss": 0.0449,
+      "step": 38912
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2248510184114563e-07,
+      "loss": 0.1341,
+      "step": 38976
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.215363633668356e-07,
+      "loss": 0.1309,
+      "step": 39040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.205876248925257e-07,
+      "loss": 0.0446,
+      "step": 39104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1963888641821574e-07,
+      "loss": 0.0743,
+      "step": 39168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1870497198256694e-07,
+      "loss": 0.0653,
+      "step": 39232
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1775623350825693e-07,
+      "loss": 0.0822,
+      "step": 39296
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.16807495033947e-07,
+      "loss": 0.0777,
+      "step": 39360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.158587565596371e-07,
+      "loss": 0.126,
+      "step": 39424
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1491001808532715e-07,
+      "loss": 0.0348,
+      "step": 39488
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1396127961101724e-07,
+      "loss": 0.0783,
+      "step": 39552
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.130125411367072e-07,
+      "loss": 0.0846,
+      "step": 39616
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.120638026623973e-07,
+      "loss": 0.0792,
+      "step": 39680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.111150641880874e-07,
+      "loss": 0.0684,
+      "step": 39744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1016632571377744e-07,
+      "loss": 0.0712,
+      "step": 39808
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.092175872394675e-07,
+      "loss": 0.0962,
+      "step": 39872
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.0826884876515757e-07,
+      "loss": 0.0995,
+      "step": 39936
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.073201102908476e-07,
+      "loss": 0.0408,
+      "step": 40000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.063713718165377e-07,
+      "loss": 0.0742,
+      "step": 40064
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.0542263334222774e-07,
+      "loss": 0.0876,
+      "step": 40128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.044738948679178e-07,
+      "loss": 0.0912,
+      "step": 40192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.0352515639360786e-07,
+      "loss": 0.0635,
+      "step": 40256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.025764179192979e-07,
+      "loss": 0.0641,
+      "step": 40320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.01627679444988e-07,
+      "loss": 0.1015,
+      "step": 40384
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.006789409706781e-07,
+      "loss": 0.0551,
+      "step": 40448
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.9973020249636807e-07,
+      "loss": 0.1139,
+      "step": 40512
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.9878146402205816e-07,
+      "loss": 0.1007,
+      "step": 40576
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.978475495864093e-07,
+      "loss": 0.1153,
+      "step": 40640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.969136351507604e-07,
+      "loss": 0.0426,
+      "step": 40704
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.959648966764505e-07,
+      "loss": 0.1175,
+      "step": 40768
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.950161582021406e-07,
+      "loss": 0.0451,
+      "step": 40832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.940674197278306e-07,
+      "loss": 0.1535,
+      "step": 40896
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.931186812535207e-07,
+      "loss": 0.0703,
+      "step": 40960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9216994277921075e-07,
+      "loss": 0.1161,
+      "step": 41024
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.912212043049008e-07,
+      "loss": 0.0631,
+      "step": 41088
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.902724658305909e-07,
+      "loss": 0.086,
+      "step": 41152
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.893237273562809e-07,
+      "loss": 0.0811,
+      "step": 41216
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.88374988881971e-07,
+      "loss": 0.097,
+      "step": 41280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8742625040766105e-07,
+      "loss": 0.0747,
+      "step": 41344
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.864775119333511e-07,
+      "loss": 0.0843,
+      "step": 41408
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.855287734590412e-07,
+      "loss": 0.0942,
+      "step": 41472
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.845800349847312e-07,
+      "loss": 0.1175,
+      "step": 41536
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.836312965104213e-07,
+      "loss": 0.0899,
+      "step": 41600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8268255803611134e-07,
+      "loss": 0.1071,
+      "step": 41664
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.817338195618014e-07,
+      "loss": 0.0559,
+      "step": 41728
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8078508108749147e-07,
+      "loss": 0.1248,
+      "step": 41792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7983634261318156e-07,
+      "loss": 0.0601,
+      "step": 41856
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7888760413887155e-07,
+      "loss": 0.049,
+      "step": 41920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7793886566456164e-07,
+      "loss": 0.0611,
+      "step": 41984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.769901271902517e-07,
+      "loss": 0.0884,
+      "step": 42048
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7604138871594176e-07,
+      "loss": 0.0759,
+      "step": 42112
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7509265024163185e-07,
+      "loss": 0.1484,
+      "step": 42176
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7414391176732184e-07,
+      "loss": 0.0447,
+      "step": 42240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7319517329301193e-07,
+      "loss": 0.0491,
+      "step": 42304
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7224643481870197e-07,
+      "loss": 0.053,
+      "step": 42368
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7129769634439206e-07,
+      "loss": 0.0832,
+      "step": 42432
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7034895787008215e-07,
+      "loss": 0.0677,
+      "step": 42496
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6940021939577213e-07,
+      "loss": 0.0599,
+      "step": 42560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.684514809214622e-07,
+      "loss": 0.1093,
+      "step": 42624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.675027424471523e-07,
+      "loss": 0.116,
+      "step": 42688
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.6655400397284235e-07,
+      "loss": 0.1072,
+      "step": 42752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.656052654985324e-07,
+      "loss": 0.1406,
+      "step": 42816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.6465652702422243e-07,
+      "loss": 0.0339,
+      "step": 42880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.637077885499125e-07,
+      "loss": 0.1058,
+      "step": 42944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.627590500756026e-07,
+      "loss": 0.1187,
+      "step": 43008
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.6181031160129265e-07,
+      "loss": 0.0695,
+      "step": 43072
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.608615731269827e-07,
+      "loss": 0.0938,
+      "step": 43136
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.599128346526727e-07,
+      "loss": 0.08,
+      "step": 43200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.589640961783628e-07,
+      "loss": 0.0911,
+      "step": 43264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.580153577040529e-07,
+      "loss": 0.0983,
+      "step": 43328
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5706661922974294e-07,
+      "loss": 0.115,
+      "step": 43392
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.56117880755433e-07,
+      "loss": 0.0913,
+      "step": 43456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5516914228112307e-07,
+      "loss": 0.0739,
+      "step": 43520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.542204038068131e-07,
+      "loss": 0.1011,
+      "step": 43584
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.532716653325032e-07,
+      "loss": 0.0595,
+      "step": 43648
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.523229268581932e-07,
+      "loss": 0.1025,
+      "step": 43712
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5137418838388327e-07,
+      "loss": 0.1022,
+      "step": 43776
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5042544990957336e-07,
+      "loss": 0.0641,
+      "step": 43840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.494767114352634e-07,
+      "loss": 0.1397,
+      "step": 43904
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.485279729609535e-07,
+      "loss": 0.136,
+      "step": 43968
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.475792344866435e-07,
+      "loss": 0.0946,
+      "step": 44032
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4663049601233357e-07,
+      "loss": 0.0963,
+      "step": 44096
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4568175753802366e-07,
+      "loss": 0.0511,
+      "step": 44160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.447330190637137e-07,
+      "loss": 0.0665,
+      "step": 44224
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.437842805894038e-07,
+      "loss": 0.0735,
+      "step": 44288
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.428355421150938e-07,
+      "loss": 0.0443,
+      "step": 44352
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4188680364078386e-07,
+      "loss": 0.1191,
+      "step": 44416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4093806516647395e-07,
+      "loss": 0.0886,
+      "step": 44480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.39989326692164e-07,
+      "loss": 0.0917,
+      "step": 44544
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.390405882178541e-07,
+      "loss": 0.0751,
+      "step": 44608
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.380918497435441e-07,
+      "loss": 0.0692,
+      "step": 44672
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3714311126923416e-07,
+      "loss": 0.1101,
+      "step": 44736
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3619437279492425e-07,
+      "loss": 0.1307,
+      "step": 44800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3524563432061434e-07,
+      "loss": 0.1512,
+      "step": 44864
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.342968958463043e-07,
+      "loss": 0.0731,
+      "step": 44928
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.333481573719944e-07,
+      "loss": 0.0688,
+      "step": 44992
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3239941889768445e-07,
+      "loss": 0.0476,
+      "step": 45056
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3145068042337454e-07,
+      "loss": 0.0511,
+      "step": 45120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3050194194906463e-07,
+      "loss": 0.0658,
+      "step": 45184
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.295532034747546e-07,
+      "loss": 0.0714,
+      "step": 45248
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.286044650004447e-07,
+      "loss": 0.0491,
+      "step": 45312
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2765572652613474e-07,
+      "loss": 0.0355,
+      "step": 45376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2670698805182484e-07,
+      "loss": 0.0322,
+      "step": 45440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.257582495775149e-07,
+      "loss": 0.034,
+      "step": 45504
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.248095111032049e-07,
+      "loss": 0.0448,
+      "step": 45568
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.23860772628895e-07,
+      "loss": 0.0289,
+      "step": 45632
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.229120341545851e-07,
+      "loss": 0.0324,
+      "step": 45696
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.2196329568027513e-07,
+      "loss": 0.0802,
+      "step": 45760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.2101455720596517e-07,
+      "loss": 0.0237,
+      "step": 45824
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.200806427703163e-07,
+      "loss": 0.0423,
+      "step": 45888
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.191319042960064e-07,
+      "loss": 0.0968,
+      "step": 45952
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1818316582169645e-07,
+      "loss": 0.0359,
+      "step": 46016
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1723442734738654e-07,
+      "loss": 0.0361,
+      "step": 46080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.162856888730766e-07,
+      "loss": 0.0676,
+      "step": 46144
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.153369503987666e-07,
+      "loss": 0.0388,
+      "step": 46208
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.143882119244567e-07,
+      "loss": 0.0332,
+      "step": 46272
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.1343947345014674e-07,
+      "loss": 0.0941,
+      "step": 46336
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.124907349758368e-07,
+      "loss": 0.0518,
+      "step": 46400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.1154199650152687e-07,
+      "loss": 0.0624,
+      "step": 46464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.105932580272169e-07,
+      "loss": 0.0516,
+      "step": 46528
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.09644519552907e-07,
+      "loss": 0.0347,
+      "step": 46592
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.086957810785971e-07,
+      "loss": 0.0452,
+      "step": 46656
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0774704260428707e-07,
+      "loss": 0.063,
+      "step": 46720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0679830412997716e-07,
+      "loss": 0.0623,
+      "step": 46784
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.058495656556672e-07,
+      "loss": 0.054,
+      "step": 46848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.049008271813573e-07,
+      "loss": 0.0659,
+      "step": 46912
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.039520887070474e-07,
+      "loss": 0.0178,
+      "step": 46976
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0300335023273737e-07,
+      "loss": 0.0412,
+      "step": 47040
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0205461175842746e-07,
+      "loss": 0.0401,
+      "step": 47104
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.011058732841175e-07,
+      "loss": 0.0287,
+      "step": 47168
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.001571348098076e-07,
+      "loss": 0.0171,
+      "step": 47232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.992083963354976e-07,
+      "loss": 0.0606,
+      "step": 47296
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9825965786118766e-07,
+      "loss": 0.0343,
+      "step": 47360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9731091938687775e-07,
+      "loss": 0.0276,
+      "step": 47424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9636218091256784e-07,
+      "loss": 0.0567,
+      "step": 47488
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.954134424382579e-07,
+      "loss": 0.0347,
+      "step": 47552
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.944647039639479e-07,
+      "loss": 0.0634,
+      "step": 47616
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9351596548963795e-07,
+      "loss": 0.0949,
+      "step": 47680
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9256722701532805e-07,
+      "loss": 0.0485,
+      "step": 47744
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.9161848854101814e-07,
+      "loss": 0.0969,
+      "step": 47808
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.906697500667082e-07,
+      "loss": 0.0709,
+      "step": 47872
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.897210115923982e-07,
+      "loss": 0.046,
+      "step": 47936
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8877227311808825e-07,
+      "loss": 0.0567,
+      "step": 48000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8782353464377834e-07,
+      "loss": 0.0712,
+      "step": 48064
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8687479616946843e-07,
+      "loss": 0.0472,
+      "step": 48128
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.859260576951584e-07,
+      "loss": 0.0687,
+      "step": 48192
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.849773192208485e-07,
+      "loss": 0.0295,
+      "step": 48256
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.840285807465386e-07,
+      "loss": 0.0404,
+      "step": 48320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8309466631088975e-07,
+      "loss": 0.0719,
+      "step": 48384
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8214592783657984e-07,
+      "loss": 0.0129,
+      "step": 48448
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.811971893622698e-07,
+      "loss": 0.0262,
+      "step": 48512
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.802484508879599e-07,
+      "loss": 0.0499,
+      "step": 48576
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7929971241364995e-07,
+      "loss": 0.0229,
+      "step": 48640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7835097393934004e-07,
+      "loss": 0.0338,
+      "step": 48704
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.774022354650301e-07,
+      "loss": 0.0614,
+      "step": 48768
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.764534969907201e-07,
+      "loss": 0.0913,
+      "step": 48832
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.755047585164102e-07,
+      "loss": 0.0481,
+      "step": 48896
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7455602004210024e-07,
+      "loss": 0.0549,
+      "step": 48960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7360728156779033e-07,
+      "loss": 0.0532,
+      "step": 49024
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7265854309348037e-07,
+      "loss": 0.0379,
+      "step": 49088
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.717098046191704e-07,
+      "loss": 0.0579,
+      "step": 49152
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.707610661448605e-07,
+      "loss": 0.0431,
+      "step": 49216
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.698123276705506e-07,
+      "loss": 0.0223,
+      "step": 49280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.6886358919624063e-07,
+      "loss": 0.0339,
+      "step": 49344
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6791485072193067e-07,
+      "loss": 0.0396,
+      "step": 49408
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.669661122476207e-07,
+      "loss": 0.0539,
+      "step": 49472
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.660173737733108e-07,
+      "loss": 0.0415,
+      "step": 49536
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.650686352990009e-07,
+      "loss": 0.083,
+      "step": 49600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.6411989682469087e-07,
+      "loss": 0.0333,
+      "step": 49664
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.6317115835038096e-07,
+      "loss": 0.0592,
+      "step": 49728
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.62222419876071e-07,
+      "loss": 0.0342,
+      "step": 49792
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.612736814017611e-07,
+      "loss": 0.0314,
+      "step": 49856
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.603249429274512e-07,
+      "loss": 0.0285,
+      "step": 49920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5937620445314116e-07,
+      "loss": 0.0208,
+      "step": 49984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5842746597883126e-07,
+      "loss": 0.0813,
+      "step": 50048
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5747872750452135e-07,
+      "loss": 0.0699,
+      "step": 50112
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.565299890302114e-07,
+      "loss": 0.0617,
+      "step": 50176
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.555812505559015e-07,
+      "loss": 0.0531,
+      "step": 50240
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5463251208159146e-07,
+      "loss": 0.0577,
+      "step": 50304
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5368377360728155e-07,
+      "loss": 0.0538,
+      "step": 50368
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5273503513297164e-07,
+      "loss": 0.0789,
+      "step": 50432
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.517862966586617e-07,
+      "loss": 0.025,
+      "step": 50496
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5083755818435177e-07,
+      "loss": 0.0663,
+      "step": 50560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.498888197100418e-07,
+      "loss": 0.0497,
+      "step": 50624
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4894008123573184e-07,
+      "loss": 0.0207,
+      "step": 50688
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4799134276142193e-07,
+      "loss": 0.0381,
+      "step": 50752
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4704260428711197e-07,
+      "loss": 0.0441,
+      "step": 50816
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.46093865812802e-07,
+      "loss": 0.0447,
+      "step": 50880
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.451451273384921e-07,
+      "loss": 0.105,
+      "step": 50944
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4419638886418214e-07,
+      "loss": 0.0551,
+      "step": 51008
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4324765038987223e-07,
+      "loss": 0.0489,
+      "step": 51072
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4229891191556227e-07,
+      "loss": 0.0243,
+      "step": 51136
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.413501734412523e-07,
+      "loss": 0.0633,
+      "step": 51200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.404014349669424e-07,
+      "loss": 0.0378,
+      "step": 51264
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3945269649263243e-07,
+      "loss": 0.0696,
+      "step": 51328
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3850395801832247e-07,
+      "loss": 0.0525,
+      "step": 51392
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3755521954401256e-07,
+      "loss": 0.012,
+      "step": 51456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3660648106970262e-07,
+      "loss": 0.0391,
+      "step": 51520
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3565774259539266e-07,
+      "loss": 0.0417,
+      "step": 51584
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3470900412108275e-07,
+      "loss": 0.0322,
+      "step": 51648
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.337602656467728e-07,
+      "loss": 0.0429,
+      "step": 51712
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3281152717246285e-07,
+      "loss": 0.0403,
+      "step": 51776
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3186278869815292e-07,
+      "loss": 0.0978,
+      "step": 51840
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3091405022384298e-07,
+      "loss": 0.0637,
+      "step": 51904
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2996531174953305e-07,
+      "loss": 0.0582,
+      "step": 51968
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2901657327522309e-07,
+      "loss": 0.0633,
+      "step": 52032
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.2806783480091315e-07,
+      "loss": 0.0419,
+      "step": 52096
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.271190963266032e-07,
+      "loss": 0.065,
+      "step": 52160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.2617035785229328e-07,
+      "loss": 0.0501,
+      "step": 52224
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2522161937798334e-07,
+      "loss": 0.0296,
+      "step": 52288
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2427288090367338e-07,
+      "loss": 0.0098,
+      "step": 52352
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2332414242936344e-07,
+      "loss": 0.0349,
+      "step": 52416
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.223754039550535e-07,
+      "loss": 0.041,
+      "step": 52480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2142666548074357e-07,
+      "loss": 0.0507,
+      "step": 52544
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.204779270064336e-07,
+      "loss": 0.0495,
+      "step": 52608
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.1952918853212367e-07,
+      "loss": 0.0757,
+      "step": 52672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1858045005781376e-07,
+      "loss": 0.0352,
+      "step": 52736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.176317115835038e-07,
+      "loss": 0.0602,
+      "step": 52800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1668297310919387e-07,
+      "loss": 0.0309,
+      "step": 52864
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.157342346348839e-07,
+      "loss": 0.0931,
+      "step": 52928
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.14785496160574e-07,
+      "loss": 0.0305,
+      "step": 52992
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1383675768626403e-07,
+      "loss": 0.0377,
+      "step": 53056
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.128880192119541e-07,
+      "loss": 0.0263,
+      "step": 53120
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1193928073764416e-07,
+      "loss": 0.0304,
+      "step": 53184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.109905422633342e-07,
+      "loss": 0.0397,
+      "step": 53248
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.100418037890243e-07,
+      "loss": 0.0294,
+      "step": 53312
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0909306531471433e-07,
+      "loss": 0.0738,
+      "step": 53376
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.081443268404044e-07,
+      "loss": 0.0298,
+      "step": 53440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0719558836609443e-07,
+      "loss": 0.046,
+      "step": 53504
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0624684989178452e-07,
+      "loss": 0.024,
+      "step": 53568
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0529811141747458e-07,
+      "loss": 0.0338,
+      "step": 53632
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0434937294316462e-07,
+      "loss": 0.056,
+      "step": 53696
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0340063446885468e-07,
+      "loss": 0.0483,
+      "step": 53760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0245189599454472e-07,
+      "loss": 0.0351,
+      "step": 53824
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.015031575202348e-07,
+      "loss": 0.1004,
+      "step": 53888
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0055441904592485e-07,
+      "loss": 0.0775,
+      "step": 53952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9960568057161492e-07,
+      "loss": 0.0388,
+      "step": 54016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9865694209730498e-07,
+      "loss": 0.0591,
+      "step": 54080
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9770820362299504e-07,
+      "loss": 0.0416,
+      "step": 54144
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967594651486851e-07,
+      "loss": 0.0373,
+      "step": 54208
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9581072667437515e-07,
+      "loss": 0.0544,
+      "step": 54272
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.948619882000652e-07,
+      "loss": 0.0256,
+      "step": 54336
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9391324972575527e-07,
+      "loss": 0.0852,
+      "step": 54400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9296451125144534e-07,
+      "loss": 0.0333,
+      "step": 54464
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.920157727771354e-07,
+      "loss": 0.0327,
+      "step": 54528
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9106703430282544e-07,
+      "loss": 0.0311,
+      "step": 54592
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9011829582851553e-07,
+      "loss": 0.0676,
+      "step": 54656
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8916955735420557e-07,
+      "loss": 0.098,
+      "step": 54720
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8823564291855672e-07,
+      "loss": 0.0668,
+      "step": 54784
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8728690444424678e-07,
+      "loss": 0.0498,
+      "step": 54848
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8633816596993685e-07,
+      "loss": 0.0681,
+      "step": 54912
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8538942749562688e-07,
+      "loss": 0.0353,
+      "step": 54976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8444068902131695e-07,
+      "loss": 0.0776,
+      "step": 55040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8349195054700704e-07,
+      "loss": 0.0636,
+      "step": 55104
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8254321207269708e-07,
+      "loss": 0.0415,
+      "step": 55168
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8159447359838714e-07,
+      "loss": 0.0646,
+      "step": 55232
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8064573512407718e-07,
+      "loss": 0.0218,
+      "step": 55296
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7971182068842835e-07,
+      "loss": 0.0818,
+      "step": 55360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.787630822141184e-07,
+      "loss": 0.0541,
+      "step": 55424
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7781434373980848e-07,
+      "loss": 0.0667,
+      "step": 55488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.768804293041596e-07,
+      "loss": 0.0717,
+      "step": 55552
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7593169082984967e-07,
+      "loss": 0.035,
+      "step": 55616
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7498295235553973e-07,
+      "loss": 0.0863,
+      "step": 55680
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.740342138812298e-07,
+      "loss": 0.0839,
+      "step": 55744
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7308547540691984e-07,
+      "loss": 0.0298,
+      "step": 55808
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.721367369326099e-07,
+      "loss": 0.0739,
+      "step": 55872
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7120282249696105e-07,
+      "loss": 0.0557,
+      "step": 55936
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7025408402265111e-07,
+      "loss": 0.0717,
+      "step": 56000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.693053455483412e-07,
+      "loss": 0.0297,
+      "step": 56064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6835660707403124e-07,
+      "loss": 0.073,
+      "step": 56128
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.674078685997213e-07,
+      "loss": 0.0389,
+      "step": 56192
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6645913012541134e-07,
+      "loss": 0.0722,
+      "step": 56256
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6551039165110144e-07,
+      "loss": 0.0544,
+      "step": 56320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6456165317679147e-07,
+      "loss": 0.0641,
+      "step": 56384
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6361291470248154e-07,
+      "loss": 0.0791,
+      "step": 56448
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.626641762281716e-07,
+      "loss": 0.1186,
+      "step": 56512
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6171543775386164e-07,
+      "loss": 0.038,
+      "step": 56576
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6076669927955173e-07,
+      "loss": 0.0557,
+      "step": 56640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5981796080524177e-07,
+      "loss": 0.0722,
+      "step": 56704
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5886922233093183e-07,
+      "loss": 0.0837,
+      "step": 56768
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5792048385662187e-07,
+      "loss": 0.0399,
+      "step": 56832
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5697174538231196e-07,
+      "loss": 0.0971,
+      "step": 56896
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5602300690800202e-07,
+      "loss": 0.0811,
+      "step": 56960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5507426843369206e-07,
+      "loss": 0.029,
+      "step": 57024
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5412552995938213e-07,
+      "loss": 0.0522,
+      "step": 57088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.531767914850722e-07,
+      "loss": 0.0415,
+      "step": 57152
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5222805301076225e-07,
+      "loss": 0.0377,
+      "step": 57216
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.512793145364523e-07,
+      "loss": 0.0351,
+      "step": 57280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5033057606214236e-07,
+      "loss": 0.0392,
+      "step": 57344
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4938183758783242e-07,
+      "loss": 0.0584,
+      "step": 57408
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4843309911352248e-07,
+      "loss": 0.0573,
+      "step": 57472
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4748436063921255e-07,
+      "loss": 0.0395,
+      "step": 57536
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4653562216490259e-07,
+      "loss": 0.068,
+      "step": 57600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4558688369059265e-07,
+      "loss": 0.0393,
+      "step": 57664
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4463814521628274e-07,
+      "loss": 0.0338,
+      "step": 57728
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4368940674197278e-07,
+      "loss": 0.0328,
+      "step": 57792
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4274066826766284e-07,
+      "loss": 0.0261,
+      "step": 57856
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4179192979335288e-07,
+      "loss": 0.0446,
+      "step": 57920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4084319131904297e-07,
+      "loss": 0.0593,
+      "step": 57984
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.39894452844733e-07,
+      "loss": 0.1374,
+      "step": 58048
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.3894571437042307e-07,
+      "loss": 0.04,
+      "step": 58112
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3799697589611314e-07,
+      "loss": 0.0511,
+      "step": 58176
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3704823742180317e-07,
+      "loss": 0.0591,
+      "step": 58240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3609949894749326e-07,
+      "loss": 0.0226,
+      "step": 58304
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.351507604731833e-07,
+      "loss": 0.034,
+      "step": 58368
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.3420202199887337e-07,
+      "loss": 0.0142,
+      "step": 58432
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.332532835245634e-07,
+      "loss": 0.0594,
+      "step": 58496
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.323045450502535e-07,
+      "loss": 0.0507,
+      "step": 58560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.3135580657594356e-07,
+      "loss": 0.0577,
+      "step": 58624
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.304070681016336e-07,
+      "loss": 0.0608,
+      "step": 58688
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2945832962732366e-07,
+      "loss": 0.0248,
+      "step": 58752
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2850959115301373e-07,
+      "loss": 0.0388,
+      "step": 58816
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.275608526787038e-07,
+      "loss": 0.0625,
+      "step": 58880
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2661211420439383e-07,
+      "loss": 0.0497,
+      "step": 58944
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.256633757300839e-07,
+      "loss": 0.0442,
+      "step": 59008
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2471463725577396e-07,
+      "loss": 0.0489,
+      "step": 59072
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2376589878146402e-07,
+      "loss": 0.0707,
+      "step": 59136
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2281716030715406e-07,
+      "loss": 0.0761,
+      "step": 59200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2186842183284415e-07,
+      "loss": 0.0605,
+      "step": 59264
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2091968335853419e-07,
+      "loss": 0.0451,
+      "step": 59328
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.1997094488422425e-07,
+      "loss": 0.0617,
+      "step": 59392
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.1902220640991431e-07,
+      "loss": 0.0501,
+      "step": 59456
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1807346793560436e-07,
+      "loss": 0.0313,
+      "step": 59520
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1712472946129443e-07,
+      "loss": 0.079,
+      "step": 59584
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1617599098698449e-07,
+      "loss": 0.1015,
+      "step": 59648
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1522725251267456e-07,
+      "loss": 0.0585,
+      "step": 59712
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1427851403836461e-07,
+      "loss": 0.0521,
+      "step": 59776
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1332977556405467e-07,
+      "loss": 0.0504,
+      "step": 59840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1238103708974472e-07,
+      "loss": 0.0409,
+      "step": 59904
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.1143229861543477e-07,
+      "loss": 0.047,
+      "step": 59968
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.1048356014112484e-07,
+      "loss": 0.064,
+      "step": 60032
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.095348216668149e-07,
+      "loss": 0.0465,
+      "step": 60096
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0858608319250497e-07,
+      "loss": 0.0636,
+      "step": 60160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0763734471819502e-07,
+      "loss": 0.0653,
+      "step": 60224
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0668860624388508e-07,
+      "loss": 0.0682,
+      "step": 60288
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0573986776957513e-07,
+      "loss": 0.0219,
+      "step": 60352
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.047911292952652e-07,
+      "loss": 0.0452,
+      "step": 60416
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0385721485961635e-07,
+      "loss": 0.059,
+      "step": 60480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0290847638530641e-07,
+      "loss": 0.0801,
+      "step": 60544
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0195973791099646e-07,
+      "loss": 0.0582,
+      "step": 60608
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0101099943668653e-07,
+      "loss": 0.0236,
+      "step": 60672
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0006226096237659e-07,
+      "loss": 0.0576,
+      "step": 60736
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.911352248806665e-08,
+      "loss": 0.0438,
+      "step": 60800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.81647840137567e-08,
+      "loss": 0.0709,
+      "step": 60864
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.721604553944676e-08,
+      "loss": 0.0626,
+      "step": 60928
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.626730706513682e-08,
+      "loss": 0.0581,
+      "step": 60992
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.531856859082687e-08,
+      "loss": 0.0422,
+      "step": 61056
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.436983011651695e-08,
+      "loss": 0.0403,
+      "step": 61120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.3421091642207e-08,
+      "loss": 0.0469,
+      "step": 61184
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.247235316789706e-08,
+      "loss": 0.0607,
+      "step": 61248
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.152361469358711e-08,
+      "loss": 0.0208,
+      "step": 61312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.057487621927718e-08,
+      "loss": 0.0468,
+      "step": 61376
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.962613774496723e-08,
+      "loss": 0.0807,
+      "step": 61440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.86773992706573e-08,
+      "loss": 0.073,
+      "step": 61504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.772866079634736e-08,
+      "loss": 0.073,
+      "step": 61568
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.677992232203742e-08,
+      "loss": 0.052,
+      "step": 61632
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.583118384772747e-08,
+      "loss": 0.0783,
+      "step": 61696
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.488244537341752e-08,
+      "loss": 0.0215,
+      "step": 61760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.393370689910759e-08,
+      "loss": 0.0627,
+      "step": 61824
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.298496842479764e-08,
+      "loss": 0.072,
+      "step": 61888
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.20362299504877e-08,
+      "loss": 0.0395,
+      "step": 61952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.108749147617777e-08,
+      "loss": 0.1183,
+      "step": 62016
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.013875300186783e-08,
+      "loss": 0.0483,
+      "step": 62080
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.919001452755788e-08,
+      "loss": 0.0571,
+      "step": 62144
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.824127605324795e-08,
+      "loss": 0.0984,
+      "step": 62208
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.7292537578938e-08,
+      "loss": 0.0771,
+      "step": 62272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.634379910462806e-08,
+      "loss": 0.0559,
+      "step": 62336
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.539506063031811e-08,
+      "loss": 0.0519,
+      "step": 62400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.444632215600819e-08,
+      "loss": 0.0566,
+      "step": 62464
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.349758368169824e-08,
+      "loss": 0.0373,
+      "step": 62528
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.25488452073883e-08,
+      "loss": 0.0581,
+      "step": 62592
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.160010673307836e-08,
+      "loss": 0.1356,
+      "step": 62656
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.06513682587684e-08,
+      "loss": 0.0374,
+      "step": 62720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.970262978445847e-08,
+      "loss": 0.0505,
+      "step": 62784
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.875389131014853e-08,
+      "loss": 0.0789,
+      "step": 62848
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.78051528358386e-08,
+      "loss": 0.0444,
+      "step": 62912
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.685641436152865e-08,
+      "loss": 0.0152,
+      "step": 62976
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.590767588721871e-08,
+      "loss": 0.0627,
+      "step": 63040
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.495893741290877e-08,
+      "loss": 0.0492,
+      "step": 63104
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.401019893859883e-08,
+      "loss": 0.0256,
+      "step": 63168
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.306146046428888e-08,
+      "loss": 0.0931,
+      "step": 63232
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.211272198997894e-08,
+      "loss": 0.0815,
+      "step": 63296
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.116398351566901e-08,
+      "loss": 0.0261,
+      "step": 63360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.021524504135907e-08,
+      "loss": 0.0329,
+      "step": 63424
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.9266506567049123e-08,
+      "loss": 0.0543,
+      "step": 63488
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.831776809273918e-08,
+      "loss": 0.0497,
+      "step": 63552
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.7369029618429245e-08,
+      "loss": 0.0509,
+      "step": 63616
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.64202911441193e-08,
+      "loss": 0.0352,
+      "step": 63680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.547155266980936e-08,
+      "loss": 0.0703,
+      "step": 63744
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.4522814195499424e-08,
+      "loss": 0.0203,
+      "step": 63808
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.357407572118948e-08,
+      "loss": 0.0462,
+      "step": 63872
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.262533724687954e-08,
+      "loss": 0.0323,
+      "step": 63936
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.167659877256959e-08,
+      "loss": 0.0481,
+      "step": 64000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.0727860298259654e-08,
+      "loss": 0.0198,
+      "step": 64064
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.977912182394971e-08,
+      "loss": 0.0386,
+      "step": 64128
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.883038334963977e-08,
+      "loss": 0.0516,
+      "step": 64192
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.7881644875329834e-08,
+      "loss": 0.0183,
+      "step": 64256
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.693290640101989e-08,
+      "loss": 0.0362,
+      "step": 64320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.598416792670995e-08,
+      "loss": 0.0174,
+      "step": 64384
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.503542945240001e-08,
+      "loss": 0.1017,
+      "step": 64448
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.408669097809007e-08,
+      "loss": 0.0516,
+      "step": 64512
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.313795250378013e-08,
+      "loss": 0.028,
+      "step": 64576
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.2189214029470186e-08,
+      "loss": 0.0999,
+      "step": 64640
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.124047555516025e-08,
+      "loss": 0.0623,
+      "step": 64704
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.029173708085031e-08,
+      "loss": 0.069,
+      "step": 64768
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.9342998606540365e-08,
+      "loss": 0.031,
+      "step": 64832
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.839426013223043e-08,
+      "loss": 0.0489,
+      "step": 64896
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.744552165792048e-08,
+      "loss": 0.0379,
+      "step": 64960
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.649678318361054e-08,
+      "loss": 0.0248,
+      "step": 65024
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5562868747961694e-08,
+      "loss": 0.0731,
+      "step": 65088
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.461413027365175e-08,
+      "loss": 0.0469,
+      "step": 65152
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.366539179934181e-08,
+      "loss": 0.036,
+      "step": 65216
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.271665332503187e-08,
+      "loss": 0.0428,
+      "step": 65280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.176791485072193e-08,
+      "loss": 0.0546,
+      "step": 65344
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.081917637641199e-08,
+      "loss": 0.059,
+      "step": 65408
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9870437902102046e-08,
+      "loss": 0.0494,
+      "step": 65472
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.8921699427792107e-08,
+      "loss": 0.059,
+      "step": 65536
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.7972960953482164e-08,
+      "loss": 0.0715,
+      "step": 65600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.7024222479172225e-08,
+      "loss": 0.0433,
+      "step": 65664
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.6075484004862282e-08,
+      "loss": 0.0305,
+      "step": 65728
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5126745530552343e-08,
+      "loss": 0.0879,
+      "step": 65792
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.41780070562424e-08,
+      "loss": 0.0691,
+      "step": 65856
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3229268581932462e-08,
+      "loss": 0.0584,
+      "step": 65920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.228053010762252e-08,
+      "loss": 0.0669,
+      "step": 65984
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1331791633312577e-08,
+      "loss": 0.0379,
+      "step": 66048
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0383053159002638e-08,
+      "loss": 0.0294,
+      "step": 66112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9434314684692695e-08,
+      "loss": 0.0282,
+      "step": 66176
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8485576210382756e-08,
+      "loss": 0.0697,
+      "step": 66240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7536837736072817e-08,
+      "loss": 0.0586,
+      "step": 66304
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6588099261762874e-08,
+      "loss": 0.0259,
+      "step": 66368
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5639360787452932e-08,
+      "loss": 0.0476,
+      "step": 66432
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4690622313142993e-08,
+      "loss": 0.0248,
+      "step": 66496
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.374188383883305e-08,
+      "loss": 0.0416,
+      "step": 66560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.279314536452311e-08,
+      "loss": 0.0324,
+      "step": 66624
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1844406890213169e-08,
+      "loss": 0.009,
+      "step": 66688
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.089566841590323e-08,
+      "loss": 0.0341,
+      "step": 66752
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.946929941593287e-09,
+      "loss": 0.0234,
+      "step": 66816
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.998191467283346e-09,
+      "loss": 0.0638,
+      "step": 66880
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.049452992973405e-09,
+      "loss": 0.0757,
+      "step": 66944
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.1007145186634646e-09,
+      "loss": 0.0294,
+      "step": 67008
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.151976044353523e-09,
+      "loss": 0.0739,
+      "step": 67072
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.203237570043583e-09,
+      "loss": 0.0545,
+      "step": 67136
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.254499095733641e-09,
+      "loss": 0.0294,
+      "step": 67200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3057606214237005e-09,
+      "loss": 0.0236,
+      "step": 67264
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3570221471137597e-09,
+      "loss": 0.0805,
+      "step": 67328
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4082836728038187e-09,
+      "loss": 0.0691,
+      "step": 67392
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5954519849387766e-10,
+      "loss": 0.0591,
+      "step": 67456
+    }
+  ],
+  "logging_steps": 64,
+  "max_steps": 67458,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 22486,
+  "total_flos": 1.432248707211264e+20,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}